Abstract: Provided is a model interpretation method, an image processing method, an electronic device and a storage medium, relating to the field of artificial intelligence, in particular to the field of deep learning. The model interpretation method includes: obtaining a token vector corresponding to an image feature input to a first model; obtaining a model prediction result output by the first model; and determining, according to a combination of an attention weight and a gradient, an association relation between the token vector input to the first model and the model prediction result output by the first model, where the association relation is used to characterize interpretability of the first model.

Abstract: The present application provides a model training, image processing method, device, storage medium, and program product relating to deep learning technology, which are able to screen auxiliary image data with image data for learning a target task, and further fuse the target image data and the auxiliary image data, so as to train a built and to-be-trained model with the fusion-processed fused image data. This implementation can increase the amount of data for training the model, and the data for training the model is determined is based on the target image data, which is suitable for learning the target task. Therefore, the solution provided by the present application can train an accurate target model even if the amount of target image data is not sufficient.

Abstract: A method, apparatus, device and readable medium for transfer learning in machine learning are provided. The method includes: constructing a target model according to the number of classes to be achieved by a target task and a duly-trained source model; obtaining a value of a regularized loss function of the corresponding target model and a value of a cross-entropy loss function of the target model, based on sets of training data in a training dataset of the target task; according to the value of the regularized loss function and the value of the cross-entropy loss function corresponding to each set of training data, updating parameters in the target model by a gradient descent method to implement the training of the target model. The above technical solution avoids excessive constraints on parameters in the prior art, thereby refraining from damaging the training effect of the source model on the target task.

Abstract: A method for processing synthetic features is provided, and includes: the synthetic features to be evaluated and original features corresponding to the synthetic features are obtained. A feature extraction is performed on the synthetic features to be evaluated based on a number S of pre-trained samples, to obtain meta features with S samples. S is a positive integer. The meta features are input into the pre-trained meta feature evaluation model for a binary classification prediction, to obtain a probability of binary classification. Quality screening is performed on the synthetic features to be evaluated according to the probability of the binary classification, to obtain second synthetic features to be evaluated. The second synthetic features are classified in a good category. The second synthetic features and original features are input into a first classifier for evaluation. classified in a poor category.

Abstract: A method for training a decision-making model parameter, a decision determination method, an electronic device, and a non-transitory computer-readable storage medium are provided. In the method, a perturbation parameter is generated according to a meta-parameter, and first observation information of a primary training environment is acquired based on the perturbation parameter. According to the first observation information, an evaluation parameter of the perturbation parameter is determined. According to the perturbation parameter and the evaluation parameter thereof, an updated meta-parameter is generated. The updated meta-parameter is determined as a target meta-parameter, when it is determined, according to the meta-parameter and the updated meta-parameter, that a condition for stopping primary training is met.

Abstract: A query method is provided and includes: acquiring association records, in which the association record is configured to indicate an execution area, execution time and user attribute data of an execution user, of a behavior; splitting the association record into behavior records based on attribute items included in the user attribute data of the association record, in which the behavior record is configured to indicate a mapping relationship between at least one of the attribute items and the execution area-the execution time; grouping the behavior records to determine behavior statistics information of each group; in which behavior records having the same attribute item, the same execution area and the same execution time belong to the same group; and displaying behavior statistics information of a target group in response to a query operation.

Abstract: A technical solution relates to a big data technology in the field of artificial intelligence technologies. The technical solution includes: acquiring training data including annotation results of a risk grade of each sample region and a risk grade of a district to which each sample region belongs; and training an initial model including an encoder, a discriminator and a classifier using the training data, and obtaining the risk prediction model using the encoder and the classifier after the training process. The encoder performs a coding operation using region features of the sample regions to obtains a feature representation of each sample region; the discriminator identifies the risk grade of the district to which the sample region belongs according to the feature representation of the sample region; the classifier identifies the risk grade of the sample region according to the feature representation of the sample region.

Abstract: A method and an apparatus for training a classification model and data classification includes: obtaining a sample set and a pre-trained classification model, wherein the classification model includes at least two convolutional layers, each convolutional layer is connected to a classification layer through a fully connected layer; inputting the sample set into the classification model, and obtaining a prediction result output by each classification layer, wherein the prediction result includes a prediction probability of a class to which each sample belongs; calculating a probability threshold of each classification layer based on the prediction result output by each classification layer; setting a prediction stopping condition for the classification mode according to the probability threshold of each classification layer.

Abstract: The disclosure provides a multi-task deployment method, and an electronic device. The method includes: obtaining N first tasks and K network models, in which N and K are positive integers greater than or equal to 1; allocating the N first tasks to the K network models differently for operation, to obtain at least one candidate combination of tasks and network models, in which each candidate combination includes a mapping relation between the N first tasks and the K network models; selecting a target combination with a maximum combination operation accuracy from the at least one candidate combination; and deploying a target mapping relation comprised in the target combination and the K network models on a prediction machine.

Abstract: The present disclosure discloses an information processing method and device, and relates to the field of artificial intelligence, in particular, to a graph neural network in the field of deep learning. A specific implementation solution according to an embodiment includes determining initial representation of edges connected between a plurality of atoms in a molecule based on three-dimensional structure information of the molecule; determining first representation of a neighbor edge of each of the atoms based on the initial representation of the edges, the neighbor edge of each of the atoms indicating at least one edge connected with each of the atoms; determining first representation of each of the atoms based on the first representation of the neighbor edge of each of the atoms; determining feature representation for characterizing the molecule based on the first representation of each of the atoms.

Abstract: A method of data interaction, a data interaction apparatus, an electronic device and a non-transitory computer readable storage medium are provided, related to field of computer technologies, and in particular to the field of artificial intelligence technologies. When the method of data interaction is applied to the first data platform, the method includes: sending a data request to a second data platform based on a first resource server provided by a cloud or based on a local server; acquiring response data fed back by the second data platform based on a second resource server provided by the cloud; at least one of the first resource server and the second resource server is dynamically created by the cloud.

Abstract: A method for determining correlation between a drug and a target, and an electronic device are provided. The method includes: establishing a spatial molecular graph of a candidate drug and the target, the spatial molecular graph including an atomic node set and an edge set, the atomic node set including atoms in the candidate drug and atoms in the target, the edge set including at least one atom connection edge; inputting a first atom feature of the atomic node set and the spatial molecular graph into a first GAT for prediction, to obtain a second atom feature of the atomic node set; and determining a parameter value of the correlation between the candidate drug and the target in accordance with the second atom feature of the atomic node set.

Abstract: A method of training a prediction model for determining molecular binding force is provided, which relates to the field of artificial intelligence, in particular to a graph neural network in the field of deep learning. The method includes: constructing a virtual complex molecule based on a three-dimensional structure information of a first molecule and a second molecule; determining a predicted binding force and a predicted interaction matrix between the first molecule and the second molecule based on the virtual complex molecule by using the prediction model, the predicted interaction matrix indicating an element-type-based and distance-based interaction between an atom in the first molecule and an atom in the second molecule; and training the prediction model by minimizing a target loss function based on a difference between the predicted binding force and a real binding force and a difference between the predicted interaction matrix and a real interaction matrix.

Abstract: Methods for monitoring an economic state and establishing an economic state monitoring model and corresponding apparatuses, and relates to the technical field of big data are disclosed. A specific implementation solution is: acquiring, from map application data, geographic location point active data in a to-be-monitored future time frame and N historical time frames before the to-be-monitored time frame respectively for a to-be-monitored region, the N being a positive integer; and inputting feature vectors of the geographic location point active data in the to-be-monitored time frame and the N historical time frames before the to-be-monitored time frame into a pre-trained economic state monitoring model, to obtain economic indicator data of the to-be-monitored region in the to-be-monitored time frame. An economic state of the to-be-monitored region in the to-be-monitored time frame in real time can be monitored, thus timely providing a reference for policy making.

Abstract: Methods for predicting an economic state and establishing an economic state prediction model and corresponding apparatuses, and relates to the technical field of big data are disclosed. A specific implementation solution is: acquiring, from map application data, geographic location point active data in N historical time frames before a to-be-predicted future time frame respectively for a to-be-predicted region, the N being a positive integer; and inputting feature vectors of the geographic location point active data in the N historical time frames before the to-be-predicted time frame into a pre-trained economic state prediction model, to obtain economic indicator data of the to-be-predicted region in the to-be-predicted time frame. An economic state of the to-be-predicted region in the to-be-predicted time frame can be predicted, thus providing a reference for policy making in advance.

Abstract: The present application provides a model training, image processing method, device, storage medium, and program product relating to deep learning technology, which are able to screen auxiliary image data with image data for learning a target task, and further fuse the target image data and the auxiliary image data, so as to train a built and to-be-trained model with the fusion-processed fused image data. This implementation can increase the amount of data for training the model, and the data for training the model is determined is based on the target image data, which is suitable for learning the target task. Therefore, the solution provided by the present application can train an accurate target model even if the amount of target image data is not sufficient.

Abstract: Embodiments of the present disclosure provide a data mining system, a data mining method, and a storage medium. The data mining system includes a transfer device, a first trusted execution space and a second trusted execution space. The transfer device is configured to receive a data calling request of the second trusted execution space, obtain data to be called from the first trusted execution space according to the data calling request, and provide the data to be called to the second trusted execution space, so as to perform data mining based on the data to be called and the mining-related data to obtain a data mining result and to provide the data mining result to a device of the data user.

Abstract: A method, apparatus, device and readable medium for transfer learning in machine learning are provided. The method includes: constructing a target model according to the number of classes to be achieved by a target task and a duly-trained source model; obtaining a value of a regularized loss function of the corresponding target model and a value of a cross-entropy loss function of the target model, based on sets of training data in a training dataset of the target task; according to the value of the regularized loss function and the value of the cross-entropy loss function corresponding to each set of training data, updating parameters in the target model by a gradient descent method to implement the training of the target model. The above technical solution avoids excessive constraints on parameters in the prior art, thereby refraining from damaging the training effect of the source model on the target task.