Device and method for measuring reliability of molecular structure prediction model

Abstract

One embodiment of the present disclosure is intended to provide a system, a computer program, a device, and a method for measuring the reliability of a molecular structure prediction model. The method may comprise the steps of: acquiring an image of a first molecular structure; acquiring a graph of the first molecular structure by using the molecular structure prediction model; performing, on the basis of the graph of the first molecular structure, image rendering on the image of the first molecular structure; and determining the reliability of the graph of the first molecular structure on the basis of the image rendering result and the graph of the first molecular structure.

Description

Device and method for measuring the reliability of a molecular structure prediction model

[0001] The present disclosure relates to a device and method for measuring the reliability of a molecular structure prediction model. More specifically, the present disclosure provides convenience by providing reliability information for the results of a molecular structure prediction model when the model predicts a predicted molecular structure.

[0002] A structural formula is a graphical representation of a chemical or molecular structure, showing how atoms are arranged in three-dimensional space. A structural formula can also explicitly or implicitly represent a molecule's chemical bonds. Unlike molecular formulas, which have a limited number of symbols and can only provide a limited description, structural formulas can provide geometric information about a molecule's structure. For example, isomers, which share the same molecular formula but differ in atomic structure or arrangement, can be expressed.

[0003] Structural formulas are often provided in image format in various documents, papers, patents, and other publications. However, unlike text, images are difficult to search, making it difficult to find documents containing the corresponding structural formulas. Consequently, various methods for retrieving images, such as structural formulas, are being developed. Models that analyze images to extract structural formulas are primarily used to create academic databases. However, if incorrect predictions lead to incorrect data being included in these databases, this can be a critical drawback for research. Therefore, there is a need for a method that provides reliability information for predicted structural formulas, allowing researchers to determine which of the predicted structures are reliable and which should be stored in the database.

[0004] The present disclosure provides a method and device for predicting molecular structures using images, in which a model predicting molecular structures provides a confidence score when predicting molecular structures.

[0005] One embodiment of the present disclosure can provide a device and method for measuring the reliability of a molecular structure prediction model.

[0006] One embodiment of the present disclosure provides a system for measuring the reliability of a molecular structure prediction model. The system includes a memory storing one or more commands; and at least one processor executing the one or more commands stored in the memory, wherein the at least one processor, by executing the one or more commands, obtains an image of a first molecular structure, obtains a graph of the first molecular structure determined using the molecular structure prediction model, performs image rendering on the image of the first molecular structure based on the graph of the first molecular structure, and determines the reliability of the graph of the first molecular structure based on the image rendering result and the graph of the first molecular structure.

[0007] In one embodiment, the at least one processor performs the image rendering by identifying at least one of a first component and a second component based on a graph of the first molecular structure, identifying a first portion corresponding to the first component in an image of the first molecular structure, identifying a second portion corresponding to the second component in the image of the first molecular structure, and distinguishing the first portion and the second portion with different marks, wherein each of the first component and the second component may include any one of a first atom, a second atom, a first bond, and a second bond.

[0008] In one embodiment, the molecular structure prediction model includes a first learning model, and the first learning model may be trained to extract a chemical table file graph using a molecular structure formula image as input.

[0009] In one embodiment, the at least one processor can output a reliability using a second learning model, taking as input the image rendering result and the graph of the first molecular structure.

[0010] In one embodiment, the second learning model may include an image backbone model that extracts features of the image rendering result; a graph backbone model that extracts features of the graph of the first molecular structure; a feature connector that connects features of the image rendering result and features of the graph of the first molecular structure; and a linear layer model that determines reliability by taking as input the output of the feature connector.

[0011] In one embodiment, the second learning model may be trained to output a first value when the image rendering result and the graph of the first molecular structure match, and to output a second value when the image rendering result and the graph of the first molecular structure do not match.

[0012] In one embodiment, a graph of a molecular structure having a reliability level greater than or equal to a predetermined level may be stored in a database.

[0013] One embodiment of the present disclosure provides a method for measuring the reliability of a molecular structure prediction model, which is performed by at least one processor. The method may include the steps of: obtaining an image of a first molecular structure; obtaining a graph of the first molecular structure using the molecular structure prediction model; performing image rendering on the image of the first molecular structure based on the graph of the first molecular structure; and determining the reliability of the graph of the first molecular structure based on the image rendering result and the graph of the first molecular structure.

[0014] In one embodiment, the step of performing image rendering on the image of the first molecular structure includes the steps of: identifying at least one of a first component and a second component based on a graph of the first molecular structure; identifying a first portion corresponding to the first component in the image of the first molecular structure; identifying a second portion corresponding to the second component in the image of the first molecular structure; and performing the image rendering by distinguishing the first portion and the second portion with different marks, wherein each of the first component and the second component may include any one of a first atom, a second atom, a first bond, and a second bond.

[0015] In one embodiment, the molecular structure prediction model includes a first learning model, and the first learning model may be trained to extract a chemical table file graph using a molecular structure formula image as input.

[0016] In one embodiment, the step of outputting the reliability of the graph of the first molecular structure may include the step of outputting the reliability of the graph of the first molecular structure using a second learning model, using the image rendering result and the graph of the first molecular structure as inputs.

[0017]

[0018] * In one embodiment, the second learning model may include an image backbone model that extracts features of the image rendering result; a graph backbone model that extracts features of the graph of the first molecular structure; a feature connector that connects features of the image rendering result and features of the graph of the first molecular structure; and a linear layer model that determines reliability by taking as input the output of the feature connector.

[0019] In one embodiment, the second learning model may be trained to output a first value when the image rendering result and the graph of the first molecular structure match, and to output a second value when the image rendering result and the graph of the first molecular structure do not match.

[0020] In one embodiment, a graph of molecular structures having a reliability greater than a certain value may be stored in a database.

[0021] One embodiment of the present disclosure includes a program stored on a recording medium to cause a computer to execute a method according to one embodiment of the present disclosure.

[0022] One embodiment of the present disclosure includes a computer-readable recording medium having recorded thereon a program for executing a method according to one embodiment of the present disclosure on a computer.

[0023] One embodiment of the present disclosure includes a computer-readable recording medium having recorded thereon a database used in one embodiment of the present disclosure.

[0024] According to one embodiment of the present disclosure, the reliability of the results of the prediction model may be provided together to provide accuracy information about the results of the prediction model, and further, may help a user decide which information to store in a database.

[0025] FIG. 1 is a diagram illustrating a method for extracting a molecular structure from an image using a first learning model according to one embodiment of the present disclosure.

[0026] FIG. 2 is a flowchart illustrating a method for providing reliability of a result value of a first learning model using a second learning model according to an embodiment of the present disclosure.

[0027] FIG. 3 is a flowchart illustrating a method for measuring the reliability of a molecular structure prediction model according to one embodiment of the present disclosure.

[0028] FIG. 4A is a diagram illustrating a method for measuring reliability when a molecular structure prediction model according to one embodiment of the present disclosure makes an incorrect prediction.

[0029] FIG. 4b is a diagram illustrating a method for measuring reliability when a molecular structure prediction model according to one embodiment of the present disclosure correctly predicts.

[0030] FIG. 5 is an example diagram of a second learning model according to one embodiment of the present disclosure.

[0031] FIG. 6 is a block diagram of a reliability measurement device for a molecular structure prediction model according to one embodiment of the present disclosure.

[0032] To clarify the technical idea of ​​the present disclosure, embodiments of the present disclosure will be described in detail with reference to the attached drawings. In describing the present disclosure, if a detailed description of a related known function or component is determined to unnecessarily obscure the gist of the present disclosure, the detailed description will be omitted. Components having substantially the same functional configuration among the drawings are given the same reference numbers and symbols as possible even if they are shown in different drawings. For convenience of explanation, devices and methods are described together when necessary. Each operation of the present disclosure does not necessarily have to be performed in the described order and may be performed in parallel, selectively, or individually.

[0033] The terms used in the embodiments of this disclosure have been selected from widely used, current terms, taking into account the functions of the present disclosure. However, these terms may vary depending on the intentions of those skilled in the art, precedents, the emergence of new technologies, etc. Furthermore, in certain cases, the applicant may arbitrarily select terms, and in such cases, their meanings will be described in detail in the description of the relevant embodiments. Therefore, the terms used in this specification should not be defined simply as names of terms, but rather based on their meanings and the overall content of the present disclosure.

[0034] Throughout this disclosure, singular expressions may include plural expressions unless the context clearly dictates otherwise. Terms such as "comprise" or "have" should be understood to indicate the presence of a feature, number, step, operation, component, part, or combination thereof, but do not preclude the presence or addition of one or more other features, numbers, steps, operations, components, parts, or combinations thereof. In other words, when it is said throughout this disclosure that a part "comprises" a certain component, unless specifically stated otherwise, this does not mean that other components may be included, but rather that other components may be excluded.

[0035] Expressions such as "at least one" modify the entire list of elements, not individual elements of the list. For example, "at least one of A, B, and C" and "at least one of A, B, or C" refer to only A, only B, only C, both A and B, both B and C, both A and C, all of A, B, and C, or any combination thereof.

[0036] In addition, terms such as “...unit”, “...module”, etc. described in the present disclosure mean a unit that processes at least one function or operation, which may be implemented as hardware or software, or a combination of hardware and software.

[0037] Throughout this disclosure, when a part is said to be "connected" to another part, this includes not only cases where the parts are "directly connected," but also cases where the parts are "electrically connected" with other elements intervening. Furthermore, when a part is said to "include" a component, this does not exclude other components, but rather includes other components, unless otherwise specifically stated.

[0038] The expression “configured to” as used throughout this disclosure can be used interchangeably with, for example, “suitable for,” “having the capacity to,” “designed to,” “adapted to,” “made to,” or “capable of.” The term “configured to” does not necessarily mean something that is “specifically designed to” in terms of hardware. Instead, in some contexts, the expression “a system configured to” can mean that the system, together with other devices or components, is “capable of.” For example, the phrase “a processor configured (or set) to perform A, B, and C” may mean a dedicated processor (e.g., an embedded processor) for performing those operations, or a generic-purpose processor (e.g., a CPU or application processor) that can perform those operations by executing one or more software programs stored in memory.

[0039] The artificial intelligence-related functions according to the present disclosure are operated via a processor and memory. The processor may be comprised of one or more processors. In this case, one or more processors may be a general-purpose processor such as a CPU, an AP, a DSP (Digital Signal Processor), a graphics-only processor such as a GPU or a VPU (Vision Processing Unit), or an artificial intelligence-only processor such as an NPU. One or more processors control the processing of input data according to predefined operating rules or artificial intelligence models stored in memory. Alternatively, if one or more processors are artificial intelligence-only processors, the artificial intelligence-only processor may be designed with a hardware structure specialized for processing a specific artificial intelligence model.

[0040] The predefined operation rules or artificial intelligence models are characterized by being created through learning. Here, being created through learning means that the basic artificial intelligence model is trained using a learning algorithm using a plurality of learning data, thereby creating a predefined operation rules or artificial intelligence model set to perform a desired characteristic (or purpose). This learning may be performed on the device itself on which the artificial intelligence according to the present disclosure is performed, or may be performed through a separate server and / or system. Examples of the learning algorithm include, but are not limited to, supervised learning, unsupervised learning, semi-supervised learning, or reinforcement learning.

[0041] Throughout the present disclosure, devices may include, but are not limited to, servers, smartphones, tablet PCs, PCs, TVs, smart TVs, mobile phones, personal digital assistants (PDAs), speakers, laptops, media players, microservers, e-book object recognition devices, digital broadcasting object recognition devices, kiosks, MP3 players, digital cameras, robot vacuum cleaners, home appliances, other mobile or non-mobile computing devices, watches, glasses, hair bands, rings, etc. having communication and data processing capabilities. In one embodiment, the devices may execute web- or module-based applications related to the system. For example, a device for measuring the reliability of a molecular structure prediction model may refer to a server, and a web-based application related to a system for measuring the reliability of a molecular structure prediction model may be executed on the server. That is, the server may provide a web service or software for measuring the reliability of a molecular structure prediction model.

[0042] One embodiment of the present disclosure aims to predict a molecular structure from an image of a molecular structure and provide reliability information for the predicted molecular structure. However, the present disclosure is not limited to the molecular structure embodiment and can be applied to various technical fields that extract image information from images rather than text.

[0043] FIG. 1 is a diagram illustrating a method for extracting a molecular structure from an image using a first learning model according to one embodiment of the present disclosure.

[0044] Referring to FIG. 1, a device including a molecular structure prediction model can identify an image (120) of a molecular structure from a document (110), analyze the image (120), and extract a molecular structure corresponding to the image. In one embodiment, a single device may include a molecular structure prediction model and a reliability measurement model, or the molecular structure prediction model and the reliability measurement model may be included in different devices, and each device may calculate an output value using the corresponding model.

[0045] In one embodiment, a device including a molecular structure prediction model can extract an image (120) of a molecular structure from a document (110). For example, the device including the molecular structure prediction model can extract the image (120) of a molecular structure from the document (110) using a panoptic segmentation technique. In addition, the device including the molecular structure prediction model can identify atoms included in the image (120) of the molecular structure and bonds between atoms based on the image (120) of the molecular structure. In one embodiment, the device including the molecular structure prediction model can analyze the types and positions of atoms, types and positions of bonds, etc. included in the image (120) of the molecular structure. For example, vertices implicitly representing carbon, atoms, superatoms, etc., their positions, and types of bonds between them (e.g., single / double / triple bonds, up / down bonds, etc.) can be identified. Optical character recognition (OCR) technology can be used to analyze the types of atoms or superatoms. In one embodiment, a device including a molecular structure prediction model can determine a structural formula (130) of a linear notation corresponding to an image (120) of a molecular structure based on the types and positions of identified atoms, types and positions of bonds, etc. The structural formula of the linear notation may include, but is not limited to, SMILES (Simplified Molecular Input Line Entry System) notation, InChI (International Chemical Identifier), WLN (Wiswesser line notation), ROSDAL (Representation of Organic Structure Descriptions Arranged Linearly), SLN (SYBYL Line Notation), etc.Furthermore, a device including a molecular structure prediction model can determine a graph corresponding to a structural formula (130) of linear notation. The graph corresponding to the structural formula (130) of linear notation includes a CT file (Chemical table file) graph, and may include, for example, a Molfile graph.

[0046] In providing a structural formula (130) or graph of the predicted linear notation in this way, in order to determine whether the predicted result is reliable, a method has been performed such as creating a chemical structural formula with the predicted result and measuring the reliability if the structural formula is capable of actual bonding. However, this method does not consider the actual image, so if it is determined that the structural formula is capable of actual bonding even with an incorrect prediction, a high reliability score is output, which increases the possibility of accumulating incorrect data. Accordingly, a method for providing a more accurate reliability score of the predicted model is required, and this will be described in more detail below with reference to FIGS. 2 to 5.

[0047] FIG. 2 is a flowchart illustrating a method for providing reliability of a result value of a first learning model using a second learning model according to an embodiment of the present disclosure.

[0048] Referring to FIG. 2, a graph (203) of a predicted first molecular structure corresponding to an image (201) of the first molecular structure can be obtained using a molecular structure prediction model (210). The methods described above with reference to FIG. 1 can be used to obtain the graph (203) of the predicted first molecular structure.

[0049] In one embodiment, image rendering may be performed based on a graph (203) of a first molecular structure predicted by a molecular structure prediction model (210). The image rendering may be performed by identifying a plurality of components based on the graph of the first molecular structure, identifying a portion corresponding to each of the plurality of components, and displaying different components with different marks in an image (201) of the first molecular structure, which is an original image. Furthermore, in order to identify a portion corresponding to each of the plurality of components in the image (201) of the first molecular structure, the positions of each of the plurality of components may also be identified. The plurality of components may include a first atom, a second atom, a third atom, a first bond, a second bond, a third bond, etc. For example, referring to the example of FIG. 2, the reliability measurement device of the molecular structure prediction model identifies eight vertices representing bromine (Br), four nitrogens (N), and carbon (C), seven single bonds, five double bonds, and one triple bond based on the graph (203) of the first molecular structure, and displays the elements as circles and the bonds as line segments, but can display different elements or different bonds using different colors. Accordingly, the image rendering result (205) of FIG. 2 can be obtained.

[0050] In one embodiment, the image rendering result (205) and the graph (203) of the predicted first molecular structure may be input to a reliability model (230), and a reliability score (207) may be obtained based on the degree of similarity between the two input information. The more similar the input information of the two input values ​​is, the higher the reliability score (207) may be obtained, and the more dissimilar the input information is, the lower the reliability score (207) may be obtained.

[0051] FIG. 3 is a flowchart illustrating a method for measuring the reliability of a molecular structure prediction model according to one embodiment of the present disclosure.

[0052] Referring to FIG. 3, in operation 310, the reliability measurement device of a molecular structure prediction model may obtain an image of a first molecular structure. In one embodiment, the reliability measurement device of a molecular structure prediction model may obtain an image of the first molecular structure by directly extracting the image of the first molecular structure from a document, or may obtain an image of the first molecular structure extracted from another device. That is, the reliability measurement device of a molecular structure prediction model or an external device may extract the image of the first molecular structure using an artificial intelligence (AI) algorithm or a predefined operation rule.

[0053] In operation 330, the reliability measurement device of the molecular structure prediction model may obtain a graph of the first molecular structure determined using the molecular structure prediction model. That is, the graph of the first molecular structure may be a graph corresponding to an image of the first molecular structure predicted by the molecular structure prediction model. In one embodiment, the molecular structure prediction model may include a first learning model trained to input an image of a molecular structure formula and output a chemical table file (CT file) graph. The CT file graph may include information on each atom of the molecule, xyz coordinate information of the corresponding atom, bonding information between atoms, etc. Similar to operation 310, the reliability measurement device of the molecular structure prediction model may obtain a graph of the first molecular structure determined using the molecular structure prediction model directly, or may obtain a graph of the first molecular structure determined using the molecular structure prediction model by receiving it from an external device.

[0054] In operation 350, the reliability measurement device of the molecular structure prediction model may perform image rendering on an image of the first molecular structure based on a graph of the first molecular structure. In one embodiment, the reliability measurement device of the molecular structure prediction model may perform image rendering by identifying a plurality of components based on the graph of the first molecular structure, identifying a portion corresponding to each of the plurality of components in the image of the first molecular structure, and distinguishing different components with different marks. Here, the plurality of components may include atoms, bonds between atoms, etc. Distinguishing with different marks may include distinguishing with different colors or different shapes, but is not limited thereto, and may include various forms that are displayed as being distinguished from each other.

[0055] For example, if the first component is carbon and the second component is nitrogen, the reliability measurement device of the molecular structure prediction model can perform image rendering by identifying the position coordinates of the first component, carbon, and carbon, in the graph of the first molecular structure, and identifying the position coordinates of the second component, nitrogen, and nitrogen, and identifying the first part corresponding to the first component and the second part corresponding to the second component in the image of the first molecular structure based on each position coordinate, and distinguishing them with different marks, such as marking the first part in yellow and marking the second part in red.

[0056] For another example, if the first component is carbon, the second component is nitrogen, the third component is a single bond, and the fourth component is a double bond, the reliability measurement device of the molecular structure prediction model can identify the positions of the carbon and carbon as the first component in the graph of the first molecular structure, identify the positions of the nitrogen and nitrogen as the second component, identify the positions of the single bond as the third component, and identify the positions of the double bond as the fourth component. In addition, the reliability measurement device of the molecular structure prediction model can perform image rendering by identifying a first part corresponding to the first component, a second part corresponding to the second component, a third part corresponding to the third component, and a fourth part corresponding to the fourth component in the image of the first molecular structure based on the positions of each component, and displaying the first and second parts corresponding to the elements in the form of circles but with different colors to indicate different elements, and displaying the third and fourth parts in the form of bars but with different colors to indicate different bonds.

[0057] In operation 370, a reliability measurement device of a molecular structure prediction model may determine the reliability of the graph of the first molecular structure based on the image rendering result and the graph of the first molecular structure. In one embodiment, the reliability of the graph of the first molecular structure may be determined using a second learning model. The second learning model may be a learning model that takes the image rendering result and the graph of the first molecular structure as inputs and outputs a reliability value (e.g., a reliability score). In one embodiment, the learning model may include an image backbone network model that extracts features of the image rendering result, a graph backbone network model that extracts features of the graph of the first molecular structure, a feature connector that connects features of the extracted image rendering result and features of the extracted graph of the first molecular structure, and an extraction network model that extracts reliability by taking the output of the feature connector as input. This will be described in more detail below with reference to FIG. 5.

[0058] In one embodiment, the second learning model may be trained to output a first value when the image rendering result and the graph of the first molecular structure match, and to output a second value when the image rendering result and the graph of the first molecular structure do not match. For example, the second learning model may be trained to output 1 when the image rendering result and the graph of the first molecular structure match during the learning step, and 0 when they do not match, so that after learning, when the prediction result of the molecular structure prediction model is correctly predicted, a value similar to 1 may be extracted, and if the prediction contains an error, a value similar to 0 may be output. The reliability measuring device of the molecular structure prediction model may determine the reliability of the graph of the first molecular structure based on the value extracted by the second learning model. For example, the reliability may be determined as a reliability value (or score).

[0059] In one embodiment, a reliability measurement device for a molecular structure prediction model provides a reliability score along with the predicted graph of the first molecular structure, thereby providing the accuracy of the predicted graph of the first molecular structure and helping determine whether to utilize it. Accordingly, a user or the user's device may decide to store the graph in a database only if the reliability score is above a predetermined level (or value), thereby enabling accurate data to be utilized for research, etc.

[0060] FIG. 4A is a diagram illustrating a method for measuring reliability when a molecular structure prediction model according to one embodiment of the present disclosure makes an incorrect prediction.

[0061] Referring to FIG. 4A, a first molecular structure can be predicted by a molecular structure prediction model based on an image (410) of the first molecular structure. For example, the first molecular structure may be a structure formed by the coexistence of single bonds and double bonds. However, unlike humans, a machine cannot require 100% accuracy in predicting the first molecular structure using the image (410) of the first molecular structure. Accordingly, a process is required to confirm whether the first molecular structure predicted based on the image (410) of the first molecular structure is correctly predicted.

[0062] In one embodiment, the reliability measurement device of the molecular structure prediction model can obtain an image (410) of the first molecular structure and a graph of the first molecular structure predicted by the first learning model. In addition, the reliability measurement device of the molecular structure prediction model can identify the type of element, the position of the element, the type of bond, the position of the bond, etc. based on the graph of the predicted first molecular structure and render this on the image (410) of the first molecular structure. For example, if the first bond (420) is identified as a double bond in the graph of the first molecular structure, the reliability measurement device of the molecular structure prediction model can generate an image rendering result (430) by displaying bonds identified as a double bond, including the first bond (440), and bonds identified as a single bond in different colors on the image (410) of the first molecular structure. That is, in reality, the first bond (420) is also a single bond, but the image rendering result may indicate that the first bond (420) has a double bond.

[0063] In one embodiment, the reliability measurement device of the molecular structure prediction model can determine the reliability of the graph of the first molecular structure based on the image rendering result. For example, the reliability measurement device of the molecular structure prediction model can identify the element portion and the bond portion by using the segmentation of the image (410) of the first molecular structure, and can identify whether they are the same element or different elements, or the same bond or different bond. If the segmentation result of the image (410) of the first molecular structure and the image rendering result (430) are similar, the reliability measurement device of the molecular structure prediction model can determine the reliability as high, and if they are not similar, the reliability as low. In the example of FIG. 4A, the image rendering result (430) determines that the first bond (440) has a double bond, but the image (410) of the first molecular structure determines that the first bond (420) has a single bond, and therefore, the reliability of the graph of the first molecular structure can be determined to be low. Alternatively, the reliability score of the graph of the first molecular structure can be determined to be low.

[0064] Alternatively, in one embodiment, the reliability measurement device for a molecular structure prediction model can determine the reliability of the graph of the first molecular structure using both the image rendering results and the predicted graph of the first molecular structure. In one embodiment, the molecular structure graph is also used to determine the reliability, thereby increasing the accuracy of the determined reliability. This will be described in more detail below with reference to FIG. 5.

[0065] FIG. 4b is a diagram illustrating a method for measuring reliability when a molecular structure prediction model according to one embodiment of the present disclosure correctly predicts.

[0066] Referring to FIG. 4B, a second molecular structure can be predicted by a molecular structure prediction model based on an image (450) of the second molecular structure. However, as described above in FIG. 4A, unlike humans, a machine cannot require 100% accuracy in predicting the second molecular structure using the image (450) of the second molecular structure. Accordingly, a process is required to confirm whether the second molecular structure predicted based on the image (450) of the second molecular structure is correctly predicted.

[0067] In one embodiment, the reliability measurement device of the molecular structure prediction model can obtain an image (450) of the second molecular structure and a graph of the second molecular structure predicted by the first learning model. In addition, the reliability measurement device of the molecular structure prediction model can identify the type of element, the position of the element, the type of bond, the position of the bond, etc. based on the graph of the predicted second molecular structure and render this on the image (450) of the second molecular structure. For example, if the graph of the second molecular structure shows that bonds except the second bond (460) and the third bond (470) are single bonds, and the second bond (460) is a bond protruding from the plane and the third bond (470) is a bond retracted from the plane, the reliability measurement device of the molecular structure prediction model can display the second bond (490), the third bond (495), and the other bonds in different colors on the image (450) of the second molecular structure to generate an image rendering result (480). Furthermore, the reliability measurement device of the molecular structure prediction model may generate an image rendering result (480) to indicate the bonding positions of the second bond and the third bond.

[0068] In one embodiment, the reliability measurement device of the molecular structure prediction model can determine the reliability of the graph of the second molecular structure based on the image rendering result (480). For example, the reliability measurement device of the molecular structure prediction model can identify the element portion and the bond portion by using the segmentation of the image (450) of the second molecular structure, and can identify whether they are the same element or different elements, or the same bond or different bond. If the segmentation result of the image (450) of the second molecular structure and the image rendering result (480) are similar, the reliability measurement device of the molecular structure prediction model can determine that the reliability is high, and if they are not similar, the reliability is low. In the example of FIG. 4B, since the image rendering result (480) and the image (450) of the second molecular structure are similar, the reliability of the graph of the second molecular structure can be determined to be high. Alternatively, the reliability score of the graph of the second molecular structure can be determined to be high.

[0069] Alternatively, in one embodiment, a reliability measurement device for a molecular structure prediction model may determine the reliability of a graph of the first molecular structure based on the image rendering results and the graph of the predicted first molecular structure. This will be described in more detail below with reference to FIG. 5.

[0070] FIG. 5 is an example diagram of a second learning model according to one embodiment of the present disclosure.

[0071] Referring to FIG. 5, the second learning model (530) for measuring reliability may include an image backbone model (540) for extracting features of an image rendering result (510), a graph backbone model (550) for extracting features of a graph of a molecular structure (520), a feature connection unit (560) for connecting features of the extracted image rendering result (510) and features of the graph of the extracted molecular structure (520), and a network model (570) for determining reliability by using the output of the feature connection unit (560) as input. The network model (570) for determining reliability may be a model that classifies connected features through a linear layer.

[0072] In one embodiment, the second learning model (530) may be an artificial intelligence model that takes as input an image rendering result (510) and a graph of a molecular structure (520) and extracts a confidence value (580). At this time, the image rendering result (510) becomes an input of the image backbone model (540), and the graph of the molecular structure (520) becomes an input of the graph backbone model (550), and thus, they may become inputs of different network models. The image backbone model (540) may extract features of the image rendering result (510), and the graph backbone model (550) may extract features of the graph of the molecular structure (520). The feature linking unit (560) may link features extracted from each backbone model, and the network model (570) that determines the confidence value may determine the confidence value based on the extracted features. For example, a higher confidence value may be output as the confidence level is higher, and a lower confidence value may be output as the confidence level is lower.

[0073] FIG. 6 is a block diagram of a reliability measurement device for a molecular structure prediction model according to one embodiment of the present disclosure.

[0074] Referring to FIG. 6, the reliability measurement device (600) of a molecular structure prediction model may include a transceiver (610), a memory (620), a database (630), and a processor (640). However, not all of the components illustrated in FIG. 6 are essential components of the reliability measurement device (600) of a molecular structure prediction model. The reliability measurement device (600) of a molecular structure prediction model may be implemented with more components than the components illustrated in FIG. 6, or the reliability measurement device (600) of a molecular structure prediction model may be implemented with fewer components than the components illustrated in FIG. 6. In addition, the transceiver (610), the memory (620), and the processor (640) may be implemented in the form of a single chip.

[0075] In one embodiment, the transceiver (610) may communicate with a terminal or other electronic device connected wired or wirelessly to the reliability measurement device (600) of the molecular structure prediction model. For example, the transceiver (610) may obtain an image of the first molecular structure, a graph of the first molecular structure determined using the molecular structure prediction model, etc. from the other electronic device.

[0076] Various types of data, such as programs and files, such as applications, can be installed and stored in the memory (620). The processor (640) can access and use data stored in the memory (620), or store new data in the memory (620). In addition, the memory (620) can store one or more instructions. The processor (640) can execute one or more instructions stored in the memory.

[0077] The processor (640) controls the overall operation of the reliability measurement device (600) of the molecular structure prediction model, and may include at least one processor, such as a CPU or a GPU. The processor (640) may control other components included in the reliability measurement device (600) of the molecular structure prediction model to perform operations for operating the reliability measurement device (600) of the molecular structure prediction model. For example, the processor (640) may obtain an image of a first molecular structure, obtain a graph of the first molecular structure determined using the molecular structure prediction model, perform image rendering on the image of the first molecular structure based on the graph of the first molecular structure, and determine the reliability of the graph of the first molecular structure based on the image rendering result and the graph of the first molecular structure.

[0078] The database (630) can store various learning data for training the learning model. Furthermore, the database (630) may store material information, phase information, simulation result information, and the like, and in various embodiments, output data output by the learning model may also be stored. While FIG. 6 illustrates the reliability measurement device (600) for a molecular structure prediction model as including the database (630), the database (630) may also be provided externally to the device. In this case, the database (630) may be connected to the reliability measurement device (600) for a molecular structure prediction model via wired or wireless connections.

[0079] Additionally, the learning model may be implemented outside the reliability measurement device (600) of the molecular structure prediction model (e.g., implemented cloud-based) or may be included within the reliability measurement device (600) of the molecular structure prediction model.

[0080] An embodiment of the present disclosure may also be implemented in the form of a recording medium containing computer-executable instructions, such as program modules, executed by a computer. Computer-readable media may be any available media that can be accessed by a computer, and includes both volatile and nonvolatile media, removable and non-removable media. Furthermore, computer-readable media may include both computer storage media and communication media. Computer storage media includes both volatile and nonvolatile, removable and non-removable media implemented in any method or technology for storing information, such as computer-readable instructions, data structures, program modules, or other data. Communication media typically contains computer-readable instructions, data structures, or program modules, and includes any information delivery media.

[0081] The above description of the present disclosure is provided for illustrative purposes only, and those skilled in the art will readily appreciate that the present disclosure can be readily modified into other specific forms without altering the technical spirit or essential characteristics of the present invention. Therefore, the embodiments described above should be understood as illustrative in all respects and not restrictive. For example, components described as being single may be implemented in a distributed manner, and similarly, components described as being distributed may be implemented in a combined manner.

[0082] The scope of the present disclosure is indicated by the claims described below rather than the detailed description above, and all changes or modifications derived from the meaning and scope of the claims and their equivalent concepts should be interpreted as being included in the scope of the present disclosure.

Claims

1. In a system for measuring the reliability of a molecular structure prediction model, memory for storing one or more instructions; and At least one processor for executing one or more instructions stored in the memory, The at least one processor executes the one or more instructions, Obtain an image of the first molecular structure, Obtaining a graph of the first molecular structure determined using the above molecular structure prediction model, Based on the graph of the first molecular structure, image rendering is performed on the image of the first molecular structure, A system for determining the reliability of the graph of the first molecular structure based on the image rendering result and the graph of the first molecular structure.

2. In the first paragraph, the at least one processor, Based on the graph of the first molecular structure, at least one of the first component and the second component is identified, Identifying a first part corresponding to the first component in the image of the first molecular structure, Identifying a second portion corresponding to the second component in the image of the first molecular structure, The image rendering is performed by distinguishing the first part and the second part with different marks, Each of the first component and the second component, A system comprising any one of a first atom, a second atom, a first bond, and a second bond.

3. In the first paragraph, the molecular structure prediction model is Includes the first learning model, The above first learning model is, A system trained to extract a chemical table file graph using a molecular structure image as input.

4. In the first paragraph, at least one processor, A system that outputs reliability using a second learning model by taking the image rendering result and the graph of the first molecular structure as input.

5. In the fourth paragraph, the second learning model, Image backbone model that extracts features of the image rendering result; A graph backbone model that extracts features of the graph of the first molecular structure; A feature linking portion connecting features of the image rendering result and features of the graph of the first molecular structure; and A system comprising a linear layer model that determines reliability by taking as input the output of the above-mentioned feature connection part.

6. In the fourth paragraph, the second learning model, A system that is trained to output a first value when the image rendering result and the graph of the first molecular structure match, and to output a second value when the image rendering result and the graph of the first molecular structure do not match.

7. In paragraph 1, A system in which a graph of a molecular structure having a reliability level above a predetermined level is stored in a database.

8. A method for measuring the reliability of a molecular structure prediction model, performed by at least one processor, A step of acquiring an image of the first molecular structure; A step of obtaining a graph of the first molecular structure using the above molecular structure prediction model; A step of performing image rendering on an image of the first molecular structure based on the graph of the first molecular structure; and A method comprising a step of determining the reliability of the graph of the first molecular structure based on the image rendering result and the graph of the first molecular structure.

9. In the 8th paragraph, the step of performing image rendering on the image of the first molecular structure is: A step of identifying at least one of the first component and the second component based on the graph of the first molecular structure; A step of identifying a first portion corresponding to the first component in the image of the first molecular structure; A step of identifying a second portion corresponding to the second component in the image of the first molecular structure; and A step of performing image rendering by distinguishing the first part and the second part with different marks, Each of the first component and the second component, A method comprising any one of a first atom, a second atom, a first bond, and a second bond.

10. In paragraph 8, the molecular structure prediction model is Includes the first learning model, The above first learning model is, A method learned to extract a chemical table file graph by inputting a molecular structure image.

11. In the 8th paragraph, the step of outputting the reliability of the graph of the first molecular structure is: A method comprising a step of outputting the reliability of the graph of the first molecular structure using a second learning model by taking the image rendering result and the graph of the first molecular structure as input.

12. In the 11th paragraph, the second learning model, Image backbone model that extracts features of the image rendering result; A graph backbone model that extracts features of the graph of the first molecular structure; A feature linking portion connecting features of the image rendering result and features of the graph of the first molecular structure; and A method comprising a linear layer model that determines reliability by taking as input the output of the above feature connection part.

13. In paragraph 11, the second learning model, A method that is trained to output a first value when the image rendering result and the graph of the first molecular structure match, and to output a second value when the image rendering result and the graph of the first molecular structure do not match.

14. In paragraph 8, A method in which a graph of a molecular structure having a reliability greater than a specific value is stored in a database.

15. A program stored on a computer-readable recording medium that causes a computer to execute the method of any one of clauses 8 to 14.

Images