33results about How to "Few features" patented technology

Composite cushioning structures with spatially variable cushioning properties and related materials and methods for producing same are disclosed. In one embodiment, gradient property foam (GPF) is provided and comprised of a two or more phase composite cushioning structure comprised of a spatially variable disposition of one or more second non-solid phase cushioning component into a first solid phase cellular cushioning component to control the local cushioning properties of the final composite cushioning structure. The second cushioning component(s) can be selected to exhibit other cushioning properties when in solid form from the cushioning properties of the first cellular cushioning component in this embodiment. In one embodiment, the second cushioning component(s) can undergo a transition from the non-solid phase into a solid phase within the first cellular cushioning component to form a bond with the first cellular cushioning component to form the composite cushioning structure.

A lamp can include an LED light source and a lens body having a first lens portion and a second lens portion arranged outside the first lens portion, the first and second lens portions being integrally formed with each other. The first lens portion can include a first light-incident surface and a refractive surface to form a main light distribution pattern condensation and refraction. The second lens portion can include a second light-incident surface, a first total-reflecting surface, a ring-shaped light projecting surface including an individual light projecting surface and a second total-reflecting surface, and a third total-reflecting surface. The second light-incident surface can be disposed beside the LED light source and can refract the light reaching the second light-incident surface to let the light enter the inside of the second lens portion. The first total-reflecting surface can totally reflect light entering through the second light-incident surface to condense the light in the front direction. The ring-shaped light projecting surface is disposed to cover an optical path range of light reflected from the first total-reflecting surface and is divided into a plurality of areas. The individual light projecting surface is provided in at least one of the plurality of divided areas and can transmit the light totally reflected from the first total-reflecting surface. The second total-reflecting surface can totally reflect the light from the first total-reflecting surface in the sideward and outward direction. The third total-reflecting surface can reflect light from the second total-reflecting surface to direct the light in the front direction.

A novel statistical learning procedure that can be applied to many machine-learning applications is presented. Although this boosting learning procedure is described with respect to its applicability to face detection, it can be applied to speech recognition, text classification, image retrieval, document routing, online learning and medical diagnosis classification problems.

Individuals want to look their best because when they look good, they feel good. It affects not only how they perceive themselves, but also how others perceive them. Unfortunately, many individuals don't look as good as they would like. And even more frustrating, they don't know how to produce their optimal facial look. Imagine a plain or average looking woman being able to transform herself into a very pretty woman. Imagine an already pretty woman being able to transform herself into a gorgeous woman. The Perfect Shape™ Makeover Process empowers individuals to achieve this magical transformation. Confidence, self-esteem, perception, and respect are enhanced in very positive ways. This is something individuals want on a universal, world-wide level. The Perfect Shape™ Makeover Process instructs the client how to use makeup to reshape each of these facial features to the ultimate of what they have to work with.

The invention discloses an event detection method based on Twitter, used to solve the real-time problem of news clue acquisition in the prior art. The event detection method based on the Twitter comprises the following steps: pre-treating tweets by using the natural language processing technology; performing online clustering on the tweets based on short text similarity; performing preliminary judgment on the event attributes of a tweet cluster by means of the characteristics of the tweets in each tweet class; and judging whether the detected event is a first event or not by means of the existing tweet event library, so as to prevent the event from repeating output. The event detection method based on the Twitter has low computational complexity and can be used for detection of sudden hotevents in large-scale flow data; and the event detection method based on the Twitter helps to obtain valuable information from a social media faster and more accurately, and can produce economic benefits directly or indirectly.

The invention relates to a real-time low-bit rate video transcoding method based on H.264, belonging to the field of multi-media signal processing and mainly solving the problem that the transcoding speed is demanded to be high at low bit rate. Video transcoding in the invention is divided into an off-line modeling stage and an online transcoding stage; a model identification technology is utilized to select a macrolbock prediction model; the macrolbock prediction model selection process is a category decision-making process in model identification; and extracted characteristics are input into a classifier to obtain a classification result, i.e. a macroblock prediction model. The classifier is obtained through offline training. The method not only can remarkably accelerate the transcoding speed but also ensures the quality of a recoded image to meet the demand of real-time performance. The invention is suitable for the fields of wireless video monitoring, internet video monitoring and wireless video on demand.

The invention discloses a systematic clustering thought-based driving behavior identification method. The driving behavior identification method comprises the following steps of building a statisticalfeature model; performing driving behavior data collection by utilizing an inertial sensor; filtering data by utilizing a Kalman filter; extracting effective driving behavior data from the filtered data by adopting an adaptive window function method, and further extracting statistical features of the effective data; performing screening to obtain optimal statistical features; and classifying thestatistical features of the driving behavior data, and identifying corresponding driving behaviors. The invention provides the systematic driving behavior identification method with extremely few features, extremely low calculation amount and extremely high precision based on a clustering thought; and the method has wide application prospects in the fields of social safety, car insurances, fleet management and the like.

The invention discloses an image segmentation level selection method based on scale perception, and aims to improve the segmentation quality of an overall image. The method comprises the following main steps: firstly, obtaining a segmentation result represented by a tree form by using a multi-level image segmentation algorithm; secondly, calculating feature vectors of all segmented regions in eachsegmentation layer, and performing quantitative description on region segmentation quality according to a visual format tower principle; thirdly, constructing a graph model of the multi-level segmented image by taking the hierarchical region with the finest segmentation granularity as a node, and finally, mapping a label to a region corresponding to the original hierarchy, and combining to obtaina final image segmentation result. According to the method, based on the scale selection principle of segmentation area quality, the problem of hierarchical selection of multi-level image segmentation is solved, and the limitation of traditional single hierarchical selection and the uncertainty of threshold parameters on segmentation hierarchical selection are overcome. As a post-processing means, the output quality of the multi-level image segmentation algorithm in a visual processing task can be improved.

The invention is applicable to the technical field of machine learning, and provides a multivariate discrete feature selection method, device, apparatus, and storage medium. The method comprises: initializing a particle swarm according to a tangent point corresponding to each feature in a target data set, obtaining the particle position of each particle, the target data set is discretized according to the particle position, the corresponding discrete data set is obtained, in accordance with discrete data set, the fitness of each particle is calculated by the fitness formula, so as to find outthe optimal position of the population of the particle swarm and the optimal position of the individual through which each particle passes, when the stop condition is met, optimal position of output population, otherwise, the particle position of each particle is updated according to the optimal position of the population and the optimal position of the individual, and the operation of data discretization and optimization is continued, so that fewer features are selected, the effect of eliminating redundant features and irrelevant features is improved, and the accuracy of classification learning algorithm is improved.

This invention provides a method for conducting a non-real time group interaction, employing a programmable voice messaging system, such as a voice message board, for storing and replaying messages from facilitators and group members. The invention provides a method for communication in rounds of discussion, whereby facilitators leave a first message and group members listen to and optionally respond to that message. The facilitators then listen to the members' responses and optionally create additional messages or optionally edit the members' messages, to guide and control the discussion. The cycle of facilitators listening, optionally responding, optionally editing followed by group members listening and optionally responding can be repeated multiple times. Also provided is a voice note pad means for users of the instant invention, whereby users can interrupt the stream of voice messages and create a new, private voice note pad message, and optionally later listen to or edit their voice note pad messages.

A radiation device (30) for providing radiation to the skin is described. The radiation device comprises a radiation guide (31) for directing radiation whereby the radiation guide (31) is configured for receiving radiation from at least one radiation source (20) in a side-lit configuration. The radiation device (30) comprises moisture transfer channels (32) inside the radiation guide (31) for enabling moisture transfer from the skin through the radiation guide (31) to the environment. The shape of the walls of the moisture transfer channels (32) is adapted for redirecting radiation in the radiation guide (31) towards the skin.

A method, system, and / or apparatus for quickly detecting blocks or regions in an image whose pixels that can be coded efficiently by a pointer to the same block occurring in a reference image. The reference image is shared between an encoder and a corresponding decoder. One embodiment of the encoder quickly detects such blocks in an image by: (a) defining features such that substantially fewer features occur in an image than pixels, (b) correlating features in an image to a corresponding feature in the reference image, (c) estimating a displacement vector representative of the displacement of the correlated features, (d) estimating a seed point within the displaced block from which to grow the block, and (e) growing the displaced block to determine its size, shape, and / or dimensions.

An embodiment of the invention discloses a license plate positioning method and positioning device, and relates to the field of image processing, and aims to solve the problem that a license plate positioning method is complex in calculation, long in consumed time and the like in the prior art. The positioning method includes the steps: acquiring sample images with known categories; calculating pixel difference characteristics of the sample images; identifying categories of the pixel difference characteristics according to the categories of the sample images; leaning the pixel difference characteristics identified the categories by the aid of a machine learning algorithm to obtain a license plate classifier; performing classification sub-images in a first sub-image set to be detected by the aid of the license plate classifier to obtain first classification results; outputting license plate positioning results according to the first classification results. The sample images comprise license plate images and non-license-plate images, and images to be positioned are traversed by the aid of a first preset detection window to obtain the first sub-image set to be detected.

Disclosed is an ultraviolet-curable composition containing an epoxy (meth)acrylate, a urethane (meth)acrylate and a silicone compound represented by the following formula (1). (CH3)3Si-O-X-O-Si(CH3)3 (1) (In the formula, X represents a group having a structure wherein n-number of structural units represented by the following formula (2): (wherein R1 represents a divalent group obtained by linking one or more groups selected from alkylene groups having 1-18 carbon atoms and divalent groups having a structure wherein alkylene groups having 2-18 carbon atoms are linked by an ether bond, and R1 may be bonded to Si directly or through an oxygen atom; and R2 represents a hydrogen atom or a methyl group) and m-number of structural units represented by the following formula (3): are linked at random or in blocks. In this connection, n is an integer of 1-15, and n / m is within the range of 1 / 5-1 / 20.) This ultraviolet-curable composition enables to obtain a light-transmitting layer which is suppressed in warping, while having excellent wear resistance.

The invention relates to a coal gangue recognition and separation method based on machine vision and capacitance fusion and a device thereof. The device comprises a retaining rack arranged on a belt conveyor rack for mounting an accessory, wherein a detection unit, a conveying unit, an electric unit and a pneumatic unit are connected to the rack through the retaining rack and are arranged around the belt conveyor reasonably. The method comprises the steps of acquiring an effective thickness of a medium and center position information of a contour thereof through a CCD camera, and combining partial voltage numerical values at two ends of a resistor to recognize coal gangue; and operating a nozzle accurately to separate the coal gangue according to the position information of the center of the contour in combination with the operating speed of a conveyor belt. The method solves the problems that it is hard to extract features of the coal gangue, the recognition precision is low, the mistake separating probability is large and the like. The method is high in processing speed and high in timeliness, and can achieve very high recognition precision with few recognition features.

A method of playing a five card draw video poker wagering game on a gaming device comprising at least a processer, input device and video monitor screen is disclosed. The method comprises providing bonus features in video poker games without requiring an extra wager in addition to a base wager, without lowering the paytable, and basing the paytable on the total wager. In lieu of an extra wager, a predetermined set of dealt hands is automatically disqualified, effectively ending the game and forfeiting the wager. All dealt hands that are not disqualified are qualified hands. The player is eligible to receive at least one randomly-provided bonus feature or guaranteed bonus feature on every qualified hand.

The invention relates to a real-time low-bit rate video transcoding method based on H.264, belonging to the field of multi-media signal processing and mainly solving the problem that the transcoding speed is demanded to be high at low bit rate. Video transcoding in the invention is divided into an off-line modeling stage and an online transcoding stage; a model identification technology is utilized to select a macrolbock prediction model; the macrolbock prediction model selection process is a category decision-making process in model identification; and extracted characteristics are input into a classifier to obtain a classification result, i.e. a macroblock prediction model. The classifier is obtained through offline training. The method not only can remarkably accelerate the transcodingspeed but also ensures the quality of a recoded image to meet the demand of real-time performance. The invention is suitable for the fields of wireless video monitoring, internet video monitoring andwireless video on demand.

The invention discloses an electrocardio recognition method with the ultralow data volume. The method is used for electrocardio signal feature extraction and belongs to the field of signal recognition. The method is characterized by comprising the following steps of (1) performing feature extraction on electrocardio signals by using a compressed sensing Chebyshev Chaos projection matrix; (2) extracting the morphological features-front and back RR intervals of the electrocardio signals; and (3) performing standardized processing on the electrocardio projection features and morphologic featuresand then feeding the processed features into a classifier to complete the electrocardio signal recognition process. Compared with the existing electrocardio signal recognition method, the electrocardio recognition method has the advantages that the 11-hour electrocardio signal data (108M) can be compressed to be 1M; the data storage space is greatly reduced; the calculation complexity degree is reduced; the electrocardio classified recognition can be performed at a higher speed; meanwhile, the recognition accuracy reaches 93.36 percent; and important significance is realized on real-time analysis and handling of heart diseases.

The embodiment of the invention discloses a Feature selection method and device for high-dimensional data and computer storage medium for high-dimensional data. The method may include initializing particles in a population according to a tangent point of a data set to be processed; According to the preset stopping criterion and the updating strategy, iteratively updating each initialized particleto obtain the updated output particle; Testing the test set by using a non-dominant solution in the output particle to determine an optimal solution particle of the test; Wherein the test set is a part of the data set to be processed, the test optimal solution is used for feature selection of the data set to be processed, and the test optimal solution is particles with the highest classification accuracy rate and the least number of features of the test set.

A method of playing a five card draw video poker wagering game on a gaming device comprising at least a processer, input device and video monitor screen is disclosed. The method comprises providing bonus features in video poker games without requiring an extra wager in addition to a base wager, without lowering the paytable, and basing the paytable on the total wager. In lieu of an extra wager, a predetermined set of dealt hands is automatically disqualified, effectively ending the game and forfeiting the wager. All dealt hands that are not disqualified are qualified hands. The player is eligible to receive at least one randomly-provided bonus feature or guaranteed bonus feature on every qualified hand.

Exemplary methods, apparatuses, and systems include a communication system accessing a request received from an application on a user device to log into a primary platform of a communication system using a user account. In response to detecting the request as an unrecognized login attempt, that the user account also grants access to a secondary platform of the communication system, and that settings of the user account include enabled push notifications via the secondary platform, the communication system provides a notification for display to a user of the user account via the secondary platform.

The invention discloses an Internet of Things equipment identification method and device based on machine learning, and the method comprises the steps: obtaining a data set containing network flows of Internet of Things equipment, carrying out the data preprocessing, obtaining a plurality of network flows, extracting a plurality of flow features from the network flows through a plurality of feature selection algorithms, so as to determine an optimal feature subset, and carrying out the recognition of the Internet of Things equipment. The method comprises the following steps: inputting a trained machine learning model to carry out Internet of Things equipment identification to obtain an equipment identification result, respectively carrying out comparative analysis based on a feature selection algorithm corresponding to the equipment identification result to complete Internet of Things equipment identification, and evaluating the effectiveness of network traffic features by adopting various feature selection algorithms. According to the method, the BGWO is introduced to determine the optimal feature subset of the identification of the Internet of Things equipment to fit the optimal model and realize the optimal accuracy among different classifiers, the reliability of the identification result is ensured by using two public Internet of Things network flow data sets, the performance of the classifiers is improved, and the BGWO realizes higher accuracy in all the classifiers.

The invention discloses a systematic driving behavior recognition method based on the idea of ​​clustering. The driving behavior recognition method includes the following steps: establishing a statistical feature model; using an inertial sensor to collect driving behavior data; using a Kalman filter to Filtering the data; using an adaptive window function method to extract effective driving behavior data on the filtered data, and further extracting the statistical features of the effective data; screening to obtain the optimal statistical features; classifying the statistical features of the driving behavior data , to identify the corresponding driving behavior. Based on the idea of ​​clustering, the present invention provides a systematic driving behavior recognition method with very few features, very low calculation amount and high precision, which has broad application prospects in the fields of social security, auto insurance, fleet management and the like.

The invention relates to a method and device for identifying and separating coal gangue based on machine vision and capacitive fusion, including a holder arranged on a belt conveyor frame for installing accessories, a detection unit, a conveying unit, an electric unit and a pneumatic The unit is connected to the frame through the cage, and is reasonably arranged around the belt conveyor; the effective thickness of the medium and the position information of the contour center are obtained through the CCD camera, and the coal gangue is identified based on the partial pressure values ​​at both ends of the resistor; according to the contour center Combined with the position information of the conveyor belt and the running speed of the conveyor belt, the nozzles can be accurately controlled to separate the gangue; it effectively solves the problems of difficulty in extracting coal gangue features, low recognition accuracy and high probability of wrong separation; the processing speed is fast, the timeliness is strong, and it can be used very quickly. Fewer recognition features to achieve high recognition accuracy.

The invention discloses a knowledge base question adding method based on a vector search engine, and the method comprises the steps: extracting scene questions in each scene of a knowledge base, converting at least one similar question of each scene question into a feature vector, and storing the feature vector into a Milvus library; storing the feature vector of each similar question in the Milvus library, and establishing an index; obtaining a newly-added question, converting the newly-added question into a feature vector, conducting similarity retrieval on the feature vector of the newly-added question in the Milvus library to obtain a first similar question and the similarity probability, wherein the similarity probability of the feature vector of the first similar question and the feature vector of the newly-added question is the highest; and determining whether to add the newly added question to the Milvus library or not based on the similarity probability.

The invention provides a speech depression state recognition method based on feature selection and transfer learning, and provides the speech depression state recognition method fusing Lasso and a transfer learning method CORAL in order to solve the two problems that the feature dimension is higher, and the feature distribution is affected by the tested individual differences except for the depression level during modeling based on speech. The method has the advantages that the redundant information in Lasso filtering features is reserved, the effective features are reserved, and the recognition precision is further improved on the basis of improving the model efficiency; 2, on the premise that the depression label information is not leaked by the transfer learning method CORAL, the feature distribution of a training set and a test set is drawn close, and the influence of other factors except the depression level on the feature distribution is reduced; the combination of the two methods can further improve the accuracy and stability of depression screening.

The invention provides a data information transmission method, apparatus, system and device. The method comprises the steps: constructing a feature value message according to a feature value corresponding to at least one feature attribute in to-be-transmitted target data, wherein the target data comprises at least one feature attribute and a feature value corresponding to the at least one featureattribute; when that a feature attribute in the target data meets a feature attribute sending condition is detected, constructing a feature attribute message associated with the feature numerical value message according to the at least one feature attribute; sending the feature attribute message and the feature numerical value message to opposite-end equipment, and therefore the opposite-end equipment recovers the target data according to each feature numerical value carried by the feature numerical value message and the feature attribute which is carried by the feature attribute message and corresponds to each feature numerical value. By using the method provided by the invention, the redundant transmission of the characteristic attributes can be reduced while the characteristic number transmission efficiency is improved.