10 results about "Loss function" patented technology

This improves the accuracy and speed of solving large-scale combinatorial optimization problems by dividing them into subproblems. [Solution] The information processing device 100 for processing combinatorial optimization problems includes: a graph creation unit 112 that creates one or more subgraphs from a main graph; a mathematical optimization unit 115 that solves the combinatorial optimization problem for each subgraph using a mathematical optimization solver; a machine learning unit 117 that trains each subGNN so that the output of the subGNN corresponding to each subgraph is close to the solution of the mathematical optimization solver; a feature vector assignment unit 118 that assigns the feature vectors at each vertex of the subGNN obtained as a result of training to the corresponding vertices of the main GNN as input to the feature vectors of the main GNN corresponding to the graph data of the main graph; and a solution output unit 119 that outputs the solution obtained as a result of training the main GNN by setting a loss function so that the machine learning unit 117 solves the combinatorial optimization problem for the main graph.

This application discloses a retrieval model training method, a retrieval-enhanced generation-based Bert-Hogg-Dubbs syndrome (BHD) identification method and system, relating to the field of rare disease identification. The training method includes acquiring positive and negative sample pairs; inputting the positive and negative sample pairs into an initial retrieval model; calculating the loss function of the initial retrieval model; and dynamically adjusting the angle margin of the loss function in real time according to a metric variance adaptive mechanism. The initial retrieval model is then optimized based on the loss function calculation results. This application forcibly expands the angle interval between BHD and non-BHD by using the angle margin of the loss function, and uses a metric variance adaptive mechanism to dynamically adjust the angle margin based on the statistical variance of the cosine similarity among all positive sample pairs in the current training batch. This solves the problem of weak image differences and blurred category decision boundaries in DCLDs caused by the highly similar imaging features of various rare diseases, thus improving the recognition accuracy of large models for query information.

The present application relates to a kind of knowledge distillation method based on inference step deconstruction and differentiating supervision, belong to artificial intelligence field.The structured thinking chain generated by acquiring teacher model is separated according to line, each inference step is deconstructed into independent unit;Each step is type labeled, according to the five-level classification system of basic calculation, basic fact, operation execution, logical reasoning, strategy planning, differentiating weight is distributed, and weight superposition is used to composite step;Weighted loss function is constructed, step weight is introduced into cross-entropy loss, and student model is distilled training.The present application focuses on key inference link in the learning process by step-level type perception and differentiating supervision, overcomes the problem of detail loss and global understanding damage caused by flat sequence supervision in traditional distillation.Experimental results show that, on mathematical reasoning task, compared with equal-weight step distillation method, accuracy is improved by 4.6 percentage points, and efficient migration of large model reasoning ability to lightweight model is realized.

The invention discloses an optimization method for solving a mixed integer linear programming problem, and the method comprises the steps: employing a bipartite graph structure composed of variable nodes and constraint nodes for the modeling of the MILP problem, and the bipartite graph structure comprises variable node features, constraint node features and edge features; variables of the bipartite graph structure are divided into stable variables and unstable variables, a corresponding total loss function is determined, and the total loss function is determined based on a cross entropy loss function and a comparison loss function; carrying out label learning on the stable variables by adopting a cross entropy loss function, and carrying out comparative learning on the unstable variables by adopting a comparative loss function; based on the total loss function, training learning of the corresponding relation of the MILP problem and the solution is carried out to obtain a corresponding prediction model, and based on the prediction model, the MILP problem is solved. According to the method, prediction precision and solution feasibility can be considered in the MILP problem solving process, and good stability and practicability are shown in various types of actual problems.

The invention relates to the technical field of computer-aided solution of physical fields or mathematical equations, and discloses a physical scene multi-region cooperative computing method based on an adaptive guaranteed structure. The method comprises the following steps: firstly, dividing a global solution domain into a plurality of mutually connected and partially overlapped sub-domains through domain decomposition, and configuring an independent physical information neural network model for each sub-domain; then circularly executing the following steps: performing multi-dimensional feature extraction in each sub-domain based on the composite feature index; constructing a probability function according to the characteristic indexes, and implementing adaptive sampling to dynamically update a training set; after each round of training is finished, dynamically optimizing the weight of the loss function according to the change trend of different loss items, and adding a physical structure retention constraint item into the total loss function; and circularly executing until a convergence condition is met, and outputting a physical field numerical solution. According to the method, precise capture of the local high-gradient region of the physical field is realized, and the stability and physical conservation of the model are ensured while the calculation precision and efficiency are improved.

The application belongs to the technical field of numerical control machine tools, and discloses a mathematical model parameter identification method for a machine tool feeding system based on gradient optimization. The method comprises the following steps: S1, establishing a mathematical model for the machine tool feeding system; S2, collecting actual operation data of the machine tool; setting initial parameters to be identified; constructing an update model for the parameters to be identified; S3, inputting a preset instruction position signal and the current parameters to be identified into the update model for the parameters to be identified to update the parameters to be identified, simulating and calculating a loss function by using the updated identification parameters, and judging the relationship between the current loss function and an optimal loss function; if the current loss function is smaller than the optimal loss function, the value of the current parameters to be identified is kept, otherwise, the number of times that the loss function increases is increased; S4, updating the number of iterations, and judging whether the current number of iterations reaches a preset maximum number of iterations; if yes, the current parameters to be identified are output; otherwise, the step S3 is returned. By the application, the identification efficiency of parameters with low sensitivity to the global loss function is improved.

Embodiments of the present application provide a training method and an inference method of a flow matching generation model and related devices, to improve the accuracy of the action sequence predicted by the trained flow matching generation model. The method of the embodiments of the present application comprises: obtaining noise, time t and environment features of a first action, wherein the environment features of the first action at least include observation values of the first action; inputting the noise, time t and environment features into an initialized flow matching generation model to obtain an output predicted velocity field vector of a conditional probability path of time t+1; calculating a loss between the predicted velocity field vector and a real velocity field vector by using a preset loss function, wherein the preset loss function includes at least one of a first loss function and a second loss function, and a third loss function; and training the initialized flow matching generation model by using the loss and a back propagation algorithm until the flow matching generation model converges, to obtain a trained flow matching generation model.

An information processing apparatus 100 for processing a combinatorial optimization problem includes: a graph creation unit 112 configured to create one or more subgraphs from a main graph; a mathematical optimization unit 115 configured to solve a combinatorial optimization problem for each of the subgraphs by a mathematical optimization solver; a machine learning unit 117 configured to train a sub-GNN corresponding to each of the subgraphs such that an output of the sub-GNN is approximate to a solution of the mathematical optimization solver; a feature vector assignment unit 118 configured to assign a feature vector at each vertex of the sub-GNN obtained as a result of the training to each corresponding vertex of a main GNN corresponding to graph data of the main graph as an input of a feature vector of the main GNN; and a solution output unit 119 configured to output a solution obtained as a result of the machine learning unit 117 training the main GNN by setting a loss function to solve the combinatorial optimization problem for the main graph.

The invention discloses a feature inversion method for continuous low-thrust orbital maneuver of a space target, and belongs to the technical field of space situation awareness. The method comprises the following steps: firstly, establishing a spacecraft motion dynamics model and a target angle measurement observation model of a tracker; secondly, constructing decision variables for low-thrust maneuvering thrust inversion, including a maneuvering starting time point, a maneuvering ending time point and a maneuvering acceleration vector, and assuming that the maneuvering acceleration vector of the continuous low-thrust maneuvering target is kept unchanged in a maneuvering time period; a distance residual error loss function and an angle measurement residual error loss function are constructed based on the hypothesis, then a low-thrust maneuvering thrust characteristic inversion problem is modeled into a multi-objective optimization problem, a non-dominated sorting multi-population genetic algorithm NSGA-II with an elitist strategy is used for optimizing and solving the multi-objective optimization problem, an optimal decision variable is obtained, and the low-thrust maneuvering thrust characteristic inversion problem is solved. The maneuvering starting time point, the maneuvering ending time point and the maneuvering acceleration vector of the low-thrust maneuvering target are obtained. According to the method, the maneuvering starting time point, the maneuvering ending time point and the maneuvering acceleration vector of the low-thrust maneuvering target are solved only by using the filtering data and the measurement data.

One embodiment provides a method and system which facilitates optimizing a pair of affine classifiers based on a diversity metric. During operation, the system defines a diversity metric based on an angle between decision boundaries of a pair of affine classifiers. The system includes the diversity metric as a regularization term in a loss function optimization for designing the pair of affine classifiers, wherein the designed pair of affine classifiers are mutually orthogonal. The system predicts an outcome for a testing data object based on the designed pair of mutually orthogonal affine classifiers.