Information providing device, information providing method, and information providing program

Abstract

A feature extraction unit (131) according to an embodiment of the present invention extracts a plurality of features by inputting a plurality of data sets into a model that outputs from a data set a feature in a dimension lower than the data set. A similarity degree calculation unit (132) calculates the degree of similarity between the plurality of features extracted by the feature extraction unit (131). The model is a model trained by multi-task learning that uses the original data set in transfer learning and that is a combination of self-supervised learning and data set classification learning.

Description

[Technical Field]

[0001] The present invention relates to an information providing device, an information providing method, and an information providing program. [Background technology]

[0002] Deep neural networks (DNNs) are capable of highly accurate predictions in image processing and natural language processing. However, training DNNs is costly.

[0003] For example, the costs involved in training a DNN include the cost of collecting a dataset including the assignment of correct labels (annotations), the computational cost of improving accuracy, and the tuning cost of exploring multiple hyperparameters for each case.

[0004] Transfer learning has been proposed as a way to reduce these costs when introducing DNNs into business.

[0005] Transfer learning is a technique that uses a dataset (source dataset) or a trained model that is different from the target dataset, and performs learning using less data or less computation time.

[0006] Transfer learning also includes techniques such as fine-tuning and domain adaptation.

[0007] Fine-tuning is a method in which a model is pre-trained on a source dataset and the trained parameters are used as initial values ​​for training the target dataset.

[0008] Domain adaptation is a method of simultaneously training the source and target datasets using the same model, and solving the tasks of the target dataset using knowledge from the source dataset. [Prior art documents] [Non-patent literature]

[0009] [Non-Patent Document 1] Kornblith, Simon, Jonathon Shlens, and Quoc V. Le. "Do Better ImageNet Models Transfer Better?" Proceedings of the IEEE / CVF Conference on Computer Vision and Pattern Recognition. 2019. Summary of the Invention [Problem to be solved by the invention]

[0010] However, conventional techniques have the problem that they cannot always perform transfer learning efficiently. Conventional transfer learning relies heavily on the intuition and experience of developers, and manual work is required for selecting the source dataset and tuning parameters.

[0011] The dataset that is effective for transfer learning is not obvious, and the results of transfer learning vary greatly depending on the relationship (similarity) between the target dataset and the source dataset. For example, a model pre-trained on ImageNet (large-scale, large-volume) may be inferior to a model that has not been pre-trained, depending on the target dataset (see, for example, Non-Patent Document 1).

[0012] On the other hand, the degree of similarity between datasets is generally unknown, and no de facto standard metric has yet emerged.

[0013] In addition, in transfer learning, it is necessary to select hyperparameters that match the target dataset and the source dataset. On the other hand, deep learning models have many hyperparameters, and tuning is also essential during transfer learning. [Means for solving the problem]

[0014] In order to solve the above-mentioned problems and achieve the objective, an information provision device includes a feature extraction unit that extracts multiple features from a dataset by inputting multiple datasets into a model that outputs features of a lower dimension than the dataset, and a similarity calculation unit that calculates the similarity between the multiple features extracted by the feature extraction unit, and is characterized in that the model has been trained by multi-task learning of self-supervised learning and dataset classification learning using a source dataset in transfer learning. [Effects of the Invention]

[0015] According to the present invention, transfer learning can be performed efficiently. [Brief explanation of the drawings]

[0016] [Figure 1] FIG. 1 is a diagram illustrating an example of the configuration of an information providing device according to the first embodiment. [Figure 2] FIG. 2 is a diagram illustrating a method for measuring the similarity. [Figure 3] FIG. 3 is a diagram illustrating a model learning method. [Figure 4] FIG. 4 is a diagram illustrating the information providing process. [Figure 5] FIG. 5 is a flowchart showing the flow of the learning process. [Figure 6] FIG. 6 is a flowchart showing the flow of the similarity measurement process. [Figure 7] FIG. 7 is a flowchart showing the flow of the information provision process. [Figure 8-1] FIG. 8-1 shows the results of the experiment. [Figure 8-2] FIG. 8-2 shows the results of the experiment. [Figure 9-1] FIG. 9-1 shows the results of the experiment. [Figure 9-2] FIG. 9-2 shows the results of the experiment. [Figure 10-1] FIG. 10-1 is a diagram showing the evaluation results. [Figure 10-2]FIG. 10-2 is a diagram showing the evaluation results. [Figure 11] FIG. 11 is a diagram illustrating an example of a computer that executes an information providing program. DETAILED DESCRIPTION OF THE INVENTION

[0017] DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS An information providing apparatus, an information providing method, and an information providing program according to the present invention will be described in detail below with reference to the accompanying drawings. However, the present invention is not limited to the following embodiments.

[0018] [Configuration of the first embodiment] FIG. 1 is a diagram illustrating an example of the configuration of an information providing device according to a first embodiment. The information providing device 10 calculates similarities between datasets and provides information based on the calculated similarities. For example, the information providing device 10 provides information for identifying a source dataset similar to a target dataset in transfer learning.

[0019] Furthermore, the information providing device 10 performs a learning process of a model for calculating the similarity. The information providing device 10 may calculate the similarity using a trained model, or may provide the trained model to another device or the like.

[0020] As shown in FIG. 1, the information providing device 10 includes an input / output unit 11, a storage unit 12, and a control unit 13.

[0021] The input / output unit 11 is an interface for inputting and outputting data. For example, the input / output unit 11 may be a communication interface such as a network interface card (NIC) for performing data communication with other devices via a network. The input / output unit 11 may also be an interface for connecting an input device such as a mouse or keyboard and an output device such as a display.

[0022] The storage unit 12 is a storage device such as an HDD (Hard Disk Drive), an SSD (Solid State Drive), or an optical disk. The storage unit 12 may be a data-rewritable semiconductor memory such as a RAM (Random Access Memory), a flash memory, or an NVSRAM (Non-Volatile Static Random Access Memory). The storage unit 12 stores an OS (Operating System) and various programs executed by the information providing device 10. The storage unit 12 also stores model information 121.

[0023] The model information 121 is information such as parameters for constructing a model, and is updated as appropriate during the learning process. The updated model information 121 may be output to another device or the like via the input / output unit 11.

[0024] The control unit 13 controls the entire information providing device 10. The control unit 13 is, for example, an electronic circuit such as a central processing unit (CPU), a micro processing unit (MPU), or a graphics processing unit (GPU), or an integrated circuit such as an application specific integrated circuit (ASIC) or a field programmable gate array (FPGA). The control unit 13 also has an internal memory for storing programs defining various processing procedures and control data, and executes each process using the internal memory. The control unit 13 also functions as various processing units by running various programs. For example, the control unit 13 has a feature extraction unit 131, a similarity calculation unit 132, a loss function calculation unit 133, an update unit 134, a candidate extraction unit 135, and a provision unit 136.

[0025] The feature extraction unit 131 extracts a plurality of feature amounts from a data set by inputting the plurality of data sets into a model that outputs feature amounts of a lower dimension than the data sets.

[0026] The similarity calculation unit 132 calculates the similarity between the plurality of feature amounts extracted by the feature extraction unit 131 .

[0027] A method for measuring the similarity by the feature extraction unit 131 and the similarity calculation unit 132 will be described with reference to Fig. 2. Fig. 2 is a diagram for explaining the method for measuring the similarity.

[0028] 2, the feature extraction unit 131 uses a model F to extract features from the data set A and the data set B. The model F is a deep neural network for feature extraction.

[0029] The data set includes a plurality of data samples, and the feature extraction unit 131 extracts a feature amount for each data sample.

[0030] In the example in Figure 2, data set A contains I data samples x A i (where i is an integer between 0 and I). The feature extraction unit 131 then extracts feature quantities f A i Extract.

[0031] Here, since the datasets used in DNNs are high-dimensional, it is difficult to directly measure the similarity between datasets. Therefore, the feature extraction unit 131 extracts features by reducing the dimension of the dataset as shown in FIG. 2. For example, A i is x A i It is lower dimensional than

[0032] Furthermore, the feature extraction unit 131 aggregates the extracted feature amounts. In the example of FIG. 2, the feature extraction unit 131 aggregates the feature amounts f A i is one feature f´ A It is summarized as follows.

[0033] In this way, the feature extraction unit 131 can aggregate the feature amounts output by the model for each data sample included in the data set into the feature amount of one data sample. For example, the feature extraction unit 131 can use statistics such as the mean and variance of each element of multiple data samples as the aggregated feature amount.

[0034] Then, the similarity calculation unit 132 calculates the similarity between the feature amounts aggregated by the feature extraction unit 131. For example, the aggregated feature amount f' A and f´ B If f and f are vectors, the similarity calculation unit 132 calculates the vector f′ A and vector f´ B Distance d AB The similarity calculation unit 132 may calculate the 2-Wasserstein distance as the distance between the vectors.

[0035] The feature extraction unit 131 extracts features from a model trained by self-supervised learning using a source dataset in transfer learning. The similarity calculation unit 132 calculates the similarity between the features of the source dataset and the features of the target dataset in transfer learning.

[0036] The loss function calculation unit 133 calculates a loss function for model training.

[0037] The learning method of model F will be explained using Figure 3. Figure 3 is a diagram for explaining the model learning method. x shown in Figure 3 is input data. D is a set of data sets. d is a data set. Model F is a feature extractor (DNN).

[0038] Model F is used to measure the similarity between target data and multiple source data when identifying source data similar to target data in transfer learning.

[0039] In this case, the information providing device 10 is assumed to have previously trained the model F with an arbitrary task such as classification using a group of source datasets. In the embodiment, the information providing device 10 adds a dataset classification task (dataset classification learning) to the pre-learning task (for example, self-supervised learning) and performs multi-task learning on the model F. The model F is a model that has been trained by multi-task learning of self-supervised learning and dataset classification learning using the source dataset in transfer learning.

[0040] Furthermore, the information providing device 10 uses MoCo (He, Kaiming, et al. "Momentum contrast for unsupervised visual representation learning," Proceedings of the IEEE / CVF Conference on Computer Vision and Pattern Recognition, 2020), a contrastive learning method, as a self-supervised learning method. Self-supervised learning does not require annotation, making it easy to handle multiple datasets together.

[0041] The dataset classification task is a task to assign a label to each dataset and classify it. The dataset classification task does not require annotation because automatic labeling is possible using only the dataset. By adding the dataset classification task, Model F can explicitly learn the features of each dataset.

[0042] As shown in FIG. 3, the loss function calculation unit 133 calculates the loss function (Self-supervised Loss) L of the pre-learning task shown in Equation (1) based on the feature values ​​obtained by inputting input data x of multiple data sets d to a model F. SSL and the loss function for the dataset classification task (Dataset Classification Loss) L shown in Equation (2). DC Calculate:

[0043]

number

[0044]

number

[0045] The q on the right side of equation (1) is the output of the model F obtained from the query image. + is the output of model F obtained from the correct key obtained by augmenting the same image as the query. K is the total number of key images including the correct key. τ is the temperature coefficient. CE in equation (2) is the cross entropy. Here, c d is a dataset classifier (fully connected neural network) that takes the feature value F(x) of the input x as input. d indicates the label assigned to the dataset d of the input data x.

[0046] Here, Self-supervised Loss is the loss for the task of generating a query and a correct key image from an input image using two different image transformations, and then finding the correct combination of the query and correct key from a set of keys obtained from other images via a DNN.

[0047] In addition, Dataset Classification Loss is the function of the feature F(x) obtained by inputting input data x into model F and the label y assigned to the dataset d of the input data x. d It is used to learn the relationship between

[0048] The loss function calculation unit 133 calculates L, which is the sum of the loss of the pre-training task and the loss of the dataset classification task, using equation (3). ALL Calculate.

[0049]

number

[0050] In equation (3), α is a hyperparameter.

[0051] The update unit 134 calculates the loss function L ALL Update the parameters of model F so that is optimized.

[0052] The parameters of model F are stored in the storage unit 12 as model information 121. The update unit 134 updates the model information 121.

[0053] The candidate extraction unit 135 and the providing unit 136 support actual transfer learning by providing information that identifies source data similar to target data. The information providing process will be described with reference to Fig. 4. Fig. 4 is a diagram for explaining the information providing process.

[0054] Dataset D in Figure 4 T is the target dataset. Also, the datasets D0 to D N are multiple source datasets.

[0055] The similarity calculator corresponds to the feature extraction unit 131 and the similarity calculation unit 132 that use the trained model F.

[0056] At this time, the similarity calculation unit 132 calculates one target data set (D T ) and multiple source datasets (D0 to D T ) and calculate the similarity between the features.

[0057] Then, based on the calculated similarity, the candidate extraction unit 135 extracts, as candidates, source datasets whose feature similarity with the target dataset is at or above a predetermined rank.

[0058] Furthermore, the providing unit 136 provides the user with information for identifying the source datasets extracted as candidates from among the source datasets.

[0059] In the example of FIG. 4, the candidate extraction unit 135 calculates the similarity d 0T , d 1T , …, d NT Create a ranking in order of similarity. For example, the top three similarities d NT , d 1T , d 0T The source dataset D corresponds to N , D1, D0 are extracted.

[0060] The providing unit 136 provides the extracted source data set D N , D1, D0 are the corresponding hyperparameters H N , H1, and H0 are provided to the user. Note that the optimal hyperparameters for each source dataset are assumed to have already been determined using a method such as grid search during the previous model construction process.

[0061] In this way, multiple combinations of source data and hyperparameters provided together may be used together in transfer learning.

[0062] [Processing of the first embodiment] The flow of processing by the information providing device 10 will be described with reference to the flowcharts shown in FIGS.

[0063] 5 is a flowchart showing the flow of the learning process. As shown in FIG. 5, first, the information providing device 10 reads learning data from the source dataset group (step S101).

[0064] Next, the information providing device 10 extracts features from the training data using the DNN model F (step S102).

[0065] Here, the information providing device 10 calculates the loss function L of the pre-learning task in the feature space. SSL (Step S103). The information providing device 10 calculates the loss function L of the dataset classification task in the feature space. DC is calculated (step S104).

[0066] The information providing device 10 calculates the sum L of the loss of the pre-training task and the loss of the dataset classification task. ALL is calculated (step S105).

[0067] Then, the information providing device 10 updates the parameters of the model F by backpropagation of the loss function (step S106).

[0068] At this time, if the maximum number of learning steps is greater than the number of learning steps (step S107, True), the information providing device 10 returns to step S101 and repeats the process. On the other hand, if the maximum number of learning steps is not greater than the number of learning steps (step S107, False), the information providing device 10 ends the process.

[0069] 6 is a flowchart showing the flow of the similarity measurement process. As shown in FIG. 6, first, the information providing device 10 reads a data sample from the source dataset (step S201).

[0070] Next, the information providing device 10 extracts features from the source data sample using the DNN model F (step S202). Furthermore, the information providing device 10 aggregates the feature vectors for each source data sample into a single feature vector (for example, a mean or variance) (step S203).

[0071] The information providing device 10 reads a data sample from the target data set (step S204).

[0072] Then, the information providing device 10 extracts features of the target data sample using the DNN model F (step S205). Furthermore, the information providing device 10 aggregates the feature vectors for each target data sample into a single feature vector, similar to the source data set (step S206).

[0073] The information providing device 10 calculates the similarity between the feature vectors of the aggregated target dataset and the transition source dataset using, for example, the 2-Wasserstein distance (step S207).

[0074] 7 is a flowchart showing the flow of the information providing process. First, the information providing device 10 calculates the similarity between the target dataset and N source datasets (step S301).

[0075] Next, the information providing device 10 calculates the data set similarity {d iT} i N The information providing device 10 sorts the source datasets (distance: ascending order / score: descending order) (step S302). Then, the information providing device 10 extracts the Top-K source dataset ids from the rankings obtained by the sorting (K≦N: any integer) (step S303).

[0076] Here, the information providing device 10 reads the datasets and hyperparameters associated with the K source dataset ids (step S304). Then, the information providing device 10 issues a URI (Uniform Resource Identifier) ​​that can be downloaded by the user, and outputs the datasets and hyperparameters (step S305).

[0077] [Advantages of the first embodiment] As described above, the feature extraction unit 131 extracts multiple feature quantities from a data set by inputting multiple data sets into a model that outputs feature quantities of lower dimensions than the data sets. The similarity calculation unit 132 calculates the similarity between the multiple feature quantities extracted by the feature extraction unit 131.

[0078] In this way, the information providing device 10 can automatically calculate the similarity between data sets. As a result, according to this embodiment, similar data sets can be identified, and transfer learning can be performed efficiently.

[0079] The feature extraction unit 131 aggregates the features output by the model for each data sample included in the data set into the features of one data sample. The similarity calculation unit 132 calculates the similarity between the features aggregated by the feature extraction unit 131.

[0080] As a result, according to this embodiment, the distance between feature amounts can be easily calculated.

[0081] The feature extraction unit 131 extracts features from a trained model using a source dataset in transfer learning. The model is trained by multi-task learning, in which a dataset classification task is added to a pre-learning task (e.g., self-supervised learning). The similarity calculation unit 132 calculates the similarity between the features of the source dataset and the features of the target dataset in transfer learning.

[0082] As described above, in this embodiment, multi-task learning, in which a dataset classification task is added to a pre-learning task that does not require annotation (e.g., self-supervised learning), can be used to efficiently train a model that measures similarity. In this embodiment, dataset classification labels are obtained from the dataset itself, eliminating the need for human labeling. Furthermore, in this embodiment, multi-task learning, in which a dataset classification task is added to a pre-learning task, makes it possible to explicitly learn the features of each dataset.

[0083] The similarity calculation unit 132 calculates the similarity between features between one target dataset and each of multiple source datasets. The provision unit 136 provides the user with information for identifying a source dataset whose feature similarity with the target dataset is at or above a predetermined rank among the source datasets.

[0084] This allows the information providing device 10 to recommend a source dataset similar to the target dataset to the user. Therefore, according to this embodiment, transfer learning can be performed efficiently.

[0085] [Evaluation experiment] An evaluation experiment conducted by actually implementing the above embodiment will now be described. In the evaluation experiment, the above embodiment was used to select a source dataset and hyperparameters (architecture) based on the similarity of the datasets.

[0086] The experimental setup is as follows. Dataset Target dataset: Oxford Pets (dog and cat images), Stanford Cars (vehicle images) Source datasets: FGVC-Aircraft (aircraft images), OxfordFlowers (flower images), StanfordDogs (dog images), Indoor67 (interior images), CUB-200-2011 (bird images) Neural network architecture: ResNet-50 Self-supervised learning task: MoCo (see references)

[0087] The evaluation items are the correlation between the source-target dataset similarity and the classification accuracy for the target task when the weights pre-trained on the source dataset are selected.

[0088] (Experiment 1) Figures 8-1 and 8-2 show the results of Experiment 1, in which the source dataset was determined based on the dataset similarity. Figures 8-1 and 8-2 are diagrams showing the results of the experiment.

[0089] In the examples of Figures 8-1 and 8-2, a feature extractor (Model F) was trained using all data in the source dataset. Then, the feature extractor was used to measure the dataset similarity (2-Wasserstein distance (2WD, lower is better)) between the target and source datasets.

[0090] Furthermore, we used the trained model of the source dataset to fine-tune it using Oxford Pets and measure the test accuracy.

[0091] Figures 8-1 and 8-2 visualize the correlation between dataset similarity and test accuracy. Figure 8-2 shows the results of multi-task learning using a dataset classification task added to the pre-training task (MoCo v2 with Dataset Cls.). D For comparison, Figure 8-1 shows the results of the experiment using a model trained only on the pre-training task (self-supervised learning model: MoCo v2). M The experimental results are shown below.

[0092] Model F M When using the model F, the correlation coefficient R is -0.333 (see Figure 8-1). D When using the model F, the correlation coefficient R is -0.797 (see Figure 8-2). M Therefore, Model F D By using this, Model F M It is easier to select a better dataset than with

[0093] (Experiment 2) 9-1 and 9-2 show the results of Experiment 2, in which the source dataset was determined based on the dataset similarity.

[0094] In the examples of Figures 9-1 and 9-2, a feature extractor (Model F) was trained using all data in the source dataset. Then, the feature extractor was used to measure the dataset similarity (2-Wasserstein distance (2WD), lower is better) between the target and source datasets.

[0095] Furthermore, we used the trained model of the source dataset and fine-tuned it with StanfordCars to measure the test accuracy.

[0096] The examples in Figures 9-1 and 9-2 visualize the correlation between dataset similarity and test accuracy. Figure 9-2 shows the correlation between dataset similarity and test accuracy. DFor comparison, Figure 9-1 shows the experimental results for Model D. M The experimental results are shown below.

[0097] Model F M When using the model F, the correlation coefficient R is +0.688 (see Figure 9-1). D When using the model F, the correlation coefficient R is -0.745 (see Figure 9-2). D By using this, Model F M Compared to the case of using the , it improves the inverse correlation and enables accurate selection of good data sets.

[0098] (Rating 1) The following shows the results of visualizing the dimensionality of vectors after feature extraction of five datasets using Model F. Figures 10-1 and 10-2 show the evaluation results.

[0099] Figure 10-1 shows Model F M Figure 10-2 shows the visualization results of the vectors after feature extraction using Model F. D This is the visualization result of the vector after feature extraction using (Dataset Cls. Enabled).

[0100] As shown in Figure 10-1, Model F M In this case, six clusters, C1, C2-1, C2-2, C3, C4, C5, and C6, were formed instead of the five clusters that should have been formed. M However, it is thought that there may be cases where the characteristics of the dataset are not fully captured and feature extraction for each dataset cannot be performed sufficiently.

[0101] In contrast, as shown in Figure 10-2, Model F D In this model, clusters C1, C2, C3, C4, C5, and C6 are formed. D In this model, the number of clusters formed matches the number of datasets. DWhen using the model F, it was confirmed that the features of each dataset could be clearly extracted. D By using, the dataset similarity can be improved significantly.

[0102] Therefore, by performing multi-task learning in which a dataset classification task was added to the pre-learning task in training Model F, we were able to confirm that the features of each dataset could be extracted more clearly.

[0103] Therefore, the information providing device 10 according to the embodiment can accurately estimate the relationship between data sets necessary for predicting a high-performance model. As a result, according to the present embodiment, similar data sets can be appropriately identified, and transfer learning can be efficiently performed.

[0104] [System configuration, etc.] Furthermore, the components of each device shown in the figure are functional concepts and do not necessarily have to be physically configured as shown. In other words, the specific form of distribution and integration of each device is not limited to that shown, and all or part of the devices can be functionally or physically distributed or integrated in any unit depending on various loads, usage conditions, etc. Furthermore, all or any part of the processing functions performed by each device can be realized by a CPU (Central Processing Unit) and a program analyzed and executed by the CPU, or can be realized as hardware using wired logic. Note that the program may be executed not only by the CPU but also by other processors such as a GPU.

[0105] Furthermore, among the processes described in this embodiment, all or part of the processes described as being performed automatically can be performed manually, or all or part of the processes described as being performed manually can be performed automatically using a known method.In addition, the information including the processing procedures, control procedures, specific names, various data and parameters shown in the above documents and drawings can be changed as desired unless otherwise specified.

[0106] [program] In one embodiment, the information providing device 10 can be implemented by installing an information providing program that executes the above-described processes as package software or online software on a desired computer. For example, by executing the above-described information providing program on an information processing device, the information processing device can function as the information providing device 10. The information processing device referred to here includes desktop and notebook personal computers. In addition, the information processing device also includes mobile communication terminals such as smartphones, mobile phones, and PHS (Personal Handyphone Systems), as well as slate terminals such as PDAs (Personal Digital Assistants).

[0107] The information providing device 10 can also be implemented as a server device that provides services related to the above processing to a client terminal device used by a user. For example, the server device is implemented as a server device that receives a target dataset and multiple source datasets as input and outputs the similarity between the target dataset and each source dataset. In this case, the server device may be implemented as a web server or as a cloud that provides services related to the above processing by outsourcing.

[0108] 11 is a diagram showing an example of a computer that executes an information provision program. The computer 1000 includes, for example, a memory 1010 and a CPU 1020. The computer 1000 also includes a hard disk drive interface 1030, a disk drive interface 1040, a serial port interface 1050, a video adapter 1060, and a network interface 1070. These components are connected by a bus 1080.

[0109] The memory 1010 includes a ROM (Read Only Memory) 1011 and a RAM (Random Access Memory) 1012. The ROM 1011 stores a boot program such as a BIOS (Basic Input Output System). The hard disk drive interface 1030 is connected to a hard disk drive 1090. The disk drive interface 1040 is connected to a disk drive 1100. A removable storage medium such as a magnetic disk or optical disk is inserted into the disk drive 1100. The serial port interface 1050 is connected to a mouse 1110 and a keyboard 1120, for example. The video adapter 1060 is connected to a display 1130, for example.

[0110] The hard disk drive 1090 stores, for example, an OS 1091, an application program 1092, a program module 1093, and program data 1094. That is, a program that defines each process of the information providing device 10 is implemented as a program module 1093 in which computer-executable code is written. The program module 1093 is stored, for example, in the hard disk drive 1090. For example, a program module 1093 for executing processes similar to those of the functional configuration of the information providing device 10 is stored in the hard disk drive 1090. The hard disk drive 1090 may be replaced with an SSD (Solid State Drive).

[0111] Furthermore, setting data used in the processing of the above-described embodiment is stored as program data 1094, for example, in the memory 1010 or the hard disk drive 1090. Then, the CPU 1020 reads the program module 1093 or the program data 1094 stored in the memory 1010 or the hard disk drive 1090 into the RAM 1012 as necessary, and executes the processing of the above-described embodiment.

[0112] The program module 1093 and program data 1094 are not limited to being stored in the hard disk drive 1090, but may also be stored in, for example, a removable storage medium and read by the CPU 1020 via the disk drive 1100 or the like. Alternatively, the program module 1093 and program data 1094 may be stored in another computer connected via a network (such as a local area network (LAN) or a wide area network (WAN)). The program module 1093 and program data 1094 may then be read by the CPU 1020 from the other computer via the network interface 1070. [Explanation of symbols]

[0113] 10 Information provision device 11 Input / output section 12 Storage section 121 Model Information 13 Control Unit 131 Feature Extraction Unit 132 Similarity calculation part 133 Loss function calculation unit 134 Update Department 135 Candidate Extraction Unit 136 Provision Department

Claims

1. a feature extraction unit that extracts a plurality of feature quantities from a dataset by inputting a plurality of datasets into a model that outputs feature quantities of a lower dimension than the dataset; a similarity calculation unit that calculates similarities between a plurality of feature amounts extracted by the feature extraction unit; and The information providing device is characterized in that the model is a model that has been trained through multitask learning of self-supervised learning and dataset classification learning using a source dataset in transfer learning.

2. the feature extraction unit aggregates the feature amounts output by the model for each data sample included in the data set into a feature amount for one data sample; 2. The information providing device according to claim 1, wherein the similarity calculation unit calculates the similarity between the feature amounts collected by the feature extraction unit.

3. The information providing device according to claim 1 , wherein the similarity calculation unit calculates a similarity between a feature amount of the source dataset and a feature amount of a target dataset in the transfer learning.

4. a providing unit for providing information to a user; the similarity calculation unit calculates similarities between features for one target dataset and each of a plurality of source datasets; The information providing device according to claim 1, characterized in that the providing unit provides a user with information for identifying a source dataset from among the source datasets whose feature similarity with the target dataset is at or above a predetermined rank.

5. An information providing method executed by an information providing device, comprising: a feature extraction step of extracting a plurality of feature quantities from a dataset by inputting a plurality of datasets into a model that outputs feature quantities of a lower dimension than the dataset; a similarity calculation step of calculating similarities between the plurality of feature amounts extracted by the feature extraction step; Including, An information provision method characterized in that the model is a model that has been trained through multi-task learning of self-supervised learning and dataset classification learning using a source dataset in transfer learning.

6. An information providing program for causing a computer to function as the information providing device according to any one of claims 1 to 4.

Images