Item classification system for use in creating a personal appearance to engender a positive self-image
An AI-driven item classification system addresses the misclassification issues in fashion recommendation systems by using machine learning to provide personalized fashion recommendations, enhancing customer satisfaction and reducing returns, and improving body image through tailored clothing suggestions.
Patent Information
- Application Number
- PCT/US2024/030061
- Authority / Receiving Office
- WO · WO
- Patent Type
- Applications
- Current Assignee / Owner
- Priority Date
- 2023-05-19
- Filing Date
- 2024-05-17
- Publication Date
- 2024-11-28
AI Technical Summary
Current recommendation systems in the fashion industry misclassify individuals' tastes and preferences, leading to a lack of trust and high return rates due to superficial categorizations that do not align with aspirations, resulting in distress and poor body image issues for customers with body dysmorphic disorder.
An AI-powered item classification system using supervised and unsupervised machine learning to provide personalized recommendations by understanding users' unique needs, preferences, and aspirations through a seven-step process, integrating with both online and brick-and-mortar platforms, and utilizing an avatar interface for a personalized stylist experience.
The system increases customer satisfaction, reduces returns, and enhances sales by offering tailored clothing suggestions that align with users' aspirations, thereby improving body image and reducing distress related to fashion choices.
Smart Images

Figure US2024030061_28112024_PF_FP_ABST
Abstract
Description
ITEM CLASSIFICATION SYSTEM FOR USE IN CREATING A PERSONAL APPEARANCE TO ENGENDER A POSITIVE SELF-IMAGECROSS-REFERENCE TO RELATED APPLICATIONS
[0001] This application claims the benefit of priority to U.S. Provisional Application No. 63 / 503,385, filed May 19, 2023, which is incorporated herein by reference in its entirety.TECHNICAL FIELD
[0002] Various of the disclosed embodiments concern an item classification system. More particularly, embodiments of the invention concern an item classification system for use in creating a personal appearance to engender a positive self-image.BACKGROUND
[0003] Body dysmorphic disorder is a mental health condition in which a person cannot stop thinking about one or more perceived defects or flaws in their appearance — a flaw that appears minor or cannot be seen by others. But the person may feel so embarrassed, ashamed, and anxious that the person may avoid many social situations.
[0004] People who suffer from body dysmorphic disorder intensely focus on appearance and body image, repeatedly checking the mirror, grooming or seeking reassurance, sometimes for many hours each day. Their perceived flaw and the repetitive behaviors cause significant distress and impact their ability to function in their daily life.
[0005] People who suffer from body dysmorphic disorder may seek out numerous cosmetic procedures to try to fix their perceived flaw. Afterward, they may feel temporary satisfaction or a reduction in their distress, but often the anxiety returns, and they may resume searching for other ways to fix their perceived flaw.
[0006] Signs and symptoms of body dysmorphic disorder include:• Being extremely preoccupied with a perceived flaw in appearance that to others cannot be seen or appears minor;• Strong belief that they have a defect in their appearance that makes them ugly or deformed;• Belief that others take special notice of their appearance in a negative way or mock them;• Engaging in behaviors aimed at fixing or hiding the perceived flaw that are difficult to resist or control, such as frequently checking the mirror, grooming, or skin picking;• Attempting to hide perceived flaws with styling, makeup, or clothes;• Constantly comparing their appearance with others;• Frequently seeking reassurance about their appearance from others;• Having perfectionist tendencies;• Seeking cosmetic procedures with little satisfaction;• Avoiding social situations; and• Preoccupation with their appearance and excessive thoughts and repetitive behaviors can be unwanted, difficult to control and so time-consuming that they can cause major distress or problems in their social life, work, school, or other areas of functioning.
[0007] People with body dysmorphic disorder may excessively focus over one or more parts of their body. The bodily feature that they focus on may change over time. The most common features people tend to fixate about include:• Face, such as nose, complexion, wrinkles, acne, and other blemishes;• Hair, such as appearance, thinning, and baldness;• Skin and vein appearance;• Breast size;• Muscle size and tone;• Genitalia; andA preoccupation with their body build being too small or not muscular.
[0008] Insight about body dysmorphic disorder varies. People who suffer from body dysmorphic disorder may recognize that their beliefs about their perceived flaws may be excessive or not be true, or think that they probably are true, or be absolutely convinced that they are true. The more convinced they are of their beliefs, the more distress and disruption they may experience in their life.
[0009] Studies show that human beings are increasingly more susceptible to stress, anxiety, and body related issues now in a generalized sense. These issues create an intensified problem when people shop online because it is a shot in the dark whether or not a product overcomes or accentuates perceived defects or flaws in their appearance.SUMMARY
[0010] We’re only alive once. And 91% of us spend every day feeling poorly about the magnificent temples that are our bodies. Embodiments of the invention use artificial intelligence (Al) and supervised / unsupervised machine learning to create a personal appearance to engender a positive self-image, based on the recognition that the clothing or costume a person wears determines the story they live. Embodiments of the invention provide an Al-implemented service that helps people write the story that they want to live through the power of dressing for the part.
[0011] In some aspects, the techniques described herein relate to a method, including: receiving, at an artificial intelligence (Al) processor, a first user input from a client device, the first user input indicative of a user shopping for an item; responsive to receiving the first user input indicative of a user shopping for an item, initiating a transmission of a questionnaire to the client device; responsive to transmitting the questionnaire to the client device, processing the following operations iteratively until a predetermined number of times, the predetermined number of times based on the questionnaire: receiving a subsequent user input from the client device, the subsequent user input including informational data about the questionnaire; retrieving, by the Al processor and from a brand tagging processor, a plurality of tagged brand items based on the subsequent user input; retrieving, by the Al processor and from a user profile processor, user profile data, the user profile data associated with the user, the retrieving based on the subsequent user input; retrieving, by the Al processor and from a user purchase and return historical processor, user purchase and return historical dataassociated with the user, the retrieving based on the subsequent user input; and using, by the Al processor, the plurality of items, the user profile data, and the user purchase and return historical data to remove undesirable items from the brand tagging data; and at end of iterative processing, transmitting, by the Al processor, remaining items of the plurality of tagged brand items to the client device.BRIEF DESCRIPTION OF THE DRAWINGS
[0012] One or more embodiments of the present disclosure are illustrated by way of example and not limitation in the figures of the accompanying drawings, in which like references indicate similar elements.
[0013] FIG. 1 A is a flow diagram showing a seven-step process according to an embodiment;
[0014] FIG. 1 B-1 C depict a table showing example data points used by the system to achieve creating a personal appearance to engender a positive self-image, according to an embodiment;
[0015] FIG. 2 is a block diagram showing an item classification system for use in creating a personal appearance to engender a positive self-image according to an embodiment;
[0016] FIG. 3A is an architectural diagram showing an item classification system for use in creating a personal appearance to engender a positive self-image according to an embodiment;
[0017] FIG. 3B is an Al architecture, according to an embodiment;
[0018] FIG. 4 is a block diagram of a computer system as may be used to implement certain features of some of the embodiments;
[0019] FIG. 5 is a schematic diagram showing measurement points generation used to create an avatar on which a garment is overlayed, according to an embodiment;
[0020] FIG. 6 is a flow diagram showing an example process for creating a personal appearance to engender a positive self-image, according to an embodiment;
[0021] FIG. 7 is a flow diagram showing an example process for generating measurement points used to create an avatar on which a garment is overlayed, according to an embodiment;
[0022] FIG. 8 is a schematic diagram showing a mirror configured to interact with a user and configured to display an avatar of the user on which a garment is overlayed, according to an embodiment; and
[0023] FIG. 9 is a schematic diagram showing an integrated clothing / inventory conveyor system, according to an embodiment.
[0024] Those skilled in the art will appreciate that the logic and process steps illustrated in the various flow diagrams discussed below may be altered in a variety of ways. For example, the order of the logic may be rearranged, sub-steps may be performed in parallel, illustrated logic may be omitted, other logic may be included, etc. One will recognize that certain steps may be consolidated into a single step and that actions represented by a single step may be alternatively represented as a collection of sub-steps. The figures are designed to make the disclosed concepts more comprehensible to a human reader. Those skilled in the art will appreciate that actual data structures used to store this information may differ from the figures and / or tables shown, in that they, for example, may be organized in a different manner; may contain more or less information than shown; may be compressed, scrambled and / or encrypted; etc.DETAILED DESCRIPTION
[0025] This innovation relates to an Al-powered system that helps customers find clothing that aligns with their personal style, body type, lifestyle, and aspirations. The system combines supervised and unsupervised machine learning techniques to provide personalized recommendations and an engaging shopping experience.The Seven-Step Process.
[0026] At the core of one embodiment of the innovation is a seven-step process that guides the customer through a series of questions and interactions:1 . **Health, Body Image, and Mindset**: The system determines the customer’s health parameters (e.g., obesity, disabilities), body image perceptions, and mindset (e.g., positive, negative) through a questionnaire.2. **Life Phase and Aspirations**: It identifies the customer’s current life phase, transitions, and aspirations to recommend clothing that aligns with their goals.3. **Personal Preferences**: The system learns about the customer’s unique preferences, such as favorite colors and styles and what makes them feel great.4. **Fit and Body Shape**: It educates the customer on proper fit and provides examples based on their body shape.5. **Brand Values**: The system considers the customer’s values regarding brand ethics and manufacturing practices.6. **Essential Items**: It recommends essential clothing items the customer should own based on their lifestyle and aspirations.7. **Budget and Shopping Habits**: The system considers the customer’s typical budget and guides when to spend more or less, as well as optimal shopping times and brands.Al Engine and Personalization.
[0027] The Al engine powers the seven-step process by presenting personalized questions and recommendations based on the customer’s responses. It uses an avatar interface to mimic a personal stylist experience, making the customer feel understood and valued.
[0028] The Al engine continuously learns from the customer’s purchase history, returns, and feedback, refining its recommendations. It employs supervised learning techniques to classify items based on predefined labels (e.g., fit, personality, lifestyle) and unsupervised learning to discover inherent patterns in customer data.In-store and Online Integration.
[0029] The system integrates with both online and brick-and-mortar retail platforms. In-store, employees use an app to augment the customer experience, offering sizing assistance, brand recommendations, and mindfulness quotes.
[0030] Online, customers take the seven-step quiz and receive personalized item recommendations and educational content. They can also create a shareable profile, connect with friends, and receive positive affirmations throughout the shopping journey.Technical Implementation.
[0031] The system architecture includes an Al engine, customer interface, item classification module, and retailer inventory system integration. The Al engine runs on a computing system with a processor, memory, and storage devices.
[0032] Computer vision techniques automate the labeling of inventory items based on predefined categories and customer types. A generative Al model can recommend tags for new items.
[0033] The system also includes features like an avatar overlay for virtual try-on and an integrated clothing / inventory conveyor system for efficient item delivery to fitting rooms.
[0034] It has been found that the majority of prior art recommendation systems and predictive models misclassify people’s tastes and preferences into superficial categories that do not align with their aspirations. The false positives from misclassification result in a lack of trust in the recommendation systems and, therefore, the high rate of returns and attrition from existing services.
[0035] In summary, this innovation provides an Al-driven, personalized shopping experience that considers the customer’s unique needs, preferences, and aspirations. This leads to increased customer satisfaction, reduced returns, and higher sales for retailers.
[0036] Various example embodiments will now be described. The following description provides certain specific details for a thorough understanding and enabling description of these examples. One skilled in the relevant technology will understand, however, that some of the disclosed embodiments may be practiced without many of these details.
[0037] Likewise, one skilled in the relevant technology will also understand that some of the embodiments may include many other obvious features not described in detail herein. Additionally, some well-known structures or functions may not be shown or described in detail below, to avoid unnecessarily obscuring the relevant descriptions of the various examples.
[0038] The terminology used below is to be interpreted in its broadest reasonable manner, even though it is being used in conjunction with a detailed description of certainspecific examples of the embodiments. Indeed, certain terms may even be emphasized below; however, any terminology intended to be interpreted in any restricted manner will be overtly and specifically defined as such in this Detailed Description section.Item Classification System for Use with Creating a Personal Appearance to Engender a Positive Self-Image
[0039] Embodiments of the invention provide a software as a service (SAAS) product and category creator for companies in the fashion retail industry, such as Nordstrom, that enables the fashion retail industry to save upwards of $218 billion per year, by offering a dramatically more personalized experience to their customers. Embodiments of the invention take the customer through a series of seven steps or categories of instruction sets (referred to herein for understandability purposes as “questions”). The customer can engage in as many or as few questions for each category as they would like. Embodiments of the invention match qualities of certain customers who do not give as much data, so that even a customer who opts out of several questions still receives personalization.
[0040] The questions asked differ wildly from what we are used to seeing on platforms such as Stitch Fix or Trunk Club. The customer is engaged through the questions the customer is asked, the order in which they are asked, the psychological response the questions provoke, and the delight that the process causes the customer.
[0041] In embodiments of the invention an Al engine retains information, gathers new information, and learns what does not work for the customer by tracking the returns the customer makes, and asking the customer what about the returns did not work. This is one of the biggest game changers of the system, in that no retailer asks why something was returned. Even when a product does not work out for the customer, embodiments of the invention still gain new data so that the retailer does not repeat the same mistake. The new information is applied across brands, not just the one specific item the customer returned. In embodiments, each brand the store offers is categorized by fit, personality, and lifestyle. The system does not bombard the customer with hundreds of generic outfit images that are all pretty much the same. Instead embodiments of the invention match the customer’s personality, shape, goals, and desires to clothing that fits those descriptors. The system also tells the customer them why this particular item or outfit is recommended, e.g.:• Because the customer is a mom who isn’t ready to stop being cool;• Because the customer is angling for a job promotion;• Because the customer just graduated with a new degree and are ready to level up;• Because the customer is a guy who’s nervous about joining the dating scene.
[0042] Simple, fun, personal questions gather this information in a safe and private way, enabling the customer to feel seen, heard, understood, and that the items recommended are chosen for the express purpose of getting them where they want to go in life.
[0043] The customer is asked questions they did not know they needed to be asked, while offering them solutions to problems they did not know they had, and providing delight they did not even know they needed. This is the secret to customer retention, and leads the purchases to be kept, not returned.
[0044] This is exactly what the inventor did as a personal stylist with the over 10,000 customers she worked with and it truly works. In embodiments of the invention Al engine is similarly trained.
[0045] The Al engine is trained by user inputs (supervised learning) and tracks returns from the customer (unsupervised). It builds generalized knowledge based on categories or types of people who happen to have similar traits. For example, consider a mom over the age of 40, who has a pear shaped figure but still wants to feel young and attractive yet still appropriate for her age. We’ll call her Alice. The Al engine does not throw hundreds of recommendations at a customer within a given day. It narrows them down and is adjustable as well. If Alice logs into Nordstrom one day, happens to have just gotten a divorce and is heading back into the dating scene, in embodiments of the invention the Al engine adapts to her new inputs and offers new, adjusted recommendations. The system continues learning about the customer. It gathers more information every day and adjusts based on the customer’s current needs. The system also observes what the customer searches for, if they are searching at all, and adds further tailoring to the recommendations.
[0046] For example, if the Al engine says that Alice is not a good fit for Valentino shoes because of their heel height and they best fit narrow feet, while she has a widefoot, yet she is searching for them and clearly interested, the system adjusts and offers her some Valentino recommendations too, but explains why, e.g.:“These have a narrow fit, so we suggest going up a size. Alternatively how about these Louboutin heels, that have a similar look but wider fit.”
[0047] In accordance with embodiment herein, when the user has a wider foot than the Valentinos were designed for, and she also has a history of purchasing Hoka One running shoes for her plantar fasciitis, then these vectors combined trigger a response from the system recommending a dress shoe that offers her stability without sacrificing the style. Supervised and unsupervised learning are used. This is because the system is configured to receive unlabeled data, requiring unsupervised techniques such as clustering.
[0048] This is the kind of personalization that does not exist yet but is possible. Embodiments of the invention learn about every brand and the specifics of each. This information can be learned in a variety of ways, e.g. in embodiments of the invention programmers input specifics about each brand and style into the Al engine so that it can start with that knowledge (supervised learning ). The Al engine also has access to the customer’s purchase history, and then integrates this information with the retailer’s system.
[0049] Average users are not aware of how many great items exist that fit their needs. Embodiments of the invention weed out all the bad items that can never work for the user, showing them a select few items that are 90% more likely to work for them, decreasing the need for endless returning and trial and error. This also increases overall sales by exposing users to brands and items they would never search for on their own. Many users do not know certain brands exist, and yet they are a perfect fit. When users experience new items that feel like they were made for them, this releases endorphins and delight, leading to higher purchases and greater pleasure in the buying experience, which compounds into more delight and trust in the seller overtime.
[0050] Significantly, embodiments of the invention decrease the options that a user is choosing from based on the herein disclosed labeling system. For example, users search for a little black dress (LBD), they see 3,000 options, 2,900 of which can never work for them. The users follow a trial and error approach, returning and returning, sometimes until they give up. In embodiments of the invention users only seethe 20-50 options that are known by the system to suit them, thereby eliminating all the trial and error for the user. By noting and tracking what customers purchase and keep vs. what they returned, the system makes increasingly accurate recommendations by eliminating bad options for the user and making more informed recommendations with each use. This is particularly helpful for underserved users, such as new mothers in need of clothing that works for their changing bodies. These users are a tremendously untapped, disappointed market that, when accessed and delighted, create an entirely new revenue stream for the seller.
[0051] The inventor has suffered from body dysmorphia and damaging self-image for her whole life. A shocking 91 % of women globally also deal with poor self-image issues on a daily basis. There is a global body image crisis, that leads women to play small, and not step into the greater roles in which they belong. The inventor sought refuge in clothing and discovered the way the right outfit would transform the way she felt about herself, her entire day would change for the better. Her decisions were more daring if she wore the right outfit.
[0052] She dresses for how she wants to feel, for the story that she wants to live today. She believes that our lives are stories we are creating as we go, and just as an actor is costumed appropriately for a role, we must dress ourselves for the story we wish to live. Her philosophy on shopping for clothes led her to become the highest earning personal stylist for Nordstrom when she was twenty, and a high-level stylist for Stitch Fix. She’s dressed and transformed over 5,000 women over the years and has identified the key problems within the retail industry that lead to us feeling so bad about ourselves. Her core belief is that clothing should be selected based on seven key principles, which all stem from the idea of choices based on attaining a specific feeling in one’s body, not a look.
[0053] We can transform our daily lives through the power of how we dress, and the first step in solving this global body image crisis is to change the way we shop. Further, fashion retailers are missing a huge gap in their services, creating a main cause in our terrible self-image, leading to bad customer experiences and missed opportunities for the business. Embodiments of the invention transform the customer experience and augment sales at the same time.
[0054] Embodiments of the invention use artificial intelligence (Al) and supervised / unsupervised machine learning to create a personal appearance to engender a positive self-image, based on the recognition that the clothing or costume a person wears determines the story they live. Embodiments of the invention provide an Al-implemented service that helps people write the story that they want to live through the power of dressing for the part.
[0055] Embodiments of the invention provide an SaaS tool that makes it possible for retailers to integrate a unique seven-step system into their sales platforms both online and in-store. This provides an empowering, nurturing tool that guides users, especially woman, through their shopping experience and changes the way they think about their bodies. It is similar having a live coach there with them, guiding them the whole way.An Embodiment of the Seven Step Process
[0056] One embodiment can be understood with reference to FIG. 1 A. FIG. 1 A is a flow diagram showing a seven-step process according to an embodiment of the invention. The seven steps comprise:
[0057] 1. Health, body image, mindset. That is, at step 100, a processor is configured to determine a user’s health, body image, and mindset parameters. Examples of health parameters can include but are not limited to obese, good health, bad health, diagnosed with a disease, anemic, and obsessive control disorder. Other examples of health parameters include but are not limited to hormonal weight changes, color blind, and physical disability. Examples of body image parameters can include but are not limited to low body image, good body image, and perceive unrealistic flaws. Other examples of body image parameters include but are not limited to age grouping parameters such as 18-24 years. Examples of mindset parameters can include but are not limited to negative mindset, positive mindset, fragile mindset, altruistic mindset, productive mindset, stoic mindset. Other examples of mindset parameters include but are not limited to life phase or transition point parameters such as but not limited to from student to first real job, retired, and working at a new company.
[0058] 2. Where does your unique strength come from? Understanding your essence. What makes you feel strong. That is, at step 1 10, the processor is furtherconfigured to determine user essence parameters. Examples of user essence parameters include but are not limited to How Do You Want To Feel? parameters, such as but not limited to powerful, stylish, and relevant. Other example user essence parameters include but are not limited to personality type parameters such as for example outgoing, ambitious, and eccentric.
[0059] 3. Fit guidance and why it’s so important. What does fit actually mean?Clothing that fits is automatically a transformation. Fit guidance. Tutorial. That is, at step 120, the processor is further configured to determine fit guidance parameters. Examples of fit guidance parameters can include but are not limited to loose fitting top, high-waisted jeans, and scoop-necked T-shirts. For instance, in an embodiment, one of the fit guidance parameters can be “tutorial”. Such tutorial parameter can cause a particular tutorial paragraph, a particular tutorial video, or a particular tutorial audio file to get sent to, displayed, or streamed for the user. It should be appreciated that other media formats, such as braille for instance, are also contemplated and included. In an embodiment, the tutorial can guide the user through the system, offering suggestions for how to use the application as well as what each parameter means. For example and consistent with embodiments herein, the tutorial can include a demonstration (demo) of a fictitious user from start to finish. That is, such fictitious user can begin the application by choosing to purchase one or more items to wear for a particular occasion. For instance, the demo can show a person wanting to put together an outfit for their first day at a new job.
[0060] In accordance with embodiments herein, other example fit guidance parameters include but are not limited to age grouping parameters (e.g., 18-24, 41 -49, and 70+), pain points parameters (e.g., petite, difficult to fit, and temporary injury), body shape challenges parameters (e.g., broad shoulders, thick waist, and long arms), size categories parameters (e.g., XXS-XS / 00-2 Women (W), L12-24 W, L / 34-36 Waist Men (M), and XL Tall (M)), and fit styles parameters (e.g., tailored, bodycon, classic, and fitted).
[0061] 4. Where and how is clothing is manufactured. Why this matters. That is, at step 130, the processor is further configured to determine clothing processing parameters. Examples of clothing processing parameters can include but are not limited to sustainably-vetted, fair-trade-vetted, made in Italy, made in France, made inthe UK, made in the USA. Other example clothing processing parameters include but are not limited to clothing care parameters (e.g., machine washable, dryer-friendly, and wrinkle resistant) and ethics parameters (e.g., natural fibers, breathable, and adjustable).
[0062] 5. Essentials of a successful wardrobe. The items themselves. That is, at step 140, the processor is further configured to determine the item parameters representing the clothing items themselves. Examples of the item parameters can include but are not limited to a particular white blouse made in the USA and a particular skirt made in France. Other example item parameters include but are not limited to wardrobe parameters (e.g., minimalist, capsule, and everyday casual) and aesthetic parameters (e.g., classic, modern, and conversative).
[0063] 6. Budget. How much a consumer or user should be spending. That is, at step 150, the processor is further configured to determine budget parameters. Examples of the budget parameters can include but are not limited to not more than x- amount of US dollars, between 1000 Euros and 1500 Euros. Other example budget parameters include but are not limited to currency ranges (e.g., $0-$50, $50-$100, and $400-$500), digital currency (e.g., Bitcoin), and distributed ledgers (e.g., blockchain- enabled payment). In addition, in accordance with an embodiment, the system can increase the number of wears for a particular item to be more conscientious of the customer’s spending limitations, thereby creating more value.
[0064] 7. The art of shopping. How to shop. Where, when, how are all specific to the user. That is, at step 160, the processor is further configured to determine art of shopping parameters. Examples of the art of shopping parameters can include but are not limited to a season parameter to indicate fall, winter, spring, or summer. Other example art of shopping parameters can include the wardrobe parameters such as for example wedding, weekend active, and travel. That is, in an embodiment, certain of the wardrobe parameters (e.g., travel) can be used by the system to activate the application or a notification to open the application, based on a predetermined date associated with a particular time of travel. In addition and in accordance with embodiments herein, the system is configured to include a conversational NLP mechanism to provide inputs that refine the prediction model.
[0065] Each of the above steps is connected through a questionnaire built through logic / if :then code, with each answer triggering a filter that narrows down the items / clothing that the user sees in their profile. The backend of this process is a brand or item tagging system, which links to the questionnaire.
[0066] Further, these input parameters are used to train the Al engine. The trained Al engine, in tandem with tracking the returns, then learns to map user input (the user’s answers) better to more closely related parameters. The more closely related parameters improve the accuracy of the ultimate clothing recommendations for the user in a subsequent use of the application to purchase clothing items.
[0067] As an example implementation, the user’s initial input for size category is XXS-XS / 00-2 W and is part of a first training set that is used to train the Al engine. The user purchases an item of such size. The first training set comprises the parameters that were set by the user’s input into the system. The Al engine is thus trained using this first training set. At some later point in time, the user returns such purchased item, because it is too small. The user returns the item for a second item, but from size category S / 4-6. The Al monitors the returned item for the user. That is, the Al reads or parses the parameters of the returned item. In addition, the Al engine learns or parses the parameters of the newly purchased second item, including the second item’s size category S / 4-6. Such learned or parsed parameters of the second purchased, including the second item’s size category S / 4-6, are used to generate a second training set for a second stage of training. The second stage of training comprises the first training set and the data reflecting the returned item, that was incorrectly detected as viable after the first training set. Subsequently, the Al engine is trained in a second stage using the second training set based on the returned item.
[0068] It should be appreciated that while some parameters are listed under particular sections, such structure is by way of example embodiment and is not meant to be limiting. For example, the fit styles parameter is listed under the fit guidance step. However, the fit styles parameter could be listed under the wardrobe category and be consistent with embodiments herein.
[0069] In addition, the enumeration of seven steps is by way of an example and is not meant to be limiting. Other structures, such as for example combining the categorical steps, eliminating categorial steps, renaming categorical steps, reorganizingcategorial steps, and adding categorical steps are contemplated and consistent with embodiments herein.
[0070] Example:
[0071] If the user chooses option 1 in drop-down menu, option 1 connects to items tagged as suitable for option 1. For example, in a non-initial training stage of the Al engine, the Al engine generates the tagged inventory items, based on the most recent training session of the Al engine. For instance, continuing with the example above, option 1 might initially connect to and display a tagged item of size XXS-XS / 00-2 W. Subsequently, after further training sessions of the Al engine, option 1 can connect to a tagged item that is of size category S / 4-6, based on the most recent training of the Al engine.
[0072] Example:Question: What are you shopping for today?Megan (end user) answers: Pants.Further specifications:Length: above ankle.Rise: midrise.Care: machine washable, no preference for material content or location of production.
[0073] In an embodiment, the system is configured to have the Al engine connect to items that match the above tags, recommending the Frank and Eileen clothing line.Data Points
[0074] An embodiment can be understood with reference to FIG. 1 B-1 C. FIG. 1 B- 1 C depict a table showing example data points used by the system to achieve creating a personal appearance to engender a positive self-image, according to an embodiment. In an embodiment, the categories include but are not limited to How Do You Want to Feel 170; Age Grouping 172; Life Phase (174); Personality Type 176; Pain Points 178; Body Shape Challenges 180; Fit Styles 182; Size Categories 184; Budget Range 186; Clothing Care 188; Ethics 190; Wardrobe 192; Aesthetic 194; and Aspiration 196. Inthe embodiment, each member of each category represents a tag associated with one or more clothing items. For example, in Life Phase 174, one tag is From student to first real job. Another example tag is Friendly from Personality Type 176. Yet another example is Timeless under Wardrobe 192.Item Classification System
[0075] FIG. 2 is a block diagram showing an item classification system for use in creating a personal appearance to engender a positive self-image according to the invention. Embodiments of the invention provide personalized shopping carts that are adapted to run on any platform. The user starts with a simple and interactive quiz 220 with an avatar. The quiz includes questions concerning, for example, style personality and shape and color. An example of such quiz includes but is not limited to that in FIG. 1 A. Examples of the questions and answers can concern the categories and elements (e.g., tags) in FIG. 1 B-1 C. The user’s journey provides a story line. The user can also upload a photograph, or their Instagram Facebook handle and the Al Engine does the rest.
[0076] In embodiments the avatar has a name, such as Audrey. The avatar resembles a human assistant and has a voice that is similar to that of a human assistant. For example, the illustrated person in 220 can be Audrey and the voice can be broadcast through the phone’s speakers. The avatar provides a user interface to an artificial intelligence (Al) system 200 by which a customized recommendation is created for the user.Process. Labeling system, then the system sorts through the tags and identifies which tags match the user profile.
[0077] In embodiments of the invention artificial intelligence learns based on the purchases that the user has kept and tracks the information from items the user has returned. Embodiments of the invention typically engage with people shopping 210 at high end fashion retailers. Uniquely, no other artificial intelligence learns from a user’s successful purchases. The user makes purchases 230 based upon the recommendations. Examples of the recommendations include but are not limited to tagged clothing items which the innovative trained Al engine had matched with the user’s profile. Initially such recommendations are based on the results of the useranswering simple and interactive quiz 220. In subsequent recommendations, the Al engine can combine results of a simple and interactive quiz 220 with non-initial training set. Further, the recommendations can include results based on the Al engine’s predictions using non-initial training sets based on the user’s returns. Further, the Al Engine tracks purchases and returns 240 for each user and learns from individual and group purchase history. Significantly, embodiments of the invention reduce return rates by initially determining what to recommend based on application of the seven-step process and by learning from each return, e.g. why the item was returned.
[0078] Companies such as Stitch Fix and Trunk Club only address surface level questions, the same questions that lead to the billions of dollars of merchandise returns and frustrated customers. Relying on antiquated definitions of body shape, style preference (boho, rock, feminine, for example) is no way to help a person find what they want or actually need. It relies on these antiquated experiences that make us feel oversimplified and bored - which lead many time to returns, while ignoring the questions and information that would actually help us find something new and exciting, and that would fit our bodies and prepare us for our next phase of life, our aspirations - which leads to keeping the item instead of returning the item. In an embodiment, the personal assistant avatar is configured to ask a user, e.g., Edith, who you are, where you are going, who you want to be, while adding the truly helpful information such as where you carry your weight, what you are used to spending and why, what has or has not worked for you in the past, and creating a wardrobe for you based on that information by using the innovative Al engine. It is key, and it is what makes people delighted, excited and eager to spend and reduces returns.Edith: “I think of getting dressed in the morning as costuming myself for the story I want to live that day.”
[0079] Embodiments of the invention learn:1 ) Your mindset about your own body.2) What phase of life are you in? What transition? Where are you going in your life? What do you aspire to?3) What makes you feel great? What’s special about you, e.g. athletic body, energy, colorful personality.4) Fit, let’s talk about body shape and show you examples of how to know when an item fits or does not.5) Where and how each brand makes their stuff, and whether or not it matters to you.6) Essential items you should own to reach your aspirations in daily life and beyond.7) Budget. How much have you typically spent in the past and why? Information on when to spend more and when to spend less.8) The art of shopping. What time of year, when, how, your brand recommendations, etc., all specific to you.
[0080] The Al engine is configured to face the user in the form of an avatar and learns based on the informational data (e.g., answers) to personality questions and questions about the user’s goals. This is similar to how a user would interact with a personal stylist. The user is not made to feel imposed upon because the questions do not concern anything that can be personally tracked back to them, which might be a concern for security reasons. The algorithm implements artificial intelligence that wants to know, for example, such information as the user’s goals, aspirations, do they have children, what are their simple needs, and what are their aspirational needs.
[0081] How is the Al engine able to sense the habits of people, giving them recommendations far beyond what they’ve encountered before?• By tracking the key labels on each item that they have kept vs. what they have returned.• A brief questionnaire asking the end user why they returned the item, e.g. fit, size, style, fabric, etc.• Draw out the customer process. How the shopper experiences this product.
[0082] Who are the target implementers or customers (e.g., business to business (B to B)) who integrate and purchase this software?
[0083] In embodiments of the invention the Al engine is configured to learn everything about what a person needs, creating delight and eagerness to spend, where there is typically frustration and sales loss and returns. This approach is especially helpful for people shopping at high end fashion retailers such as Nordstrom or Net-a-porter, but embodiments of the invention also apply to other industries or knowledge domains such as but not limited to auto shops, hardware stores, and Amazon. The experience is as if the retailer has an army of Audreys (e.g., digital personal assistants or avatars) selling for them all day long, and the customer or user can experience Audrey’s level of personalized service. One trick is in having the Al engine be configured such as to be incredibly dexterous and able to sense the habits in people, giving them recommendations far beyond what they have simply purchased before.
[0084] Following is a complete example use case of a user, Edith, using the innovative system.Edith: one use case If / then
[0085] Ten categorized types: [Here, on a digital display. Edith is presented with ten categorization types that were predefined in the system.]If: Single professional [Here, Edith had selected a particular category: single professional.]Then: filter all brands tagged as: [Here, the system collects brands that were previously tagged for such categorized type.]Q: What life transition are you in? [Here, the system presents (e.g., on the display or could be audio, etc.) a question to Edith.]If: Leveling up (I want higher pay, more respect, higher likelihood of finding love) [Here, Edith had selected leveling up.]Q: What can I help you find today? [Here, based on the selected leveling up, the innovative system presents this question to the user.]If: Work clothes [Here, Edith had selected work clothes.]Then: What kind of work do you do? [Here, the system consequently asks a question, as shown.]
[0086] It should be appreciated that one skilled in the art would recognize the pattern here of the innovative system executing predetermined paths based on if / then scenarios. The if / then scenarios are powered by the user’s answers to the presented questions. The innovative Al engine is trained continuously based on the user’sanswers. The trained Al engine presents increasingly relevant and refined questions to the user as the Al engine continues to be trained. The further trained Al then presents further refined and more relevant questions and recommendations.Edith: one use case If / then - continued:If: Indoors, sitting at a desk most of the day.Then: Is it cold in your office? Do you always find yourself needing layers?Then: body image> What are your pain points?If: I always hate my body when I try on clothes. Everything makes me wish I was skinnier.Then: filter out all brands tagged as: [previously tagged to correlate with items that could make a person feel that they are too fat or large]Q: tell us about your body shape. Where do you carry your weight? Proportions?If: I’m narrow in the hips but broad in the shoulders. Most things make me look top heavy and my shoulders constricted.Q: what kind of footwear do you usually prefer to work in?If: flats, but sometimes a nice heel.Then: what style preferences do you have? Absolutely no bare arms, cardigans vs. blazers, pant fit, wide vs straight, ankles showing or floor length, etc.Q: Clothing Care: dry cleaning?If: Yes but only two or three times per year.Then: Filter out all items that require dry cleaning other than suits, outerwear and sweaters.Q: Are you willing to use alterations?Include information about dry cleaning best practices for different types of items.Include cleaning information for other items as well.Brand assessment: What brands already work well for you?If: I’m not crazy about anything I currently have.Pants:Tops:Shoes:Then: Are there any brands you hate? That really don’t work for you?If: [depends on the brands that user inputs]Sizing: what sizes are you in those brands?
[0087] The system is configured to present an Information section on how to know if something fits. Images showing difference between a good fitting top and ill fitting, pants, jackets, etc. How to know when you have got the right fit.Budget: What are you accustomed to spending on work clothing items?What compels you to spend more?If: great quality fabric, great fit, or made in America.Then: [The innovative systems presents the following recommended exercise.]Exercise: Think of your clothing as an investment in your future self. You are clothing yourself for your dreams, your future, your aspirations. Invest in killer items that elevate, and less quantity.Shopping ethics: Do you care where your clothing is made?Fabric sourcing?Would you be willing to spend more on items that are ethically produced?If: I prefer to buy “Made in America” but am open to other ethically produced brands.Add ons!Hey, you, you are doing incredible things with your life and you deserve a great handbag to complete your look.Undergarments? How about a great pair of Spanks?Pajamas?Special designer splurge item?
[0088] It should be appreciated that in the above use case, as the use case illustrates the user-facing display outputs (questions and inputs), the system is configured to filter continually the collection of items based on the user input. The system associates the input by the user with tags and then uses such tags to weed out or remove from the present collection the items based on those tags.
[0089] Further, in an embodiment, the Al engine can be configured to use a trained large language model (LLM) to generate a plurality of candidate prompts (e.g., instruction sets or questions including the questions above), automatically score the candidates, and select the top candidate or candidates, depending.Daily Needs Vs. Aspirational Needs
[0090] In an embodiment, labelling each item in a store’s inventory is as follows:
[0091] Customized key categories / personalities (proprietary definitions of customer types / tags) upon thorough review of store inventory (embodiments of the invention automate this with increasingly sophisticated computer vision). The system is configured to link (automatically or manually) the personality tags to each item in the store’s system, allowing the quiz questionnaire results to filter through the tags, leaving only the items that match the quiz results for that day.
[0092] In an embodiment, the generative Al engine is trained to recommend tags for items, analogously to a generative Al engine generating recommended prompts or instruction sets.
[0093] In embodiments of the invention, the Al / tagging system learns from each customer which item types they return or do not like in their returns quiz, further narrowing down the recommendations each time the customer shops.In-Store App Functionality
[0094] In-store, a hardware / software application (app) serves as an on-site guide for the employee, enabling them to augment the process for the customer. The app is configured to track how the customer wants to feel, track what the customer has already purchased, offer a sizing platform, branding recommendations, colors, new styles, price points, etc. The Al engine provides instructional information (e.g., tutorials) and information concerning mindfulness of quotes to offer and a printable / shareable profile for the client.
[0095] There is a mindfulness / encouragement aspect to the invention. How does this factor in? There is a shareable profile for the client (e.g., manifested as a digital profile stored in the app, the client’s client device, and system 200). There is a separate set of inspirational / encouraging quotes matched with each of the core personalities. The customer is surrounded by positive vibes when looking at their page, reminded of how badass they are, and how capable and deserving they are of looking and feeling phenomenal.
[0096] Embodiments of the invention provide the customer with a profile that they can use outside of the computer. They can access it at brick and mortar stores too. Or they can print and share it online with friends.
[0097] In embodiments of the invention there is a social aspect. Customers can share and network with their friends through the system. A friend can go in and see another friend’s profile and make recommendations through such profile.Online App Functionality
[0098] Every site takes the customer through a quiz based on feeling. Using helpful, inspirational quotes, likening it to headspace; considering mindset, fit guides, e.g., what doesn’t fit vs. what does. Customer education is integrated into the shopping process. The customer can opt out if they choose to do so. Initially, the customer is greeted with a guide. How do you want to feel today?System Architecture
[0099] FIG. 3A is an architectural diagram showing an item classification system for use in creating a personal appearance to engender a positive self-image according to the invention. In FIG. 3A, a customer 350 interacts with an Al engine 352. For the seven step process the customer is guided through a questionnaire 353 and via lf:Then logic 354, as discussed above, the Al engine learns about the customer and can begin making recommendations to the customer. The Al engine also uses a user profile 355 to incorporate additional information about the customer into the recommendation process. As part of this process the Al engine also interacts with a brand tagging module 357, again as described above. User interaction with the system is tracked. Accordingly, user purchases and returns are noted 356 and also used by the Al engine in the recommendation process. Finally, community information 359 comprising similarly situated customers and products is also used to train and update the Al engine. In this way, the Al engine is subjected to both supervised learning and continual unsupervised learning based on varying inputs from disparate data sources.
[0100] The customer’s goal is to purchase the clothing best suited for them which minimizes returns because of the high level of satisfaction the customer has with their purchase. A key aspect of the invention is the customer’s interaction with a retailer 358, either on-line or in the store itself. Once the Al engine is trained it makes recommendations to the customer while the customer is shopping. As such, the Al engine also ingests information from the retailer and provides information to the retailer, for example to make recommendations to the retailer’s customers. In this way, the Al engine continues to improve its recommendations by unsupervised learning that results in machine learning.
[0101] For purposes of the discussion herein machine learning involves algorithms that build a model based on sample data, known as training data, to make predictions or decisions without being explicitly programmed to do so. Machine learning algorithms are used in a wide variety of applications, such as in medicine, email filtering, speech recognition, agriculture, and computer vision, where it is difficult or unfeasible to develop conventional algorithms to perform the needed tasks.
[0102] Further, as used herein supervised learning (SL) is a machine learning paradigm for problems where the available data consists of labeled examples, meaningthat each data point contains features (covariates) and an associated label. The goal of supervised learning algorithms is learning a function that maps feature vectors (inputs) to labels (output), based on example input-output pairs. It infers a function from labeled training data consisting of a set of training examples. In supervised learning, each example is a pair consisting of an input object (typically a vector) and a desired output value (also called the supervisory signal). A supervised learning algorithm analyzes the training data and produces an inferred function, which can be used for mapping new examples. An optimal scenario allows for the algorithm to correctly determine the class labels for unseen instances. This requires the learning algorithm to generalize from the training data to unseen situations in a “reasonable” way. This statistical quality of an algorithm is measured through the so-called generalization error.
[0103] Further, as used herein unsupervised learning is a type of algorithm that learns patterns from untagged data. The goal is that through mimicry, which is an important mode of learning in people, the machine is forced to build a concise representation of its world and then generate imaginative content from it.
[0104] In contrast to supervised learning where data is tagged by an expert, e.g. tagged as a “ball” or “fish,” unsupervised methods exhibit self-organization that captures patterns as probability densities or a combination of neural feature preferences encoded in the machine’s weights and activations. The other levels in the supervision spectrum are reinforcement learning where the machine is given only a numerical performance score as guidance, and semi-supervised learning where a small portion of the data is tagged.An Exemplary Al Architecture
[0105] An embodiment of the Al engine can be understood with reference to FIG. 3B. FIG. 3B is an Al architecture, according to an embodiment. As shown, the Al system 300 can include a set of layers, which conceptually organize elements within an example network topology for the Al system’s architecture to implement a particular Al model 330. Generally, an Al model 330 is a computer-executable program implemented by the Al system 300 that analyses data to make predictions. Information can pass through each layer of the Al system 300 to generate outputs for the Al model 330. The layers can include a data layer 302, a structure layer 304, a model layer 306, and an application layer 308. The algorithm 316 of the structure layer 304 and themodel structure 320 and model parameters 322 of the model layer 306 together form the example Al model 330. The optimizer 326, loss function engine 324, and regularization engine 328 work to refine and optimize the Al model 330, and the data layer 302 provides resources and support for application of the Al model 330 by the application layer 308.
[0106] The data layer 302 acts as the foundation of the Al system 300 by preparing data for the Al model 330. As shown, the data layer 302 can include two sub-layers: a hardware platform 310 and one or more software libraries 312. The hardware platform 310 can be designed to perform operations for the Al model 330 and include computing resources for storage, memory, logic and networking, such as the resources described in relation to FIG. 3A. The hardware platform 310 can process amounts of data using one or more servers. The servers can perform backend operations such as matrix calculations, parallel calculations, machine learning (ML) training, and the like. Examples of servers used by the hardware platform 310 include central processing units (CPUs) and graphics processing units (GPUs). CPUs are electronic circuitry designed to execute instructions for computer programs, such as arithmetic, logic, controlling, and input / output (I / O) operations, and can be implemented on integrated circuit (IC) microprocessors. GPUs are electric circuits that were originally designed for graphics manipulation and output but may be used for Al applications due to their vast computing and memory resources. GPUs use a parallel structure that generally makes their processing more efficient than that of CPUs. In some instances, the hardware platform 310 can include Infrastructure as a Service (laaS) resources, which are computing resources, (e.g., servers, memory, etc.) offered by a cloud services provider. The hardware platform 310 can also include computer memory for storing data about the Al model 330, application of the Al model 330, and training data for the Al model 330. The computer memory can be a form of random-access memory (RAM), such as dynamic RAM, static RAM, and non-volatile RAM.
[0107] The software libraries 312 can be thought of as suites of data and programming code, including executables, used to control the computing resources of the hardware platform 310. The programming code can include low-level primitives (e.g., fundamental language elements) that form the foundation of one or more low- level programming languages, such that servers of the hardware platform 310 can use the low-level primitives to carry out specific operations. The low-level programminglanguages do not require much, if any, abstraction from a computing resource’s instruction set architecture, allowing them to run quickly with a small memory footprint. Examples of software libraries 312 that can be included in the Al system 300 include Intel Math Kernel Library, Nvidia cuDNN, Eigen, and Open BLAS.
[0108] The structure layer 304 can include an ML framework 314 and an algorithm 316. The ML framework 314 can be thought of as an interface, library, or tool that allows users to build and deploy the Al model 330. The ML framework 314 can include an open-source library, an application programming interface (API), a gradient-boosting library, an ensemble method, and / or a deep learning toolkit that work with the layers of the Al system facilitate development of the Al model 330. For example, the ML framework 314 can distribute processes for application or training of the Al model 330 across multiple resources in the hardware platform 310. The ML framework 314 can also include a set of pre-built components that have the functionality to implement and train the Al model 330 and allow users to use pre-built functions and classes to construct and train the Al model 330. Thus, the ML framework 314 can be used to facilitate data engineering, development, hyperparameter tuning, testing, and training for the Al model 330. Examples of ML frameworks 314 that can be used in the Al system 300 include TensorFlow, PyTorch, Scikit-Learn, Keras, Cafffe, LightGBM, Random Forest, and Amazon Web Services.
[0109] The algorithm 316 can be an organized set of computer-executable operations used to generate output data from a set of input data and can be described using pseudocode. The algorithm 316 can include complex code that allows the computing resources to learn from new input data and create new / modified outputs based on what was learned. In some implementations, the algorithm 316 can build the Al model 330 through being trained while running computing resources of the hardware platform 310. This training allows the algorithm 316 to make predictions or decisions without being explicitly programmed to do so. Once trained, the algorithm 316 can run at the computing resources as part of the Al model 330 to make predictions or decisions, improve computing resource performance, or perform tasks. The algorithm 316 can be trained using supervised learning, unsupervised learning, semi-supervised learning, and / or reinforcement learning.
[0110] Using supervised learning, the algorithm 316 can be trained to learn patterns (e.g., map input data to output data) based on labeled training data. The training data may be labeled by an external user or operator. For instance, a user may collect a set of training data, such as by capturing data from sensors, images from a camera, outputs from a model, and the like. In an example implementation, training data can include but are not limited to the data from the lookup table in FIG. 1 B-1 C; freeform user input data which are amply shown and discussed herein; detailed return information, and also amply described herein. The user may label the training data based on one or more classes and trains the Al model 330 by inputting the training data to the algorithm 316. The algorithm determines how to label the new data based on the labeled training data. The user can facilitate collection, labeling, and / or input via the ML framework 314. In some instances, the user may convert the training data to a set of feature vectors for input to the algorithm 316. Once trained, the user can test the algorithm 316 on new data to determine if the algorithm 316 is predicting accurate labels for the new data. For example, the user can use cross-validation methods to test the accuracy of the algorithm 316 and retrain the algorithm 316 on new training data if the results of the cross-validation are below an accuracy threshold.
[0111] Supervised learning can involve classification and / or regression. Classification techniques involve teaching the algorithm 316 to identify a category of new observations based on training data and are used when input data for the algorithm 316 is discrete. Said differently, when learning through classification techniques, the algorithm 316 receives training data labeled with categories (e.g., classes) and determines how features observed in the training data (e.g., from natural language user input answers and from details of the return data) relate to the categories (e.g., Size Categories 184). Once trained, the algorithm 316 can categorize new data by analyzing the new data for features that map to the categories. Examples of classification techniques include boosting, decision tree learning, genetic programming, learning vector quantization, k-nearest neighbor (k-NN) algorithm, and statistical classification.
[0112] Regression techniques involve estimating relationships between independent and dependent variables and are used when input data to the algorithm 316 is continuous. Regression techniques can be used to train the algorithm 316 to predict or forecast relationships between variables. To train the algorithm 316 using regression techniques, a user can select a regression method for estimating theparameters of the model. The user collects and labels training data that is input to the algorithm 316 such that the algorithm 316 is trained to understand the relationship between data features and the dependent variable(s). Once trained, the algorithm 316 can predict missing historic data or future outcomes based on input data. Examples of regression methods include linear regression, multiple linear regression, logistic regression, regression tree analysis, least squares method, and gradient descent. In an example implementation, regression techniques can be used, for example, to estimate and fill-in missing data for machine-learning based pre-processing operations.
[0113] Under unsupervised learning, the algorithm 316 learns patterns from unlabeled training data. In particular, the algorithm 316 is trained to learn hidden patterns and insights of input data, which can be used for data exploration or for generating new data. Here, the algorithm 316 does not have a predefined output, unlike the labels output when the algorithm 316 is trained using supervised learning. Said another way, unsupervised learning is used to train the algorithm 316 to find an underlying structure of a set of data, group the data according to similarities, and represent that set of data in a compressed format. The system 200 or platform can use unsupervised learning to identify patterns in the questionnaire 353 and its corresponding user input (e.g., to identify particular items by a particular retailer 358) and so forth. In some implementations, performance of the Al engine 352 that can use unsupervised learning is improved, as described herein.
[0114] A few techniques can be used in supervised learning: clustering, anomaly detection, and techniques for learning latent variable models. Clustering techniques involve grouping data into different clusters that include similar data, such that other clusters contain dissimilar data. For example, during clustering, data with possible similarities remain in a group that has less or no similarities to another group. Examples of clustering techniques density-based methods, hierarchical based methods, partitioning methods, and grid-based methods. In one example, the algorithm 316 may be trained to be a k-means clustering algorithm, which partitions n observations in k clusters such that each observation belongs to the cluster with the nearest mean serving as a prototype of the cluster. Anomaly detection techniques are used to detect previously unseen rare objects or events represented in data without prior knowledge of these objects or events. Anomalies can include data that occur rarely in a set, a deviation from other observations, outliers that are inconsistent with the rest of the data,patterns that do not conform to well-defined normal behavior, and the like. When using anomaly detection techniques, the algorithm 316 may be trained to be an Isolation Forest, local outlier factor (LOF) algorithm, or K-nearest neighbor (k-NN) algorithm. Latent variable techniques involve relating observable variables to a set of latent variables. These techniques assume that the observable variables are the result of an individual’s position on the latent variables and that the observable variables have nothing in common after controlling for the latent variables. Examples of latent variable techniques that may be used by the algorithm 316 include factor analysis, item response theory, latent profile analysis, and latent class analysis.
[0115] In an embodiment, the model layer 306 implements the Al model 330 using data from the data layer and the algorithm 316 and ML framework 314 from the structure layer 304, thus enabling decision-making capabilities of the Al system 300. The model layer 306 includes a model structure 320, model parameters 322, a loss function engine 324, an optimizer 326, and a regularization engine 328.
[0116] The model structure 320 describes the architecture of the Al model 330 of the Al system 300. The model structure 320 defines the complexity of the pattern / relationship that the Al model 330 expresses. Examples of structures that can be used as the model structure 320 include decision trees, support vector machines, regression analyses, Bayesian networks, Gaussian processes, genetic algorithms, and artificial neural networks (or, simply, neural networks). The model structure 320 can include a number of structure layers, a number of nodes (or neurons) at each structure layer, and activation functions of each node. Each node’s activation function defines how to node converts data received to data output. The structure layers may include an input layer of nodes that receive input data, an output layer of nodes that produce output data. The model structure 320 may include one or more hidden layers of nodes between the input and output layers. The model structure 320 can be an Artificial Neural Network (or, simply, neural network) that connects the nodes in the structured layers such that the nodes are interconnected. Examples of neural networks include Feedforward Neural Networks, convolutional neural networks (CNNs), Recurrent Neural Networks (RNNs), Autoencoder, and Generative Adversarial Networks (GANs).
[0117] The model parameters 322 represent the relationships learned during training and can be used to make predictions and decisions based on input data. Themodel parameters 322 can weight and bias the nodes and connections of the model structure 320. For instance, when the model structure 320 is a neural network, the model parameters 322 can weight and bias the nodes in each layer of the neural networks, such that the weights determine the strength of the nodes and the biases determine the thresholds for the activation functions of each node. The model parameters 322, in conjunction with the activation functions of the nodes, determine how input data is transformed into desired outputs. The model parameters 322 can be determined and / or altered during training of the algorithm 316.
[0118] The loss function engine 324 can determine a loss function, which is a metric used to evaluate the Al model’s 330 performance during training. For instance, the loss function engine 324 can measure the difference between a predicted output of the Al model 330 and the actual output of the Al model 330 and is used to guide optimization of the Al model 330 during training to minimize the loss function. The loss function may be presented via the ML framework 314, such that a user can determine whether to retrain or otherwise alter the algorithm 316 if the loss function is over a threshold. In some instances, the algorithm 316 can be retrained automatically if the loss function is over the threshold. Examples of loss functions include a binary-cross entropy function, hinge loss function, regression loss function (e.g., mean square error, quadratic loss, etc.), mean absolute error function, smooth mean absolute error function, log-cosh loss function, and quantile loss function.
[0119] The optimizer 326 adjusts the model parameters 322 to minimize the loss function during training of the algorithm 316. In other words, the optimizer 326 uses the loss function generated by the loss function engine 324 as a guide to determine what model parameters lead to the most accurate Al model 330. Examples of optimizers include Gradient Descent (GD), Adaptive Gradient Algorithm (AdaGrad), Adaptive Moment Estimation (Adam), Root Mean Square Propagation (RMSprop), Radial Base Function (RBF) and Limited-memory BFGS (L-BFGS). The type of optimizer 326 used may be determined based on the type of model structure 320 and the size of data and the computing resources available in the data layer 302.
[0120] The regularization engine 328 executes regularization operations. Regularization is a technique that prevents over- and under-fitting of the Al model 330. Overfitting occurs when the algorithm 316 is overly complex and too adapted to thetraining data, which can result in poor performance of the Al model 330. Underfitting occurs when the algorithm 316 is unable to recognize even basic patterns from the training data such that it cannot perform well on training data or on validation data. The regularization engine 328 can apply one or more regularization techniques to fit the algorithm 316 to the training data properly, which helps constraint the resulting Al model 330 and improves its ability for generalized application. Examples of regularization techniques include lasso (L1 ) regularization, ridge (L2) regularization, and elastic (L1 and L2 regularization).
[0121] The application layer 308 describes how the Al system 300 is used to solve problems or perform tasks. In an example implementation, the application layer 308 can include the Al engine 352 of the system 200.Computer Implementation
[0122] FIG. 4 is a block diagram of a computer system as may be used to implement certain features of some of the embodiments. The computer system may be a server computer, a client computer, a personal computer (PC), a user device, a tablet PC, a laptop computer, a personal digital assistant (PDA), a cellular telephone, an iPhone, an iPad, a Blackberry, a processor, a telephone, a web appliance, a network router, switch or bridge, a console, a hand-held console, a (hand-held) gaming device, a music player, any portable, mobile, hand-held device, wearable device, or any machine capable of executing a set of instructions (sequential or otherwise) that specify actions to be taken by that machine.
[0123] The computing system 400 may include one or more central processing units (“processors”) 403, memory 410, input / output devices 425, e.g. keyboard and pointing devices, touch devices, display devices, storage devices 420, e.g. disk drives, and network adapters 430, e.g. network interfaces, that are connected to an interconnect 415. The interconnect 415 is illustrated as an abstraction that represents any one or more separate physical buses, point to point connections, or both connected by appropriate bridges, adapters, or controllers. The interconnect 415, therefore, may include, for example, a system bus, a Peripheral Component Interconnect (PCI) bus or PCI-Express bus, a HyperTransport or industry standard architecture (ISA) bus, a small computer system interface (SCSI) bus, a universal serial bus (USB), IIC (12C) bus, oran Institute of Electrical and Electronics Engineers (IEEE) standard 1394 bus, also called Firewire.
[0124] The memory 410 and storage devices 420 are computer-readable storage media that may store instructions that implement at least portions of the various embodiments. In addition, the data structures and message structures may be stored or transmitted via a data transmission medium, e.g. a signal on a communications link. Various communications links may be used, e.g. the Internet, a local area network, a wide area network, or a point-to-point dial-up connection. Thus, computer readable media can include computer-readable storage media, e.g. non-transitory media, and computer-readable transmission media.
[0125] The instructions stored in memory 410 can be implemented as software and / or firmware to program the processor 405 to carry out actions described above. In some embodiments, such software or firmware may be initially provided to the processing system 400 by downloading it from a remote system through the computing system 400, e.g. via network adapter 430.
[0126] The various embodiments introduced herein can be implemented by, for example, programmable circuitry, e.g. one or more microprocessors, programmed with software and / or firmware, or entirely in special-purpose hardwired (non-programmable) circuitry, or in a combination of such forms. Special-purpose hardwired circuitry may be in the form of, for example, one or more ASICs, PLDs, FPGAs, efc.Avatar Overlay
[0127] An embodiment can be understood with reference to FIG. 5. FIG. 5 is a schematic diagram showing measurement points generation used to create an avatar on which a garment is overlayed, according to an embodiment. System 200 generates an avatar 1 associated with a user. A camera such as a smartphone camera scans the user’s body. The camera is in communication with a measuring tool. The tool maps particular features of the body to eight measurement points: shoulder; bust; waist; hip; thigh; knee; calf; and feet. It should be appreciated that other measurement points (e.g., neck size) are contemplated herein this discussion. Also measured are ratio and width values for the particular user and for the associated avatar. Also, system 200 is configured to receive garment ratios and fabric values. Such garment ratios and fabricvalues are generated by camera scans for example but not limited to such. In an embodiment, the camera is configured to scan an image of the garment and determine therefrom certain proportions and fabric content. In other embodiment, the garment ratio and fabric values are entered manually. Based on the generated avatar and the generated garment and ratio fabric values, system 200 is configured to overlay such garment over the avatar. Such overlay demonstrates to the shopper what the item would look like on them. Further, it should be appreciated that this innovative integration of the eight measurement points distinguishes improvements over certain traditional types. For example, the eight measurement points more accurate describe a person’s body type rather than the current apple, pear, straight, or hourglass body types.
[0128] An embodiment can be understood with reference to FIG. 7. FIG. 7 is a flow diagram showing an example process for generating measurement points used to create an avatar on which a garment is overlayed, according to an embodiment. At step 710, the body scanner takes measurements, creating a proportionally accurate avatar. At step 720, a second scanner measures portions of the garment and tracks fabric content (e.g., in the back-end of system 200). The garment measurements are associated with or tagged to the eight measurement points. At step 730, the shopper avatar is shown virtually wearing the clothing item, demonstrating how the item would fit the shopper.An Exemplary Embodiment for Creating a Personal Appearance to Engender a Positive Self-Image
[0129] An embodiment can be understood with reference to FIG. 6. FIG. 6 is a flow diagram showing an example process for creating a personal appearance to engender a positive self-image, according to an embodiment. In accordance with embodiments herein, the initial questionnaire (diagnostic / triangulation) is of needs that are met and not met. Building on what already works for the customer and addressing what has not worked for the customer in the past.
[0130] System 200 is configured to capture women characters 602 which are categories of who individuals are and who they want to be. Examples are depicted but are not limited to shown in the matrix, which includes young professional-edgy; cool fun mom, and the like. System 200 is further configured to receive and process personal information 604. Examples of personal information 604 include but are not limited tothe categories Personality / getting to know you, Body image, and Body shape. Further, system 200 is configured to receive data indicative of style preferences 606. Examples of style preferences 606 include but are not limited to pant rise, pant length, and sleeve lengths. System 200 is further configured to receive data indicative of brand assessment 608. For instance, brand assessment 608 includes but is not limited to brands that one already loves and works for them, brands they really do not like, and styles which one already knows do not work for them. System 200 is further configured to receive data indicative of sizing 610. The user can be asked questions such as but not limited to what your sizes in those brands are, both the good and the bad sizes. System 200 is further configured to be triggered to present on the display or via audio or any other medium information to educate the user how to know when something fits them. System 200 is further configured to receive data indicative of a budget section 612. Examples of instructions or questions used as an aid to obtain this information from the user include but are not limited to asking what are you accustomed to spending? When do you feel compelled to spend more? Further the budget section 612 processor can be configured to be triggered to present on the display or via audio or any other medium information to educate the user about sale psychology. System 200 is further configured to receive data indicative of subcategories 614. For instance, the subcategories 614 processor can be configured to generate subcategories for each type of answer a person gives. There can be different brand tags depending on which of the five (or any other predetermined number) answers a customer chooses. Subcategories 614 processor accesses a brand tags / categorizations 616 processor that stores a lookup table of subcategories of brands, such as but not limited to Frank and Eileen, Veronica Beard, and so on as shown. Brand tags / Categorizations 616 processor matches up each life category with brands (e.g., as shown). System 200 is further configured to receive data indicative of shopping ethics 618. Examples of shopping ethics 618 include but are not limited to data indicative of where fabrics or the person’s clothes are made, does the customer care about these things, would the customer be open to spending more on items that are ethically produced, as shown. System 200 is further configured to receive data indicative of an Add on section 620. Examples of section 620 include but are not limited to handbags, jewelry, undergarments, and a “just cuz you deserve it” piece.
[0131] It should be appreciated that in an embodiment, system 200 can receive the data described above sequentially, as indicated by the arrows. When sequentially, an embodiment can have Al engine 352 remove undesirable items as the process progresses, based on progressive user input. Further, in an embodiment, one or more of the steps (e.g., budget section 612 or section 620) can be omitted. Further, in an embodiment, such steps can be omitted by an implementer or by a user via a customized configuration or by the customer choosing to opt-out of providing user input to any of the categories.
[0132] In an embodiment, in addition to being business to business (B2B), the system (e.g., system 200) is configured to be a gateway to every clothing brand possible (e.g., in the public domain) so that the system filters and sees / presents everything (e.g., every item) available to the customer in the one stop shop. Thus, the system is configured so that it is possible for the customer to imagine, indeed view, everything they can view in any department store or solo business on one website. Such customer can shop for everything at the same time (e.g., during a single session or saved sessions).Style Mirror
[0133] An embodiment can be understood with reference to FIG. 8. FIG. 8 is a schematic diagram showing a style mirror 820 configured to engage in an interaction 840 with a user 810 and configured to display an avatar 830 of the user on which a garment is overlayed, according to an embodiment. The interaction 840 with the avatar stylist is by text, image, voice, and touchscreen option capabilities. For instance, style mirror 820 can show a streaming or video of a person asking the user 810 questions. Style mirror 820 can be configured to receive sound signals and interpret such signals as the language user 810 speaks. Further, style mirror 820 can receive user input via a virtual keyboard on the screen of style mirror 820 or via speaking words and phrases to style mirror 820. In an embodiment, the interaction 840 includes the questionnaire process. Style mirror 820 is configured to process one mode of how the shopper (e.g., user 810) interacts with the seven part questionnaire of FIG. 1A, for example. In addition, style mirror 820 is configured to show the shopper wearing an item of clothing in the shopper’s reflection. For instance, style mirror 820 can show the shopperreflection, that is showing the shopper wearing an item of clothing in the department they’re passing through or browsing.Clothing / lnventory Conveyor System
[0134] An embodiment can be understood with reference to FIG. 9. FIG. 9 is a schematic diagram showing an integrated clothing / inventory conveyor system, according to an embodiment. The innovative clothing / inventory conveyor system is configured to transport items between departments, backstock, and fitting rooms. Items are scanned so that a shopper can have a specific stock-keeping unit (SKU) delivered to a fitting room. Shoppers can interact with and request items through the style mirror 820.Clauses
[0135] In some embodiments, a programmable logic device (for example a field programmable gate array) may be used to perform some or all of the functionalities of the methods described herein. In some embodiments, a field programmable gate array may cooperate with a microprocessor in order to perform one of the methods described herein. Generally, the methods are performed by any hardware apparatus.
[0136] The system described herein may be implemented using a hardware apparatus, or using a computer, or using a combination of a hardware apparatus and a computer.
[0137] The system described herein, or any components of the system described herein, may be implemented at least partially in hardware and / or in software.
[0138] The methods described herein may be performed using a hardware apparatus, or using a computer, or using a combination of a hardware apparatus and a computer.
[0139] The methods described herein, or any components of the system described herein, may be performed at least partially by hardware and / or by software.
[0140] Some example solutions are listed below.
[0141] Clause 1. A method, comprising: receiving, at an artificial intelligence (Al) processor (352), a first user input from a client device, the first user input indicative of a user (350) shopping for an item; responsive to receiving the first user input indicative ofa user (350) shopping for an item, initiating a transmission of a questionnaire (353) to the client device; responsive to transmitting the questionnaire (353) to the client device, processing the following operations iteratively until a predetermined number of times, the predetermined number of times based on the questionnaire (353): receiving a subsequent user input from the client device, the subsequent user input comprising informational data about the questionnaire (353); retrieving, by the Al processor (352) and from a brand tagging processor (357), a plurality of tagged brand items based on the subsequent user input; retrieving, by the Al processor (352) and from a user profile processor (355), user profile data, the user profile data associated with the user (350), the retrieving based on the subsequent user input; retrieving, by the Al processor (352) and from a user purchase and return historical processor (356), user purchase and return historical data associated with the user (350), the retrieving based on the subsequent user input; and using, by the Al processor (352), the plurality of items, the user profile data, and the user purchase and return historical data to remove undesirable items from the brand tagging data; and at end of iterative processing, transmitting, by the Al processor (352), remaining items of the plurality of tagged brand items to the client device; wherein the Al processor (352) further comprises: generating a first training set, the first training set comprising parameters set by a first instance of the plurality of items, the user profile data, and the user purchase and return historical data; training a neural network (320, 322, 316) in a first stage using the first training set; retrieving updated return historical data (356); generating a second training set for a second stage of training comprising the first training set and the updated return historical data; and training the neural network (320, 322, 316) in a second stage using the second training set.
[0142] Clause 2. The method of clause 1 , wherein the iterative process uses if :then logic (354) so that each subsequent user input about the questionnaire (353) triggers a particular filter that narrows down the retrieved plurality of items.
[0143] Clause 3. The method of clause 1 , wherein the brand tagging processor (357) is linked to the questionnaire (353), each employing a same set of tags, the tags associated with the plurality of tagged brand items.
[0144] Clause 4. The method of clause 1 , wherein the questionnaire (353) is based on seven principles comprising parameters indicative of: determining a user’sbody image; determining parameters indicative of what makes a user (350) feel strong; determine fit guidance parameters; determine clothing processing parameters; determine parameters indicative of essentials of a successful wardrobe; determine budget parameters; and determine parameters indicative of the art of shopping; and wherein the questionnaire (353) parameters are mapped to particular tagged brand items based in part on the user profile data.
[0145] Clause 5. The method of clause 1 , the Al processor (352) further comprising: asking the user (350) via the client device what about returned items did not work; receiving and parsing an answer from the client device; transmitting an update to the user profile processor (355) or the user purchase and return historical processor (356) based on the answer; and using the answer, train an Al engine (320, 322, 316) associated with the Al processor (352).
[0146] Clause 6. The method of clause 1 , wherein specifics about each brand and style are input into an Al engine (320, 322, 316) of the Al processor (352) so that the Al engine (320, 322, 316) can start with that knowledge.
[0147] Clause 7. The method of clause 1 , wherein removing undesirable items further comprises matching (352) personality, shape, goals, and desires of the user (350) to clothing that fits those descriptors.
[0148] Clause 8. The method of clause 1 , further comprising an avatar (830) overlay process, the avatar (830) overlay process comprising: receiving first measurements from a body scanner (710); generating a proportionally-accurate avatar (830) based on the received measurements; receiving second scanner measurements of proportions of a garment (720); receiving a fabric content type of the garment; virtually displaying on a display (730) the proportionally-accurate avatar (830) wearing a visual representation of the garment, wherein the garment is visually rendered to appear made of the fabric content.
[0149] Clause 9. The method of clause 8, wherein the first measurements and the second scanner measurements comprise measurements (1 ) of: shoulder; bust; waist; hip; thigh; knee; calf; and feet.
[0150] Clause 10. The method of clause 8, wherein the display is a style mirror (820) and the style mirror (820) is configured to interact with the user (810) and configured to display the avatar (830) of the user (810) on which a garment is overlayed.
[0151] Clause 1 1 . The method of clause 1 , further comprising providing a clothing / inventory conveyor system that is integrated with Al processor (352, 300), wherein the clothing / inventory conveyor system is configured to deliver the remaining items of the plurality of tagged brand items to the user (810).
[0152] Clause 12. A system, comprising: an artificial intelligence (Al) engine (352) configured to: receive a first user input from a device of a user (350) (“client device”), the first user input indicative of a user (350) shopping for an item; and initiate, responsive to receiving the first user input indicative of a user (350) shopping (210) for an item, a transmission of a questionnaire (353) to the client device; process, responsive to transmitting the questionnaire (353) to the client device, subsequent user input iteratively until a predetermined number of times, the predetermined number of times based on the questionnaire (353); to transmit, at end of the iterative processing, remaining items of a plurality of tagged brand items to the client device; an if :then logic (354) processor configured to parse each subsequent user input about the questionnaire (353) and to trigger a particular filter that narrows the plurality of tagged brand items; a brand tagging processor (357) in communication with the Al engine (352) and configured to tag brand items and to transmit to the Al engine (352) a specific plurality of tagged brand items based on the subsequent user input; a user profile processor (355) in communication with the Al engine (352) and configured to transmit user profile data based on the subsequent user input, wherein the user profile data is associated with the user (350); a user purchase and return history processor (356) in communication with the Al engine (352) and configured to transmit to the Al engine (352) user purchase and return historical data, based on the subsequent user input, wherein the user purchase and return historical data is associated with the user (350); wherein to process iteratively the subsequent user input, the Al engine (352) is further configured to: receive the subsequent user input from the client device, the subsequent user input comprising informational data about the questionnaire (353); retrieve from the brand tagging processor (357), the plurality of tagged brand items based on the subsequent user input; retrieve from the user profile processor (355) the user profile data associated with the user (350) based on the subsequent user input; retrieve fromthe user purchase and return historical processor (356) the user purchase and return historical data associated with the user (350) based on the subsequent user input; and use the plurality of items, the user profile data, and the user purchase and return historical data to remove undesirable items from the brand tagging data; and wherein the Al engine (352) is further configured to: create a first training set, the first training set comprising parameters set by a first instance of the plurality of items, the user profile data, and the user purchase and return historical data; train a neural network (320, 322, 316) in a first stage using the first training set; create a second training set for a second stage of training comprising the first training set and the updated return historical data; and train the neural network (320, 322, 316) in a second stage using the second training set.
[0153] Clause 13. The system of clause 12, wherein the Al engine (352) is in communication with a community information (359) network and is further configured to receive community information from the community information (359) network and to use the community information to remove undesirable items from the brand tagging data.
[0154] Clause 14. The system of clause 12, wherein the Al engine (352) is in communication with a retailer (358) and is further configured to receive information depicting the interaction between the user (350) and the retailer (358) and to use the information depicting the interaction to remove undesirable items from the brand tagging data or to make recommendations to customers of the retailer (358).
[0155] Clause 15. The system of clause 12, wherein the questionnaire (353) is based on seven principles comprising parameters indicative of: determining a user’s body image; determining parameters indicative of what makes a user (810) feel strong; determine fit guidance parameters; determine clothing processing parameters; determine parameters indicative of essentials of a successful wardrobe; determine budget parameters; and determine parameters indicative of the art of shopping; and wherein the questionnaire (353) parameters are mapped to particular tagged brand items based in part on the user profile data.
[0156] Clause 16. A computer program product, the computer program product being embodied in a non-transitory computer readable storage medium and comprising computer instructions for: receiving, at an artificial intelligence (Al) processor (352), afirst user input from a client device, the first user input indicative of a user (350) shopping for an item; responsive to receiving the first user input indicative of a user (350) shopping for an item, initiating a transmission of a questionnaire (353) to the client device; responsive to transmitting the questionnaire (353) to the client device, processing the following operations iteratively until a predetermined number of times, the predetermined number of times based on the questionnaire (353): receiving a subsequent user input from the client device, the subsequent user input comprising informational data about the questionnaire (353); retrieving, by the Al processor (352) and from a brand tagging processor (357), a plurality of tagged brand items based on the subsequent user input; retrieving, by the Al processor (352) and from a user profile processor (355), user profile data, the user profile data associated with the user (350), the retrieving based on the subsequent user input; retrieving, by the Al processor (352) and from a user purchase and return historical processor (356), user purchase and return historical data associated with the user (350), the retrieving based on the subsequent user input; and using, by the Al processor (352), the plurality of items, the user profile data, and the user purchase and return historical data to remove undesirable items from the brand tagging data; and at end of iterative processing, transmitting, by the Al processor (352), remaining items of the plurality of tagged brand items to the client device; wherein the Al processor (352) further comprises: generating a first training set, the first training set comprising parameters set by a first instance of the plurality of items, the user profile data, and the user purchase and return historical data; training a neural network (320, 322, 316) in a first stage using the first training set; retrieving updated return historical data (356); generating a second training set for a second stage of training comprising the first training set and the updated return historical data; and training the neural network (320, 322, 316) in a second stage using the second training set.
[0157] Clause 17. The computer program product of clause 16, wherein the iterative process uses if :then logic (354) so that each subsequent user input about the questionnaire (353) triggers a particular filter that narrows down the retrieved plurality of items.
[0158] Clause 18. The computer program product of clause 16, wherein the questionnaire (353) is based on seven principles comprising parameters indicative of: determining a user’s body image; determining parameters indicative of what makes auser (350) feel strong; determine fit guidance parameters; determine clothing processing parameters; determine parameters indicative of essentials of a successful wardrobe; determine budget parameters; and determine parameters indicative of the art of shopping; and wherein the questionnaire (353) parameters are mapped to particular tagged brand items based in part on the user profile data.
[0159] Clause 19. The computer program product of clause 16, wherein the computer instructions further comprise instructions for the Al processor (352) further comprising: asking the user (350) via the client device what about returned items did not work; receiving and parsing an answer from the client device; transmitting an update to user profile processor (355) or the user purchase and return historical processor (356) based on the answer; and using the answer, train an Al engine (320, 322, 316) associated with the Al processor (352).
[0160] Clause 20. The computer program product of clause 16, wherein the brand tagging processor (357) is linked to the questionnaire (353), each employing a same set of tags, the tags associated with the plurality of tagged brand items.Remarks
[0161] The above description and drawings are illustrative and are not to be construed as limiting. Numerous specific details are described to provide a thorough understanding of the disclosure. However, in certain instances, well-known details are not described in order to avoid obscuring the description. Further, various modifications may be made without deviating from the scope of the embodiments.
[0162] Reference in this specification to “one embodiment” or “an embodiment” means that a particular feature, structure, or characteristic described in connection with the embodiment is included in at least one embodiment of the disclosure. The appearances of the phrase “in one embodiment” in various places in the specification are not necessarily all referring to the same embodiment, nor are separate or alternative embodiments mutually exclusive of other embodiments. Moreover, various features are described which may be exhibited by some embodiments and not by others. Similarly, various requirements are described which may be requirements for some embodiments but not for other embodiments.
[0163] The terms used in this specification generally have their ordinary meanings in the art, within the context of the disclosure, and in the specific context where each term is used. Certain terms that are used to describe the disclosure are discussed above, or elsewhere in the specification, to provide additional guidance to the practitioner regarding the description of the disclosure. For convenience, certain terms may be highlighted, for example using italics and / or quotation marks. The use of highlighting has no influence on the scope and meaning of a term; the scope and meaning of a term is the same, in the same context, whether or not it is highlighted. It will be appreciated that the same thing can be said in more than one way. One will recognize that “memory” is one form of a “storage” and that the terms may on occasion be used interchangeably.
[0164] Consequently, alternative language and synonyms may be used for any one or more of the terms discussed herein, nor is any special significance to be placed upon whether or not a term is elaborated or discussed herein. Synonyms for certain terms are provided. A recital of one or more synonyms does not exclude the use of other synonyms. The use of examples anywhere in this specification including examples of any term discussed herein is illustrative only and is not intended to further limit the scope and meaning of the disclosure or of any exemplified term. Likewise, the disclosure is not limited to various embodiments given in this specification.
[0165] Without intent to further limit the scope of the disclosure, examples of instruments, apparatus, methods and their related results according to the embodiments of the present disclosure are given above. Note that titles or subtitles may be used in the examples for convenience of a reader, which in no way should limit the scope of the disclosure. Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this disclosure pertains. In the case of conflict, the present document, including definitions will control.
Claims
CLAIMSI / We claim:1 . A method, comprising: receiving, at an artificial intelligence (Al) processor (352), a first user input from a client device, the first user input indicative of a user (350) shopping for an item; responsive to receiving the first user input indicative of a user (350) shopping for an item, initiating a transmission of a questionnaire (353) to the client device; responsive to transmitting the questionnaire (353) to the client device, processing the following operations iteratively until a predetermined number of times, the predetermined number of times based on the questionnaire (353): receiving a subsequent user input from the client device, the subsequent user input comprising informational data about the questionnaire (353); retrieving, by the Al processor (352) and from a brand tagging processor (357), a plurality of tagged brand items based on the subsequent user input; retrieving, by the Al processor (352) and from a user profile processor (355), user profile data, the user profile data associated with the user (350), the retrieving based on the subsequent user input; retrieving, by the Al processor (352) and from a user purchase and return historical processor (356), user purchase and return historical data associated with the user (350), the retrieving based on the subsequent user input; and using, by the Al processor (352), the plurality of items, the user profile data, and the user purchase and return historical data to remove undesirable items from the brand tagging data; and at end of iterative processing, transmitting, by the Al processor (352), remaining items of the plurality of tagged brand items to the client device;wherein the Al processor (352) further comprises: generating a first training set, the first training set comprising parameters set by a first instance of the plurality of items, the user profile data, and the user purchase and return historical data; training a neural network (320, 322, 316) in a first stage using the first training set; retrieving updated return historical data (356); generating a second training set for a second stage of training comprising the first training set and the updated return historical data; and training the neural network (320, 322, 316) in a second stage using the second training set.
2. The method of claim 1 , wherein the iterative process uses if :then logic (354) so that each subsequent user input about the questionnaire (353) triggers a particular filter that narrows down the retrieved plurality of items.
3. The method of claim 1 , wherein the brand tagging processor (357) is linked to the questionnaire (353), each employing a same set of tags, the tags associated with the plurality of tagged brand items.
4. The method of claim 1 , wherein the questionnaire (353) is based on seven principles comprising parameters indicative of: determining a user’s body image; determining parameters indicative of what makes a user (350) feel strong; determine fit guidance parameters; determine clothing processing parameters; determine parameters indicative of essentials of a successful wardrobe; determine budget parameters; and determine parameters indicative of the art of shopping; and wherein the questionnaire (353) parameters are mapped to particular tagged brand items based in part on the user profile data.
5. The method of claim 1 , the Al processor (352) further comprising: asking the user (350) via the client device what about returned items did not work; receiving and parsing an answer from the client device;transmitting an update to the user profile processor (355) or the user purchase and return historical processor (356) based on the answer; and using the answer, train an Al engine (320, 322, 316) associated with the Al processor (352).
6. The method of claim 1 , wherein specifics about each brand and style are input into an Al engine (320, 322, 316) of the Al processor (352) so that the Al engine (320, 322, 316) can start with that knowledge.
7. The method of claim 1 , wherein removing undesirable items further comprises matching (352) personality, shape, goals, and desires of the user (350) to clothing that fits those descriptors.
8. The method of claim 1 , further comprising an avatar (830) overlay process, the avatar (830) overlay process comprising: receiving first measurements from a body scanner (710); generating a proportionally-accurate avatar (830) based on the received measurements; receiving second scanner measurements of proportions of a garment (720); receiving a fabric content type of the garment; virtually displaying on a display (730) the proportionally-accurate avatar (830) wearing a visual representation of the garment, wherein the garment is visually rendered to appear made of the fabric content.
9. The method of claim 8, wherein the first measurements and the second scanner measurements comprise measurements (1 ) of: shoulder; bust; waist; hip; thigh; knee; calf; and feet.
10. The method of claim 8, wherein the display is a style mirror (820) and the style mirror (820) is configured to interact with the user (810) and configured to display the avatar (830) of the user (810) on which a garment is overlayed. re1 1 . The method of claim 1 , further comprising providing a clothing / inventory conveyor system that is integrated with Al processor (352, 300), wherein the clothing / inventory conveyor system is configured to deliver the remaining items of the plurality of tagged brand items to the user (810).
12. A system, comprising: an artificial intelligence (Al) engine (352) configured to: receive a first user input from a device of a user (350) (“client device”), the first user input indicative of a user (350) shopping for an item; and initiate, responsive to receiving the first user input indicative of a user (350) shopping (210) for an item, a transmission of a questionnaire (353) to the client device; process, responsive to transmitting the questionnaire (353) to the client device, subsequent user input iteratively until a predetermined number of times, the predetermined number of times based on the questionnaire (353); to transmit, at end of the iterative processing, remaining items of a plurality of tagged brand items to the client device; an if :then logic (354) processor configured to parse each subsequent user input about the questionnaire (353) and to trigger a particular filter that narrows the plurality of tagged brand items; a brand tagging processor (357) in communication with the Al engine (352) and configured to tag brand items and to transmit to the Al engine (352) a specific plurality of tagged brand items based on the subsequent user input; a user profile processor (355) in communication with the Al engine (352) and configured to transmit user profile data based on the subsequent user input, wherein the user profile data is associated with the user (350); a user purchase and return history processor (356) in communication with the Al engine (352) and configured to transmit to the Al engine (352) user purchase and return historical data, based on the subsequent user input, wherein the user purchase and return historical data is associated with the user (350);wherein to process iteratively the subsequent user input, the Al engine (352) is further configured to: receive the subsequent user input from the client device, the subsequent user input comprising informational data about the questionnaire (353); retrieve from the brand tagging processor (357), the plurality of tagged brand items based on the subsequent user input; retrieve from the user profile processor (355) the user profile data associated with the user (350) based on the subsequent user input; retrieve from the user purchase and return historical processor (356) the user purchase and return historical data associated with the user (350) based on the subsequent user input; and use the plurality of items, the user profile data, and the user purchase and return historical data to remove undesirable items from the brand tagging data; and wherein the Al engine (352) is further configured to: create a first training set, the first training set comprising parameters set by a first instance of the plurality of items, the user profile data, and the user purchase and return historical data; train a neural network (320, 322, 316) in a first stage using the first training set; create a second training set for a second stage of training comprising the first training set and the updated return historical data; and train the neural network (320, 322, 316) in a second stage using the second training set.
13. The system of claim 12, wherein the Al engine (352) is in communication with a community information (359) network and is further configured to receive community information from the community information (359) network and to use the community information to remove undesirable items from the brand tagging data.
14. The system of claim 12, wherein the Al engine (352) is in communication with a retailer (358) and is further configured to receive information depicting theinteraction between the user (350) and the retailer (358) and to use the information depicting the interaction to remove undesirable items from the brand tagging data or to make recommendations to customers of the retailer (358).
15. The system of claim 12, wherein the questionnaire (353) is based on seven principles comprising parameters indicative of: determining a user’s body image; determining parameters indicative of what makes a user (810) feel strong; determine fit guidance parameters; determine clothing processing parameters; determine parameters indicative of essentials of a successful wardrobe; determine budget parameters; and determine parameters indicative of the art of shopping; and wherein the questionnaire (353) parameters are mapped to particular tagged brand items based in part on the user profile data.
16. A computer program product, the computer program product being embodied in a non-transitory computer readable storage medium and comprising computer instructions for: receiving, at an artificial intelligence (Al) processor (352), a first user input from a client device, the first user input indicative of a user (350) shopping for an item; responsive to receiving the first user input indicative of a user (350) shopping for an item, initiating a transmission of a questionnaire (353) to the client device; responsive to transmitting the questionnaire (353) to the client device, processing the following operations iteratively until a predetermined number of times, the predetermined number of times based on the questionnaire (353): receiving a subsequent user input from the client device, the subsequent user input comprising informational data about the questionnaire (353); retrieving, by the Al processor (352) and from a brand tagging processor (357), a plurality of tagged brand items based on the subsequent user input;retrieving, by the Al processor (352) and from a user profile processor (355), user profile data, the user profile data associated with the user (350), the retrieving based on the subsequent user input; retrieving, by the Al processor (352) and from a user purchase and return historical processor (356), user purchase and return historical data associated with the user (350), the retrieving based on the subsequent user input; and using, by the Al processor (352), the plurality of items, the user profile data, and the user purchase and return historical data to remove undesirable items from the brand tagging data; and at end of iterative processing, transmitting, by the Al processor (352), remaining items of the plurality of tagged brand items to the client device; wherein the Al processor (352) further comprises: generating a first training set, the first training set comprising parameters set by a first instance of the plurality of items, the user profile data, and the user purchase and return historical data; training a neural network (320, 322, 316) in a first stage using the first training set; retrieving updated return historical data (356); generating a second training set for a second stage of training comprising the first training set and the updated return historical data; and training the neural network (320, 322, 316) in a second stage using the second training set.
17. The computer program product of claim 16, wherein the iterative process uses if :then logic (354) so that each subsequent user input about the questionnaire (353) triggers a particular filter that narrows down the retrieved plurality of items.
18. The computer program product of claim 16, wherein the questionnaire (353) is based on seven principles comprising parameters indicative of: determining a user’s body image; determining parameters indicative of what makes a user (350) feel strong; determine fit guidance parameters; determine clothing processing parameters;determine parameters indicative of essentials of a successful wardrobe; determine budget parameters; and determine parameters indicative of the art of shopping; and wherein the questionnaire (353) parameters are mapped to particular tagged brand items based in part on the user profile data.
19. The computer program product of claim 16, wherein the computer instructions further comprise instructions for the Al processor (352) further comprising: asking the user (350) via the client device what about returned items did not work; receiving and parsing an answer from the client device; transmitting an update to user profile processor (355) or the user purchase and return historical processor (356) based on the answer; and using the answer, train an Al engine (320, 322, 316) associated with the Al processor (352).
20. The computer program product of claim 16, wherein the brand tagging processor (357) is linked to the questionnaire (353), each employing a same set of tags, the tags associated with the plurality of tagged brand items.
Citation Information
Patent Citations
Virtual Personal Shopping System
US20180218436A1
Virtual representation creation of user for fit and style of apparel and accessories
US20190266806A1
System and method for real-time calibration of virtual apparel using stateful neural network inferences and interactive body measurements
US20220258049A1
Matching Cosmetics and Skin Care Products Based on Skin Tone and Skin Condition Scanning
US20230074782A1
Systems and methods for intelligent product classification using product titles
WO2022172067A1
