19results about How to "Efficient learning process" patented technology

The invention discloses a dynamic cross-modal Hash retrieval method and system based on concept vector learning. The method comprises the following steps: constructing a concept vector matrix based on current label information and a Hadamard matrix; concept vectors in the concept vector matrix are used for representing that the samples with the same label have the same and unchanged vectors; constructing a data similarity matrix based on the current label information and the historical label information; obtaining a target hash code based on the concept vector matrix and the data similarity matrix; obtaining a target hash function based on the target hash code; analyzing the to-be-queried data according to the target hash function to obtain a to-be-queried hash code corresponding to the to-be-queried data; and performing similarity calculation on the to-be-queried hash code and a target hash code in a database to obtain a target retrieval result. According to the method, the defect of the concept drift problem in the dynamic data environment can be overcome, the adaptability of the cross-modal Hash model to the concept drift is improved, the dynamic cross-modal Hash retrieval performance is improved, and the method can be widely applied to the technical field of information processing.

The invention discloses an efficient progressive strategy enhancement (EPPE) framework based on deep reinforcement learning (DRL) and used for a path planning scene, and the framework assists a DRL algorithm and is used for solving dual challenges of training efficiency and combinatorial optimization faced by DRL in path planning. The framework combines the advantages of sparse rewards and process rewards, the sparse rewards help the intelligent agent to realize a global optimal strategy, and the process rewards provide real-time feedback, so that the intelligent agent can adjust the strategy in time. The strategy learning efficiency is remarkably improved, and the reward coupling problem caused by process rewards is effectively solved, so that the global optimization of the strategy is realized. The method can be applied to the research field of robot path planning, and plays an important role in the application fields of automatic driving, logistics distribution and the like.

The invention relates to the field of artificial intelligence, and provides an information recommendation method and device, equipment and a storage medium. The method comprises the steps that a recommendation model conducts recommendation estimation on multiple pieces of candidate recommendation information obtained through retrieval, target recommendation information recommended to a target object is determined based on corresponding information recommendation estimation values, and the recommendation model is obtained by adopting a loop iteration mode and based on first training samples of classification targets with the positive and negative sample number greatly different. In each round of iterative training, each first training sample is divided into a common sample or a difficult sample, a first weight value with a relatively low weight value is given to a target weight of the common sample, a second weight value with a relatively high weight value is given to a target weight of the difficult sample, and the second weight value is in positive correlation with the learning difficulty of the corresponding difficult sample; based on the first recommendation estimated value of each training sample and the target weight, model parameters are adjusted, and a model capable of correctly distinguishing positive and negative sample interfaces is adopted, so that the recommendation accuracy is improved.

The invention belongs to the technical field of aerospace, and particularly discloses a spacecraft cluster game method and system based on generative adversarial reinforcement learning, and the method comprises the steps: building a relative motion dynamics model of pursuit and escape parties, and building a game relation model of the pursuit and escape parties based on a two-person zero sum game model; solving a Nash equilibrium solution of the pursuit game by using a differential game method based on the game relationship model of pursuit and escape parties, and generating an expert database; designing and generating an adversarial reinforcement learning algorithm based on the expert database; and solving the game relationship between the pursuing and escaping parties by using a generative adversarial reinforcement learning algorithm to obtain an optimal game strategy. According to the method, the optimal control method and the artificial intelligence method are combined to solve the approaching strategy of the spacecraft cluster virtual main satellite to the non-cooperative target, the problems that convergence is difficult, reward setting is difficult and the effect is poor when the artificial intelligence method is used for solving the orbit game problem are solved, and finally efficient and accurate approaching to the target is achieved.

The invention discloses a semi-supervised adaptive unstructured scene segmentation method, and relates to the technical field of computer vision. The method comprises the following steps: acquiring label-free image data of an unstructured scene; inputting the weakly enhanced image into a teacher model to generate a first prediction result; inputting the strong enhancement image into the student model to generate a second prediction result; calculating the JS divergence between the first prediction result and the second prediction result, and dynamically adjusting the attenuation coefficient of the index moving average according to the JS divergence; updating teacher model parameters according to the adjusted index moving average attenuation coefficient to obtain an unstructured scene segmentation model; and inputting a to-be-predicted unstructured scene image into the trained unstructured scene segmentation model, and outputting an unstructured scene segmentation result. According to the method, the teacher model weight updating rate is reduced in the model divergence significant stage such as the initial training stage or the difficult sample processing stage so as to suppress noise propagation, and the prediction precision of unstructured scene segmentation is effectively improved.

The invention discloses a human body orientation and action form identification method, system and device and a medium, and relates to the technical field of human body orientation and action form identification, and the method comprises the steps: obtaining analog signals outputted by a sensor under different conditions; converting the preprocessed analog signals into waveform signals, and analyzing each waveform signal in a time domain and a frequency domain to obtain first waveform data corresponding to each waveform signal; training the preset initial model until the preset initial model reaches a preset training ending condition; acquiring second waveform data corresponding to the to-be-detected signal, and inputting the second waveform data into the identification model for processing to obtain a target distance and a target form of the to-be-detected signal; by means of the mode, while action form and distance recognition is achieved, the cost and power consumption of the system are greatly reduced, and the problems that an existing video image form recognition mode is high in cost, complex in algorithm, difficult to maintain in the later period and the like are effectively solved.

The application discloses a LIBS element quantitative analysis method based on double-branch feature fusion and electronic equipment, the method comprises the following steps: preprocessing the collected LIBS spectrum signal, including baseline correction and spectrum resampling; inputting the spectrum signal data after preprocessing into a double-branch feature extraction network, extracting local features through a CNN branch, and extracting global features through an MLP branch in parallel; performing adaptive weighted fusion on the local features and the global features through a gating fusion module to obtain fusion features; inputting the fusion features into a WMA-MLP model, the WMA-MLP model is based on MLP, integrates a multi-head self-attention mechanism and a residual module, is used for modeling the global dependency relationship between features, and outputs a final element quantitative analysis result. Through the parallelly arranged CNN and MLP branch feature extraction structures, more comprehensive spectrum feature representation can be obtained, and the accuracy and robustness of the spectrum analysis model are improved.

The invention discloses a public security event identification method based on small sample learning and a knowledge graph, which comprises the following steps: generating a virtual scene similar to a real event scene through the combination of small sample data and the knowledge graph, and dynamically activating nodes related to the event through a knowledge activation threshold value, freezing and storing the unrelated nodes, and dynamically releasing frozen nodes; identifying dynamic evolution of the event by monitoring node relation fluctuation in the knowledge graph; a multidimensional event analysis result is generated through mutual reinforcement of a graph fission mechanism and an interactive training mechanism; generating a risk path identification result through the event aggregation trigger point; the scheme has the characteristics of realizing efficient learning of small samples, enhancing semantic understanding ability, dynamically adapting to diversified scenes, improving event identification flexibility and constructing a global risk assessment framework.

The application discloses a kind of brain-like reinforcement learning method and system based on layered experience playback, it is related to reinforcement learning and brain-like computing technical field.The method comprises the following steps: S1, collecting observation data and pre-processing;S2, initialize experience buffer pool and actor network, critic network and corresponding target network, parameter initialization;S3, initialize exploration noise and select action from actor network according to current state and execute, store the obtained experience sample to experience buffer pool;S4, obtain new sample from experience buffer pool, carry out short-term memory pool update;S5, use attention discrimination module to determine whether part of experience in short-term memory experience pool is transferred to long-term memory experience pool;S6, the parameters of actor network, critic network and corresponding target network are updated.The application uses the above method, improves the experience utilization rate of intelligent agent, improves the performance of reinforcement learning, and has wide application potential in many fields.

The present application relates to a kind of small sample disease classification methods based on association learning, comprising: first, build a small sample asphalt disease data set containing multiple asphalt disease types;Second, using the principle of association learning, develop a set of deep learning model and its training strategy suitable for small sample disease classification;Then, by setting "precision-speed" comprehensive evaluation index and a variety of damage simulation mode, evaluate the comprehensive performance and robustness of the model;Finally, the model is further adjusted and optimized, and deployed to road disease inspection equipment, realize the accurate classification of road disease.Compared with the prior art, the present application aims to solve the problem that traditional deep learning method excessively depends on training data size, alleviate the deficiency of existing method under small sample condition, improve the classification accuracy and system robustness, and provide a new solution for road disease detection and classification.

The invention discloses a Double DQN-based rare association rule mining method, which belongs to the technical field of data mining, and comprises the following steps of: obtaining a data set of data to be mined, and preprocessing the data set; constructing a rare association rule mining model based on a Double DQN algorithm, and taking the preprocessed data set as an environment of the rare association rule mining model; generating a state variable according to environment initialization; calculating a state change value according to the state variable; the item with the highest state change value is selected to be changed to obtain a new state value, and an association rule is generated; a reward value is generated based on the association rule. According to the method, the Double DQN is applied to mining of the rare association rule, so that the mining effect and efficiency of the high rare association rule are improved.

This invention proposes a robotic arm control method and apparatus. The method includes: constructing a simulated environment for robotic arm control; acquiring environmental state information, including the pose and velocity of the robotic arm and the position of the object to be manipulated; inputting the environmental state information as input to a symbolic network, the output of which is the robotic arm motion value, including the joint velocities of the robotic arm; selecting a suitable path from the symbolic network to generate a symbolic strategy; and deploying the robotic arm control task according to the symbolic strategy. This method can improve the efficiency of symbolic strategy learning, thereby learning robotic arm control with less interactive data and improving the accuracy of robotic arm control.

The invention relates to the technical field of smart homes, and discloses a defrosting detection method for an air conditioner, which comprises the following steps: acquiring real-time operation condition data of the air conditioner in a heating mode; different types of feature parameters in the real-time operation condition data are subjected to differential preprocessing, feature data after differential preprocessing are obtained, and differential preprocessing comprises feature explicit and feature implicit; inputting the preprocessed feature data into a neural network model to obtain a frosting degree prediction result; and defrosting control is conducted according to the frosting degree prediction result. Thus, the dynamic mode of the frosting state can be learned more accurately and efficiently, the detection precision, robustness and generalization ability under complex and changeable actual working conditions are remarkably improved, and real self-adaptive and high-reliability intelligent defrosting control is achieved. The invention further discloses a defrosting detection device for the air conditioner, the air conditioner and a computer readable storage medium.

The present invention relates to the field of model construction technology, and in particular to a method for constructing a transaction decision model for logistics data analysis. The method comprises the following steps: obtaining comprehensive logistics data; performing data preprocessing on the comprehensive logistics data to generate standard comprehensive logistics data; performing cargo value assessment on the standard comprehensive logistics data to generate cargo value assessment data; performing initial transportation mode analysis on the logistics transportation cargo information data based on the cargo value assessment data to generate initial logistics transportation mode data; performing first logistics transaction fee confirmation on the initial logistics transportation mode data based on the cargo value assessment data to obtain first logistics transportation transaction data; performing intelligent perception of logistics transportation cargo using sensors to obtain logistics cargo intelligent perception data. The present invention improves the accuracy and comprehensiveness of transaction decision model construction through standardized data processing, dynamic transportation mode selection, reverse logistics cost analysis and machine learning decision making.

The invention relates to a brain signal basic processing system and processing method for a multi-mode electroneurographic signal, and the system comprises an electroneurographic signal abnormity detection module which is configured to output a qualified multi-mode electroneurographic signal; the intracerebral and extracerebral unified spatial position coding module is configured to output a multi-modal electroneurographic signal with position codes; the dynamic brain region spatial convolution module is configured to perform adaptive spatial weighting on the multi-modal neuroelectric signals and convert the heterogeneous multi-channel neuroelectric signals into isomorphic single-channel feature sequences; and the composite self-supervised learning module is configured to execute self-supervised training combining mask prediction and autoregression prediction on the isomorphic single-channel feature sequence, and output a brain signal basic model after pre-training is completed, so as to complete downstream task evaluation based on the brain signal basic model after pre-training is completed. According to the method, the performance of multi-dimensional downstream tasks in the brain-computer interface field is improved, and the requirement of the downstream tasks for the data scale is reduced.

This invention discloses a multi-UAV task allocation and fusion method based on improved reinforcement learning, comprising: determining an initial parameter set for multi-UAV task allocation; fragmenting and decomposing multi-task objectives to determine cluster centers; using an adaptive clustering strategy based on center points to screen candidate clustering strategies and determine the optimal cluster set; determining a dynamic decay gradient strategy for the improved Q-learning network based on the action value function of basic reinforcement learning; determining a spiral cross-assignment mechanism for task redistribution based on a task redistribution mechanism with weighted coefficients; and determining a multi-UAV task allocation strategy based on the optimal cluster set of task points, the dynamic decay gradient strategy of the improved Q-learning network, and the task redistribution mechanism to complete the multi-UAV task allocation. This invention can effectively avoid unnecessary redundant allocation, improve the time consumption and range of task allocation, and increase the efficiency of task allocation.

The invention discloses a robot teaching method, device and equipment based on a virtual demonstrator and a medium. The method is applied to a virtual demonstrator system, and comprises the following steps: receiving a voice instruction input by an operator and a selected to-be-trained robot; converting the voice instruction into a text instruction, and generating an initial code instruction from the text instruction through a preset large model; and correcting the initial code instruction to determine a target code instruction, and sending the target code instruction to the to-be-trained robot through a virtual demonstrator. By implementing the method, the problem of low teaching training efficiency of the industrial robot in the prior art can be solved.

The application relates to a dynamic gradient compression learning method of a federal learning system and belongs to the field of artificial intelligence. The method comprises the following steps: constructing a federal learning model, a server distributing different sampling communication bandwidths to clients, and the clients obtaining local models after training; considering resource constraints and gradient states of each client local model, calculating a pruning rate and a quantization level of a dynamic gradient compression strategy in this round; according to the pruning rate, each client performs a sparse operation on the gradient of the client to obtain a sparse gradient; according to the quantization level, each client performs a quantization and encoding operation on the sparse gradient to obtain compressed gradient data; each client uploads the compressed gradient data to the server, the server decompresses the gradient data to obtain decompressed gradient data; and the server aggregates the decompressed gradient data according to elements and updates a global model until the model converges. The application realizes efficient communication federal learning with better performance in a resource-limited scene.

The embodiments of the present invention disclose a method and module for autonomously constructing complex products based on human-machine hybrid enhancement, the method comprising: S10, establishing a parameterized knowledge model mapping from key feature states to actions based on human knowledge, encoding the parameters to be measured in the knowledge model into an output vector of a neural network; S20, encoding the input vector of the neural network according to the key feature states of the environment and the output value of the reward function; S40, intelligent agent configuration; S50, generating training data mapping from key feature states to actions based on human-known empirical knowledge, using the training data to reversely fit the neural network parameters, and driving the knowledge model to deduce in the training environment; S60, continuously optimizing the neural network parameters based on reinforcement learning until the training of the neural network is completed, and finally achieving the self-adaptation of the intelligent agent.