Apparatus and method for determining experimental priority of candidate reactant combinations for synthesizing a target product using a neural network.

Abstract

To provide an apparatus for synthesizing target products by using neural networks and a method for the same.SOLUTION: An apparatus for synthesizing target products by using neural networks comprises: a memory which stores at least one program; and a processor which executes at least one program, and is configured to: predict candidate reactant combinations for generating the target product by using a pre-trained retrosynthesis prediction model; predict a prediction product with respect to each of the candidate reactant combinations by using a pre-trained reaction prediction model; and determine an experimental priority order of the candidate reactant combinations based on a result of comparing the target product with the prediction product.SELECTED DRAWING: Figure 2

Description

[Technical Field]

[0001] This invention utilizes a neural network to synthesize target products. Determine the experimental priority of candidate reactant combinations. This relates to the apparatus and the method thereof. [Background technology]

[0002] A neural network is a computer science architecture that models the biological brain. With the development of neural network technology, various types of electronic systems are now utilizing neural networks to analyze input data and extract useful information.

[0003] There is a need for technology that can use neural networks to derive effective and appropriate reactants and provide a variety of candidate reactants. [Prior art documents] [Patent Documents]

[0004] [Patent Document 1] Japanese Patent Publication No. 2020-107338 [Overview of the Initiative] [Problems that the invention aims to solve]

[0005] The present invention has been made in view of the above-mentioned prior art, and the object of the present invention is to synthesize a target product using a neural network. Determine the experimental priority of candidate reactant combinations. The objective is to provide an apparatus and a method thereof. [Means for solving the problem]

[0006] To achieve the above objective, a target product is synthesized using a neural network according to one aspect of the present invention. Determine the experimental priority of candidate reactant combinations. The method is, Based on categorical latent variables Using a pre-trained retrosynthesis predictive model The aforementioned Target product If The process involves predicting combinations of candidate reactants to achieve the desired reaction, and then using a previously trained reaction prediction model. The combination of candidate reactants predicted by the previously trained retrosynthesis prediction model and the categorical latent variables shared with the previously trained retrosynthesis prediction model are input. A step of predicting the predicted product for each combination of the candidate reactants, and based on the comparison result of the target product and the predicted product, The likelihood of agreement between the predicted product and the target product is shown in descending order. The process includes the step of determining the experimental priority of the candidate reactant combinations.

[0007] To achieve the above objective, a target product is generated using a neural network according to one aspect of the present invention. Determine the experimental priority of candidate reactant combinations. The device comprises a memory in which at least one program is stored, and a processor that executes the at least one program, the processor is Based on categorical latent variables Using a pre-trained retrosynthesis predictive model The aforementioned Target product If We predict the candidate reactant combinations to achieve the reaction and use a previously trained reaction prediction model. The combination of candidate reactants predicted by the previously trained retrosynthesis prediction model and the categorical latent variables shared with the previously trained retrosynthesis prediction model are input. Predict the predicted product for each combination of the candidate reactants, and based on the comparison result between the target product and the predicted product, The likelihood of agreement between the predicted product and the target product is shown in descending order. Determine the experimental priority for the combinations of candidate reactants. [Effects of the Invention]

[0008] According to the neural network device of the present invention, the combination of candidate reactants derived via the retrosynthesis prediction model is verified via the reaction prediction model, thereby preventing cases where combinations of candidate reactants that violate the structural formula grammar are output and cases where the target product cannot be synthesized in string format. [Brief explanation of the drawing]

[0009] [Figure 1] This is a block diagram showing the hardware configuration of a neural network device according to one embodiment. [Figure 2]A diagram for explaining a process of determining an experimental priority order of a combination of candidate reactants based on a comparison result between a target product and a predicted product according to an embodiment. [Figure 3] A diagram for explaining a categorical latent variable. [Figure 4] A diagram for explaining operations performed by an inverse synthesis prediction model and a reaction prediction model according to an embodiment. [Figure 5] A diagram for explaining operations performed by an inverse synthesis prediction model and a reaction prediction model according to an embodiment. [Figure 6] A diagram for explaining operations performed by an inverse synthesis prediction model and a reaction prediction model according to an embodiment. [Figure 7] A flowchart for explaining an operation method of a neural network device according to an embodiment. [Figure 8] A flowchart for explaining a learning method of an inverse synthesis prediction model and a reaction prediction model according to an embodiment. [Figure 9] A flowchart for explaining an operation method of an inverse synthesis prediction model according to an embodiment. [Figure 10] A flowchart for explaining an operation method of a reaction prediction model according to an embodiment. [Figure 11] A flowchart for explaining a method of determining an experimental priority order of a combination of candidate reactants according to an embodiment.

Embodiments for Carrying Out the Invention

[0010] Hereinafter, specific examples of embodiments for carrying out the present invention will be described in detail with reference to the drawings. In this specification, phrases such as "in some embodiments" or "in one embodiment" described in various places do not necessarily indicate the same embodiment.

[0011] Embodiments of the present invention are represented by functional block configurations and various processing stages. Some or all of such functional blocks are embodied by various hardware and / or software configurations that perform specific functions. For example, a functional block of the present invention may be embodied by one or more microprocessors or by a circuit configuration for a given function. Alternatively, for example, a functional block of the present invention may be embodied by various programming or scripting languages. A functional block may be embodied by an algorithm executed by one or more processors. The present invention also employs prior art for electronic environment configuration, signal processing, and / or data processing. Terms such as “mechanism,” “element,” “means,” and “configuration” are general, mechanical, and not limited to physical configurations.

[0012] Furthermore, the connecting lines or members shown in the drawings between components are merely illustrative examples of functional and / or physical or circuit connections. In actual devices, connections between components are indicated by a variety of functional, physical, or circuit connections that may be substituted or added.

[0013] In this specification, the term "structure," as used in neural network systems, refers to the atomic-level structure of a substance. A structure is a structural formula based on the bonds between atoms. For example, a structure can be expressed as a simple string (one-dimensional) representation. Examples of string representations of structures include SMILES (simplified molecular-input line-entry system) codes, SMARTS (smiles arbitrary target specification) codes, or InChi (international chemical identifier) ​​codes.

[0014] Furthermore, a descriptor is an index used to represent the characteristics of a substance, and is a value obtained by performing relatively simple calculations on a given substance. In one embodiment, the descriptor includes a descriptor of quantitative structure-property relationships (QSPR) composed of values ​​that can be calculated immediately, such as a molecular structure fingerprint (e.g., Morgan Fingerprint, ECFP (extended connectivity fingerprint)) indicating whether or not a particular part of the structure is included, or the molecular weight or the number of substructures (e.g., rings) contained within the molecular structure.

[0015] Furthermore, material properties refer to the characteristics of a material and are real values ​​that are measured through experiments or calculated through simulations. For example, if the material is a display material, the properties would be the transmission wavelength and emission wavelength for light, and if the material is a battery material, they would be the voltage. Unlike representations, calculating material properties requires complex simulations and consumes a lot of time.

[0016] Figure 1 is a block diagram showing the hardware configuration of a neural network device according to one embodiment.

[0017] The neural network device 100 can be implemented in a variety of devices, such as PCs (personal computers), server devices, mobile devices, and embedded devices. Specific examples include smartphones, tablet devices, AR (augmented reality) devices, IoT (Internet of Things) devices, autonomous vehicles, robotics, and medical devices that utilize neural networks for speech recognition, image recognition, and image classification, but it is not limited to these. Furthermore, the neural network device 100 corresponds to a dedicated hardware accelerator (HW accelerator) mounted on the aforementioned devices. The neural network device 100 is a hardware accelerator such as an NPU (neural processing unit), TPU (Tensor processing unit), or Neural Engine, which are dedicated modules for driving neural networks, but it is not limited to these.

[0018] Referring to Figure 1, the neural network device 100 includes a processor 110, memory 120, and a user interface 130. Figure 1 only illustrates the components relating to this embodiment. Therefore, it will be obvious to those skilled in the art that the neural network device 100 may further include other general-purpose components in addition to those shown in Figure 1.

[0019] The processor 110 plays a role in controlling the overall functions for executing the neural network device 100. For example, the processor 110 controls the neural network device 100 overall by executing a program stored in the memory 120 within the neural network device 100. The processor 110 is embodied by, but is not limited to, a CPU (central processing unit), GPU (graphics processing unit), AP (application processor), etc., provided within the neural network device 100.

[0020] Memory 120 is hardware that stores various data processed within the neural network device 100. For example, memory 120 stores data processed by the neural network device 100 and data being processed. Memory 120 also stores applications, drivers, etc., driven by the neural network device 100. Memory 120 includes RAM (random access memory) such as DRAM (dynamic random access memory) and SRAM (static random access memory), ROM (read-only memory), EEPROM (electrically erasable programmable read-only memory), CD-ROM (compact disc read-only memory), Blu-ray, or other optical disc storage, HDD (hard disk drive), SSD (solid static drive), or flash memory.

[0021] The processor 110 drives at least one of the following: a retrosynthesis prediction model, a synthesis prediction model, a yield prediction model, and a reaction scheme model.

[0022] The aforementioned retrosynthesis prediction model, synthesis prediction model, yield prediction model, and reaction scheme model are embodied as transformer models. The neural network device 100 drives the transformer models, enabling parallel processing and rapid computation of data.

[0023] The processor 110 uses the test dataset to train the transformer model, during which categorical latent variables are determined or learned. For example, the processor 110 learns the conditional probability distribution of the categorical latent variables for each input of the test product. The prediction yield rate and reaction method are also provided to train the conditional probability distribution of the latent variables.

[0024] The test dataset includes test reactants and test products corresponding to each combination of test reactants. Furthermore, the test dataset includes the predicted yield of the test products under experimental conditions and the reaction scheme for each combination of test reactants.

[0025] Experimental conditions refer to the various conditions set to carry out an experiment that uses reactants to produce a product. For example, experimental conditions include at least one of the following: catalyst, base, solvent, reagent, temperature, and reaction time.

[0026] The predicted yield represents the expected yield of the product produced from the reactants when the reactants are subjected to a predetermined reaction scheme and experimental conditions. The predicted yield differs from the actual measured yield obtained through experiments.

[0027] A reaction scheme refers to a chemical reaction method used to produce a product from reactants. For example, when producing a halide (R2-BY2) from the reactant organoboron (R1-BY2), the reaction scheme is the Suzuki-Miyaura scheme. Multiple reaction schemes are provided based on the structural information of the reactants and the products.

[0028] The processor 110 uses a pre-trained retrosynthesis prediction model to predict candidate reactant combinations for producing the target product.

[0029] The processor 110 inputs the chemical structure of the target product in a string (one-dimensional) format. The structure is a structural formula based on the linkage relationships between atoms. For example, the processor 110 inputs the chemical structure of the target product in the format of SMILES code, SMARTS code, or InChi code. In one embodiment, information regarding the representation and physical properties of the target product is input along with the chemical structure in string format.

[0030] The processor 110 drives a pre-trained retrosynthesis prediction model, uses the string of the target product as input data to perform calculations, and generates candidate reactant combinations as output data in string format based on the calculation results.

[0031] The processor 110 uses a previously trained reaction prediction model to predict the predicted product for each combination of candidate reactants.

[0032] The processor 110 drives a pre-trained reaction prediction model, uses candidate reactant combinations as input data to perform calculations, and generates predicted products as string-formatted output data based on the calculation results.

[0033] The processor 110 drives a pre-trained yield prediction model, using candidate reactant combinations as input data to perform calculations, and based on the calculation results, generates the predicted yield of the target product and the reaction scheme for producing the target product as output data. The predicted yield of the target product is used as data for training a categorical latent variable.

[0034] Furthermore, the processor 110 drives a pre-trained reaction pattern prediction model, uses the target product as input data to perform calculations, and generates a reaction pattern for producing the target product as output data based on the calculation results. The reaction pattern for producing the target product is used as data for training categorical latent variables.

[0035] The processor 110 compares the target product with the predicted product and determines the experimental priority of candidate reactant combinations based on the comparison results. The optimal candidate reactant combination for producing the target product is then output according to priority.

[0036] The user interface 130 refers to an input means for providing feedback on experimental results. For example, the user interface 130 may include, but is not limited to, a keypad, a dome switch, a touchpad (such as a contact-type capacitive touchpad, a pressure-type resistive touchpad, an infrared sensing touchpad, a surface ultrasonic conduction touchpad, an integral tension measurement touchpad, or a piezoelectric effect touchpad), a jog wheel, or a jog switch. The processor 110 updates the transformer model based on the feedback of the experimental results.

[0037] In the following description of this embodiment, as described above, a method for preferentially providing optimized experimental conditions using the neural network device 100 will be explained in detail. The method described below is carried out by the processor 110, memory 120, and user interface 130 of the neural network device 100.

[0038] Figure 2 illustrates the process of determining the experimental priority of candidate reactant combinations based on the comparison results of the target product and the predicted product according to one embodiment, and Figure 3 illustrates categorical latent variables.

[0039] Referring to Figure 2, the processor 110 receives the chemical structure of the target product 210 as input in string format.

[0040] The target product 211, in string format, is input to the retrosynthetic predictive model 220. Additionally, the categorical latent variable 230 is input to the retrosynthetic predictive model 220.

[0041] The categorical latent variable 230 is a variable that is not directly observable but influences information regarding candidate reactant combinations 240. In one embodiment, the categorical latent variable 230 indicates the type of reaction. The type of reaction includes reaction forms such as decomposition, combustion, metathesis, and displacement; experimental conditions such as catalyst, base, solvent, reagent, temperature, and reaction time; and reaction schemes such as Suzuki-Miyaura.

[0042] The categorical latent variable 230 contains multiple classes. These classes are generated according to the type of reaction. The categorical latent variable 230 changes the cluster of candidate reactants or combinations of candidate reactants.

[0043] The retrosynthetic predictive model 220 is a probabilistic model that depends on an unobserved categorical latent variable. In one embodiment, the retrosynthetic predictive model is a Gaussian mixture model consisting of multiple normal distributions having different parameters (e.g., expected value, variance, etc.) determined by the categorical latent variable 230, and the information regarding candidate reactants or combinations of candidate reactants is determined by the categorical latent variable 230.

[0044] The processor 110 outputs a combination of candidate reactants 240 for generating the target product 210 using a pre-trained retrosynthesis prediction model 220. The combination of candidate reactants 240 is output in string format. In other words, the processor 110 outputs a combination of candidate reactants 240 in string format based on the target product 211 and the categorical latent variable 230. In one embodiment, the processor 110 outputs a combination of candidate reactants 240 that maximizes both the class likelihood for the target product input and the likelihood of the retrosynthesis prediction result.

[0045] The candidate reactant combination 240 represents the set of reactant combinations predicted by the processor 110 to produce the target product 211. The candidate reactant combination 240 may contain multiple reactants, but is not limited to that number. In other words, the candidate reactant combination 240 may contain only one reactant.

[0046] On the other hand, since the retrosynthesis prediction model 220 is input with categorical latent variables 230, a diversity of candidate reactant combinations 240 is ensured, as shown in Figure 3.

[0047] Referring to Figure 3, when the processor 110 sets one of the classes included in the categorical latent variable 230 as the initial token and performs decoding based on the set initial token, multiple combinations of candidate reactants (220a, 220b, 220c) are generated for a single target product 210. The combinations of candidate reactants (220a, 220b, 220c) are clustered according to the class-specific conditional probabilities of the combinations of candidate reactants (220a, 220b, 220c) based on the input of the target product 210. Figure 3 illustrates the clustering process of the combinations of candidate reactants (220a, 220b, 220c) based on the input of the target product 210 when there are three classes.

[0048] The processor 110 calculates the class likelihood of each of the candidate reactant combinations 240 based on the input of the target product 211. The processor 110 also selects a predetermined number of final candidate reactant combinations from the candidate reactant combinations 240 based on the class likelihood of each of the candidate reactant combinations 240. In one embodiment, the processor 110 can also calculate the class log likelihood of each of the candidate reactant combinations 240 and select a predetermined number of final candidate reactant combinations based on the class log likelihood. The final candidate reactant combinations are input into the reaction prediction model 250 and used to predict the predicted product 260.

[0049] Referring again to Figure 2, the candidate reactant combinations 240 are input into the reaction prediction model 250. The categorical latent variables 230 are also input into the reaction prediction model 250.

[0050] The processor 110 uses a pre-trained reaction prediction model 250 to output a prediction product 260 for each of the candidate reactant combinations 240. The prediction product 260 is output in string format. In other words, the processor 110 outputs a prediction product 261 in string format based on the candidate reactant combinations 240 and the categorical latent variables 230. In one embodiment, the processor 110 outputs a prediction product 260 that maximizes both the class-specific likelihood and the likelihood of the reaction prediction result for the input of the candidate reactant combinations 240.

[0051] The predicted product 260 refers to the product predicted from the reactants included in the candidate reactant combination 240.

[0052] On the other hand, since the categorical latent variables 230 are also input to the reaction prediction model 250, diversity of the predicted products 260 is also ensured.

[0053] The processor 110 compares the predicted product 260 and the target product 210 for each input of the candidate reactant combination 240. Based on the comparison results between the predicted product 260 and the target product 210, the processor 110 also determines the experimental priority of the candidate reactant combination 240.

[0054] Processor 110 outputs 270 combinations of candidate reactants that have been rearranged and prioritized by experimental order. Processor 110 outputs 280 combinations of candidate reactants that have been rearranged in structural formula form.

[0055] The neural network device 100 of the present invention verifies the combination of candidate reactants 240 derived via the retrosynthesis prediction model 220 via the reaction prediction model 250, thereby preventing cases where a combination of candidate reactants 240 that violates the structural formula grammar is output and cases where the target product 210 cannot be synthesized in string format.

[0056] Figure 4 is a diagram illustrating the calculations performed in the retrosynthesis prediction model and reaction prediction model according to one embodiment.

[0057] Referring to Figure 4, the processor 110 receives information about the target product 210. This information is represented in string format. The string format of the target product 210 is input to the retrosynthesis prediction model 220.

[0058] The processor 110 predicts a categorical latent variable 230 based on information about the target product 210. The categorical latent variable 230 refers to a variable that is not directly observed but influences information about candidate reactant combinations 240, and includes multiple classes.

[0059] The processor 110 predicts candidate reactant combinations 240 for producing the target product 210 by driving a retrosynthesis prediction model 220 based on information about the target product 210 and a categorical latent variable 230.

[0060] The retrosynthesis prediction model 220 includes a shared word embedder 411, position encoding units (412a, 412b, 412c, 412d) (position encoding unit 412 if distinction is not necessary), a shared encoder 413, a retrosynthesis decoder 414, and a shared word generator 415. In Figure 4, the word embedding unit 411, position encoding unit 412, retrosynthesis encoder unit 413, retrosynthesis decoder unit 414, and word generator unit 415 are shown as any single unit included in the retrosynthesis prediction model 220, but the word embedding unit 411, position encoding unit 412, retrosynthesis encoder unit 413, retrosynthesis decoder unit 414, and word generator unit 415 represent layers included in the retrosynthesis prediction model 220.

[0061] The word embedding unit 411 embeds the input data character by character. The word embedding unit 411 maps the target product 210 in string format to a vector of a predefined dimension. The word embedding unit 411 also maps the categorical latent variable 230 to a vector of a predefined dimension.

[0062] The position encoding unit 412 performs positional encoding to identify the position of each character contained in the input data. In one embodiment, the position encoding unit 412 encodes the input data using sine waves of different frequencies.

[0063] The first position encoding unit 412a performs position encoding of the target product 210 in string format. The first position encoding unit 412a combines the position information with the embedded input data and provides it to the inverse synthesis encoder unit 413.

[0064] The second position encoding unit 412b performs position encoding of the categorical latent variable 230 as an initial token. The second position encoding unit 412b combines the position information with the embedded input data and provides it to the desynthesized decoder unit 414.

[0065] The inverse synthesis encoder unit 413 includes a self-attention sub-hierarchy and a feed-forward sub-hierarchy. For the sake of explanation, Figure 4 shows the inverse synthesis encoder unit 413 as a single unit, but the inverse synthesis encoder unit 413 is actually composed of N encoders stacked on top of each other.

[0066] The inverse synthesis encoder unit 413 identifies information of interest from the input sequence of the target product 210 via a self-attention sub-hierarchy. The identified information is transmitted to a feedforward sub-hierarchy. The feedforward sub-hierarchy includes a feedforward neural network, which outputs a transformed sequence of the input sequence. The transformed sequence is provided to the inverse synthesis decoder unit 414 as the encoder output of the inverse synthesis encoder unit 413.

[0067] Similar to the reverse synthesis encoder unit 413, for the sake of explanation, the reverse synthesis decoder unit 414 is shown as a single unit in Figure 4, but the reverse synthesis decoder unit 414 is actually composed of N decoders stacked on top of each other.

[0068] The inverse synthesis decoder unit 414 includes a self-attention sub-hierarchy, an encoder-decoder attention sub-hierarchy, and a feedforward sub-hierarchy. The encoder-decoder attention sub-hierarchy differs from the self-attention sub-hierarchy in that the query is the decoder vector, and the key and value are the encoder vectors.

[0069] On the other hand, the residual connection and normalization subhierarchies are applied individually to all lower levels, and masking is applied to the self-attention subhierarchy so that the current output position is not used as information about the next output position. In addition, the decoder output is either linearly transformed or the softmax function is applied.

[0070] The retrosynthesis decoder unit 414 outputs an output sequence corresponding to the input sequence of the categorical latent variable 230 and the input sequence of the target product 210 using a beam search procedure. The output sequence is converted into a string format by the word generator unit 415 and output. As a result, the retrosynthesis prediction model 220 outputs a string format combination 240 of candidate reactants corresponding to the input of the target product 210. The categorical latent variable 230 is also shared with the reaction prediction model 250 and used when calculating the predicted product.

[0071] The processor 110 predicts the predicted product 260 for each of the candidate reactant combinations 240 by driving a reaction prediction model 250 based on information about the candidate reactant combinations 240 and a categorical latent variable 230.

[0072] The reaction prediction model 250 includes a word embedding unit 411, a position encoding unit 412, a reaction prediction encoder unit 413, a reaction prediction decoder unit 416, and a word generator unit 415. As will be described later, the inverse synthesis prediction model 220 and the reaction prediction model 250 share an encoder unit, so the inverse synthesis encoder unit 413 and the reaction prediction encoder unit 413 are named the shared encoder unit 413. In Figure 4, the word embedding unit 411, the position encoding unit 412, the reaction prediction encoder unit 413, the reaction prediction decoder unit 416, and the word generator unit 415 are shown as any one of the parts included in the reaction prediction model 250, but the word embedding unit 411, the position encoding unit 412, the reaction prediction encoder unit 413, the reaction prediction decoder unit 416, and the word generator unit 415 represent the layers included in the reaction prediction model 250.

[0073] The calculation methods for the reaction prediction model 250 and the retrosynthesis prediction model 220 are similar to each other, except for the types of input sequences and output sequences. In other words, the word embedding unit 411 takes the candidate reactant combinations 240 and the categorical latent variables 230 as input data and embeds the input data character by character. The third position encoding unit 412c combines position information with the embedded input data and provides it to the reaction prediction encoder unit 413, and the fourth position encoding unit 412d position encodes the categorical latent variables 230 as initial tokens and provides them to the reaction prediction decoder unit 416.

[0074] The reaction prediction encoder unit 413 identifies information that needs attention from among the input sequences of candidate reactant combinations 240 via the self-attention sub-hierarchy and transmits it to the feedforward sub-hierarchy, which then outputs a transformed sequence using the feedforward neural network.

[0075] The reaction prediction decoder unit 416 includes a self-attention sub-hierarchy, an encoder-decoder attention sub-hierarchy, and a feedforward sub-hierarchy. Furthermore, the residual connection sub-hierarchy and the normalization sub-hierarchy are applied individually to all lower-level sub-hierarchies, and masking is applied to the self-attention sub-hierarchy so that the current output position is not used as information regarding the next output position. Additionally, the decoder output is either linearly transformed or subjected to a softmax function.

[0076] The reaction prediction decoder unit 416 outputs an output sequence corresponding to the input sequence of the categorical latent variable 230 and the input sequence of the candidate reactant combination 240 using a beam search procedure. The output sequence is converted into a string format by the word generator unit 415 and output. As a result, the reaction prediction model 250 outputs a predicted product 261 in string format corresponding to the input of the candidate reactant combination 240.

[0077] The processor 110 compares the target product 211 in string format with the predicted product 261 in string format and determines the experimental priority of the candidate reactant combination 240 based on the comparison result. The processor 110 determines the experimental priority of the candidate reactant combination 240 based on whether the target product 211 in string format and the predicted product 261 in string format match, and outputs the reordered candidate reactant combination 270 based on the determined experimental priority.

[0078] On the other hand, in Figure 4, since the source language and target language are identical in string format (e.g., SMILES), the retrosynthesis prediction model 220 and the reaction prediction model 250 share the word embedding unit 411, the word generator unit 415, and the encoder unit. Furthermore, since the retrosynthesis prediction model 220 and the reaction prediction model 250 share parameters such as the categorical latent variable 230, model complexity is reduced and model normalization is facilitated.

[0079] Figure 5 is a diagram illustrating the calculations performed in the retrosynthesis prediction model and reaction prediction model according to one embodiment.

[0080] The difference from Figure 4 is that the reaction scheme is provided as an input value for a categorical latent variable. In Figures 5 and 4, the same reference numerals are used for the same components, and redundant explanations for the same components are omitted.

[0081] Referring to Figure 5, the neural network device 100 further includes a reaction type prediction model (type classifier) ​​417. In Figure 5, the reaction type prediction model 417 is shown as part of the retrosynthesis prediction model 220, but in one embodiment, the reaction type prediction model 417 is a separate configuration distinct from the retrosynthesis prediction model 220.

[0082] The reaction method prediction model 417 predicts the reaction method based on the encoded information about the target product 210. In one embodiment, the reaction method prediction model 417 obtains at least one of the physical property information, phenotypic information, and structural information of the target product 210 from the encoded representation of the target product 210. The reaction method prediction model 417 also predicts the reaction method 510 for producing the target product 210 based on at least one of the physical property information, phenotypic information, and structural information of the target product 210.

[0083] The processor 110 receives one of the predicted response modes 510 as an expected response mode 511. The expected response mode 511 is input to the processor 110 via user input through the user interface 130. The expected response mode 511 is used as an input value within the categorical latent variable 230.

[0084] The processor 110 decodes the encoded target product 210 and expected reaction scheme 511, and outputs a combination of candidate reactants 240 corresponding to the expected reaction scheme 511. For example, if the expected reaction scheme 511 is the Suzuki-Miyaura scheme, the processor 110 outputs a combination of candidate reactants 240 that produce the target product 210 using the Suzuki-Miyaura scheme.

[0085] By providing the expected reaction scheme 511 as an input value for the categorical latent variable 230, the optimal combination of candidate reactants 240 can be output even in environments where reagent use is restricted.

[0086] Figure 6 is a diagram illustrating the calculations performed in the retrosynthesis prediction model and reaction prediction model according to one embodiment.

[0087] The difference from Figure 5 is that the yield under experimental conditions is utilized when learning the categorical latent variable 230. Similarly, in Figure 6, the same reference numerals are used for the same components as in Figure 5, and redundant explanations for the same components are omitted.

[0088] Referring to Figure 6, the neural network device 100 further includes a yield prediction model 610. In Figure 6, the yield prediction model 610 is shown as being distinct from the retrosynthesis prediction model 220, but in one embodiment, the yield prediction model 610 is a part of the retrosynthesis prediction model 220.

[0089] The yield prediction model 610 is implemented using a transformer model. The yield prediction model 610 is trained on a test dataset and outputs a predicted yield of the target product 210 under experimental conditions.

[0090] The yield prediction model 610 includes a word embedding unit 411, a fifth position encoding unit 412e, a yield prediction encoder unit 413, a yield prediction unit 611, and a reaction condition prediction unit 612. The yield prediction model 610 shares the word embedding unit 411 with the retrosynthesis prediction model 220 and the reaction prediction model 250. The yield prediction model 610 also shares the encoder unit with the retrosynthesis prediction model 220 and the reaction prediction model 250. The yield prediction unit 611 and the reaction condition prediction unit 612 perform the same functions as the decoder units included in the retrosynthesis prediction model 220 and the reaction prediction model 250.

[0091] The experimental condition prediction unit 612 predicts experimental conditions for 270 rearranged candidate reactant combinations for producing the target product 210. The yield prediction unit 611 also predicts the yield of the target product 210 when the experiment is conducted under the experimental conditions predicted by the experimental condition prediction unit 612.

[0092] The predicted yield of the target product 210 predicted by the yield prediction unit 611 is used when learning the categorical latent variable 230. The processor 110 outputs the predicted yield of the target product 210 via a predetermined display means. This allows the user to select the optimal combination of candidate reactants 240, taking the predicted yield into consideration.

[0093] Figure 7 is a flowchart illustrating the operation method of a neural network device according to one embodiment.

[0094] Referring to Figure 7, in step S710, the processor 110 uses a pre-trained retrosynthesis prediction model 220 to predict candidate reactant combinations 240 for generating the target product 210.

[0095] When information about the target product 210 is input to the processor 110, it predicts a categorical latent variable 230 containing multiple classes based on the information about the target product 210. The processor 110 also autoregressively predicts candidate reactant combinations 240 based on the categorical latent variable 230.

[0096] If x is the information about the target product 210, z is the categorical latent variable 230, and y is the combination of candidate reactants 240, then the categorical latent variable 230 and the combination of candidate reactants 240 based on the input of the target product 210 are predicted by the likelihood P(z,y|x). The processor 110 predicts the categorical latent variable 230 and the combination of candidate reactants 240 based on the likelihood P(z,y|x) value.

[0097] In step S720, the processor 110 uses a pre-trained reaction prediction model 250 to predict the predicted product 260 for each of the candidate reactant combinations 240.

[0098] The processor 110 obtains information about the predicted product 260 based on the categorical latent variable 230 and the combination of candidate reactants 240.

[0099] In step S730, the processor 110 determines the experimental priority of candidate reactant combinations 240 based on the comparison results between the target product 210 and the predicted product 260.

[0100] The processor 110 determines the experimental priority of candidate reactant combinations 240 based on whether the target product 210 and the predicted product 260 match.

[0101] Predicted product 260 If the filename is TIFF0007851699000001.tif9128, the probability of agreement between the target product 210 and the predicted product 260, given the input of the categorical latent variable 230 and the combination of candidate reactants 240, is the likelihood. Determined by TIFF0007851699000002.tif10128. Processor 110 determines likelihood The experimental priority of the 240 candidate reactant combinations is determined based on the value of TIFF0007851699000003.tif11128, in descending order.

[0102] Figure 8 is a flowchart illustrating a learning method for a retrosynthesis prediction model and a reaction prediction model according to one embodiment.

[0103] Referring to Figure 8, in step S810, the processor 110 receives input for the combination of test reactants and the test product corresponding to each of the combinations of test reactants.

[0104] In one embodiment, the processor 110 receives input including the predicted yield of the test product under experimental conditions and the reaction method for each combination of test reactants.

[0105] In step S820, the processor 110 learns a categorical latent variable 230 containing multiple classes based on the combination of test reactants and the test product.

[0106] In one embodiment, the processor 110 can also learn a categorical latent variable 230 based on the predicted yield provided by a previously trained yield prediction model.

[0107] The processor 110 learns the conditional probability distribution of the categorical latent variables 230 for each input of the test product.

[0108] In one embodiment, information regarding the test product is x ts Therefore, information about the test reactants is y ts If so, the processor 110 calculates the likelihood P(z,y) of the inverse composite prediction model 220 as shown in equation 1 below. ts |x ts ) and the likelihood P(x) of the reaction prediction model 250 ts |y tsThe categorical latent variable 230 is set such that the cross-entropy loss of ,z) is minimized.

[0109]

number

[0110] For example, processor 110 learns a categorical latent variable 230 using an expectation-maximization algorithm (hard EM).

[0111] When the current model parameter is θ, the processor 110 iteratively performs the following steps: first, estimating a categorical latent variable 230 that minimizes the cross-entropy loss L, as shown in Equation 2 below; and second, updating the parameter in response to the input variable, as shown in Equation 3 below. By iteratively updating the model, the processor 110 derives the optimal categorical latent variable 230.

[0112]

number

[0113]

number

[0114] On the other hand, in equations 2 and 3, L h This means that processor 110 used a hard expectation maximization algorithm, and in equation 3, θ' represents the updated parameter.

[0115] Figure 9 is a flowchart illustrating the operation method of a retrosynthetic prediction model according to one embodiment.

[0116] Referring to FIG. 9, at step S910, the processor 110 inputs information regarding the target product 210.

[0117] The processor 110 inputs the chemical structure of the target product 210 in string format. For example, the processor 110 inputs the chemical structure of the target product 210 in the form of a SMILES code, a SMARTS code, or an InChi code.

[0118] At step S920, the processor 110 predicts a categorical latent variable 230 including a plurality of classes based on the information regarding the target product 210.

[0119] The processor 110 drives the retrosynthesis prediction model 220 to predict the categorical latent variable 230 as the first token. According to one embodiment, the expected reaction mode of the target product 210 is provided as an input value of the categorical latent variable 230. When the expected reaction mode 511 is provided as an input value of the categorical latent variable 230, an optimal combination of candidate reactants 240 is output even in an environment where reagent use is restricted.

[0120] At step S930, the processor 110 obtains information regarding the combination of candidate reactants 240 based on the information regarding the target product 210 and the categorical latent variable 230.

[0121] The processor 110 automatically and recursively predicts tokens regarding the combination of candidate reactants 240 based on the information regarding the target product 210 and the predicted categorical latent variable 230. The likelihoods of the categorical latent variable 230 and the combination of candidate reactants 240 due to the input of the target product 210 are determined by Equation 4 below.

[0122]

Equation

[0123] In Equation 4, T is the length of the token sequence, and y <t is y tThis refers to the target token preceding it.

[0124] The processor 110 outputs a combination of candidate reactants 240 that maximizes both the class-specific likelihood for the input of the target product 210 and the likelihood of the retrosynthesis prediction result for the inputs of the target product 210 and the categorical latent variable 230.

[0125] In one embodiment, the processor 110 selects a predetermined number of final candidate reactant combinations based on class likelihoods and the likelihood of retrosynthesis prediction results. For example, the processor 110 selects a predetermined number of sets of combinations of categorical latent variables 230 and candidate reactants 240 in descending order of likelihood based on the input of the target product 210. The sets include pairs of categorical latent variables 230 and final candidate reactant combinations 240. These pairs of categorical latent variables 230 and final candidate reactant combinations 240 are input into the reaction prediction model 250 and used to calculate the predicted product 260.

[0126] Figure 10 is a flowchart illustrating the operation method of a reaction prediction model according to one embodiment.

[0127] Referring to Figure 10, in step S1010, the processor 110 receives information about the candidate reactant combinations 240.

[0128] The processor 110 receives the structures of the candidate reactant combinations 240 as input in string format. In this case, the string format of the candidate reactant combinations 240 is the same as the string format of the target product 210. For example, if the string format of the target product 210 is a SMILES code, then the string format of the candidate reactant combinations 240 is also a SMILES code.

[0129] In step S1020, the processor 110 receives a categorical latent variable 230 that includes multiple classes.

[0130] In other words, the processor 110 is input a pair of categorical latent variables 230 and a combination of final candidate reactants 240.

[0131] In step S1030, the processor 110 obtains information about the predicted product for each of the candidate reactant combinations 240 and the categorical latent variable 230.

[0132] The processor 110 generates tokens for the prediction product 260 based on the likelihood P(x|z,y). The prediction product 260 is also used to validate the inverse synthesis prediction model 220.

[0133] Figure 11 is a flowchart illustrating a method for determining the experimental priority of candidate reactant combinations according to one embodiment.

[0134] Referring to Figure 11, in step S1110, the processor 110 determines whether the predicted product 260 for each input of the candidate reactant combination 240 matches the target product 210.

[0135] The processor 110 determines whether the target product 210 and the predicted product 260 match based on the input of a categorical latent variable 230 and a combination of candidate reactants 240, using the likelihood calculation. Determined by TIFF0007851699000008.tif13128.

[0136] In step S1120, the processor 110 determines the experimental priority of candidate reactant combinations 240 based on whether the predicted product 260 matches the target product 210.

[0137] Processor 110 has a likelihood The experimental priority of the 240 candidate reactant combinations is determined based on the value of TIFF0007851699000009.tif13128, in descending order.

[0138] The above-described embodiments can be created using programs executed on a computer and are implemented by a general-purpose digital computer that runs the program using a computer-readable recording medium. Furthermore, the data structures used in the above-described embodiments are recorded on a computer-readable recording medium by various means. Computer-readable recording media include magnetic recording media (e.g., ROM (read-only memory), floppy disks, hard disks, etc.) and optically readable media (e.g., CD-ROM (compact disc read-only memory), DVD (digital versatile disc), etc.).

[0139] Although embodiments of the present invention have been described in detail above with reference to the drawings, the present invention is not limited to the embodiments described above, and can be modified and implemented in various ways without departing from the technical spirit of the present invention. [Explanation of symbols]

[0140] 100 Neural Network Devices 110 processors 120 memory 130 User Interface 210, 211 target product 220 Inverse Synthesis Predictive Models 230 Categorical Latent Variables 240 combinations of candidate reactants 250 Reaction Prediction Models 260, 261 Predicted Products 270, 280 rearranged candidate reactant combinations 411 Word Embedding Department 412 Position Encoding Section 413 Inverse synthesis encoder unit, reaction prediction encoder unit, yield prediction encoder unit 414 Retrosynthesis Decoder Unit, Reaction Prediction Decoder Unit 415 Word Generator Section 416 Reaction prediction decoder unit 417 Reaction Scheme Prediction Model 510 Reaction Method 511 Expected Response Method 610 Yield Prediction Model 611 Yield prediction unit 612 Experimental Condition Prediction Unit

Claims

1. A method for determining the experimental priority of candidate reactant combinations for synthesizing a target product using a neural network, The steps include: predicting candidate reactant combinations for synthesizing the target product using a previously trained retrosynthesis prediction model based on categorical latent variables; A step of using a previously trained reaction prediction model to input the combination of candidate reactants predicted by the previously trained retrosynthesis prediction model, and the categorical latent variable shared with the previously trained retrosynthesis prediction model, and predicting the predicted product for each of the combinations of candidate reactants; A method characterized by comprising the step of determining the experimental priority of candidate reactant combinations in descending order of likelihood, based on the results of comparing the target product with the predicted product.

2. When using the aforementioned previously trained retrosynthesis prediction model and the aforementioned previously trained reaction prediction model, A step of inputting a combination of test reactants and a test product corresponding to each of the combinations of test reactants, The process includes a step of learning a categorical latent variable that includes multiple classes based on the combination of test reactants and the test product, The method according to claim 1, wherein the categorical latent variable is a variable that affects information regarding the combination of candidate reactants, and indicates the type of reaction, including the reaction form, experimental conditions, and reaction method, and is generated in correspondence with the class.

3. The method according to claim 2, characterized in that the step of learning the categorical latent variable is to learn the conditional probability distribution of the categorical latent variable with respect to each input of the test product.

4. The method according to claim 2, characterized in that the step of learning the categorical latent variable is to learn the categorical latent variable based on the predicted yield provided by a previously trained yield prediction model.

5. The method according to 2, characterized in that the learned categorical latent variable is used to predict the combination of candidate reactants and the predicted product.

6. The step of predicting the combination of candidate reactants is: The step of inputting information about the target product, A step of predicting the categorical latent variable, which includes multiple classes, based on the information regarding the target product; The step includes obtaining information on the combination of candidate reactants based on the information on the target product and the categorical latent variable, The method according to claim 1, characterized in that the information regarding the categorical latent variable and the combination of candidate reactants is predicted and obtained by the value of likelihood P(z, y | x) when the information regarding the target product is x, the categorical latent variable is z, and the combination of candidate reactants is y.

7. The step of obtaining information regarding the combination of candidate reactants is: A step of calculating the class-specific likelihood of each combination of candidate reactants based on the input of the target product, A step of calculating the likelihood of the retrosynthesis prediction result with respect to the input of the target product and the categorical latent variable, The method according to 6, comprising the step of selecting a predetermined number of combinations of final candidate reactants based on the class-specific likelihoods and the likelihood of the retrosynthesis prediction results.

8. The method according to 6, characterized in that the step of predicting the categorical latent variable is provided with an expected reaction scheme for producing the target product as an input value for the categorical latent variable.

9. The aforementioned prediction step of the predicted product is: The step of inputting information regarding the combination of candidate reactants, The steps include inputting the categorical latent variable which includes multiple classes, The step includes obtaining information about the predicted product for each of the candidate reactant combinations based on the information about the candidate reactant combinations and the categorical latent variable, The information regarding the predicted product is If the information relating to the target product is x, the categorical latent variable is z, and the combination of candidate reactants is y, then the likelihood The method according to claim 1, characterized in that it is obtained by the value of 1.

10. The step of determining the experimental priority of the candidate reactant combinations is: A step of determining whether the predicted product and the target product match based on each input of the candidate reactant combinations, The method according to claim 1, comprising the step of determining the experimental priority of candidate reactant combinations based on whether the predicted product matches the target product.

11. A device that uses a neural network to determine the experimental priority of candidate reactant combinations for synthesizing a target product, Memory in which at least one program is stored, A processor that executes at least one of the aforementioned programs, The aforementioned processor, Based on categorical latent variables, a previously trained retrosynthesis prediction model is used to predict candidate reactant combinations for synthesizing the target product. Using a previously trained reaction prediction model, the combinations of candidate reactants predicted by the previously trained retrosynthesis prediction model, and the categorical latent variables shared with the previously trained retrosynthesis prediction model are input, and the predicted products for each of the combinations of candidate reactants are predicted. The apparatus is characterized by determining the experimental priority of candidate reactant combinations in descending order of likelihood, based on the comparison results between the target product and the predicted product.

12. The previously trained retrosynthesis prediction model and the previously trained reaction prediction model learn the categorical latent variable, which includes multiple classes, based on the combination of test reactants and the test product corresponding to each of the combinations of test reactants. The apparatus according to claim 11, wherein the categorical latent variable is a variable that affects information regarding the combination of candidate reactants, and indicates the type of reaction including the reaction form, experimental conditions, and reaction method, and is generated in correspondence with the class.

13. The apparatus according to claim 12, characterized in that the previously trained retrosynthesis predictive model and the previously trained reaction predictive model learn the conditional probability distribution of the categorical latent variable with respect to each input of the test product.

14. The apparatus according to claim 12, characterized in that the previously trained retrosynthesis prediction model and the previously trained reaction prediction model learn the categorical latent variables based on the predicted yields provided by the previously trained yield prediction model.

15. The apparatus according to claim 12, characterized in that the learned categorical latent variables are used to predict the combination of candidate reactants and the predicted products.

16. The aforementioned processor, Information regarding the target product is entered, Based on the information regarding the target product, the categorical latent variable, which includes multiple classes, is predicted. Based on the information regarding the target product and the categorical latent variables, information regarding the combination of candidate reactants is obtained. The apparatus according to claim 11, characterized in that the information regarding the categorical latent variable and the combination of candidate reactants is predicted and obtained by the value of likelihood P(z, y | x) when the information regarding the target product is x, the categorical latent variable is z, and the combination of candidate reactants is y.

17. The aforementioned processor, The class-specific likelihood of each combination of candidate reactants based on the input of the target product is calculated. The likelihood of the retrosynthesis prediction result for the input of the target product and the categorical latent variable is calculated. The apparatus according to claim 16, characterized in that it selects a predetermined number of combinations of final candidate reactants based on the class-specific likelihood and the likelihood of the retrosynthesis prediction result.

18. The apparatus according to claim 16, characterized in that the expected reaction method for producing the target product is provided as an input value for the categorical latent variable.

19. The aforementioned processor, Input information regarding the combination of candidate reactants, Input the categorical latent variable containing multiple classes, Based on the information regarding the combination of candidate reactants and the categorical latent variable, information regarding the predicted product for each of the combinations of candidate reactants is obtained. The information regarding the predicted product is If the information relating to the target product is x, the categorical latent variable is z, and the combination of candidate reactants is y, then the likelihood The apparatus according to claim 11, characterized in that it is obtained by the value of .

20. The aforementioned processor, Determine whether the predicted product and the target product match for each input of the candidate reactant combination. The apparatus according to claim 11, characterized in that the experimental priority of the candidate reactant combinations is determined based on whether or not the predicted product matches the target product.

Images