7 results about "Multi aspect" patented technology

The invention discloses a next interest point recommendation method based on multi-aspect continuous dynamic preference modeling of hypergraph ODE. According to the method, modeling can be efficiently carried out on multi-aspect dynamic preferences of the user from a continuous angle, and more personalized recommendation is provided for the user. Multi-aspect user preferences are decoupled by constructing a multi-view hypergraph from collaborative, geographic, and transformational aspects. By expanding discrete hypergraph convolution operation into a continuous form based on an ordinary differential equation (ODE), continuous evolution of multi-aspect user dynamic preferences is modeled. User preferences under different views are integrated by adopting a self-adaptive fusion mechanism, and a dynamic relationship between the user preferences is captured. In addition, an inter-view and in-view comparative learning normal form is designed to enhance the expressivity and robustness of interest point representation. A large number of experiments carried out on four real spatio-temporal data sets with different scales verify that the method has remarkable advantages and efficiency.

The application provides a system and method for multi-angle examination and evaluation of engineering contractors and employees, which can instantly manage and record multi-aspect data related to the examination of the contractors and employees, determine examination results by using corresponding examination models according to examination requirements, provide data display services and examination result output services matched with instructions in response to the operation of users, identify data attributes and decide storage paths based on received data, decide management action signals based on received instructions, distribute and store multiple data, support real-time data state updating in response to management action instructions, and form prompt instructions according to set rules on data state information. The system can overcome the defects of the prior art, such as complicated data entry, limited application range of evaluation and analysis, and insufficient reliability, form a comprehensive business data examination and management system for engineering contractors and employees, and provide business monitoring and credit repair and update services based on examination data.

The application provides a person image clustering method and device, relates to the technical field of image clustering, and is used for improving the accuracy of person image clustering. In the person image clustering method, the self-organizing mapping network is used to process the multi-aspect attribute features of the lens, so as to determine the winning neuron corresponding to the lens, and the winning neurons to which the multiple lenses respectively belong are used to determine the cluster to which the multiple lenses belong, so that a spatial domain clustering manner for person images is provided. Since the clustering results of the multiple lenses are not affected by the quality of the person images, the reliability of the lens clustering is improved, and the reliability of the person image clustering is improved.

The application realizes a commodity search method with query-related multi-aspect explainability by means of a method in the field of artificial intelligence. The search information of an input commodity is searched and the query commodity result is fed back by using a multi-path query perception graph convolution network for explainable commodity search. The multi-path query perception graph convolution network for explainable commodity search comprises two components: a query perception graph convolution sorter and a query perception multi-path reasoner. The query perception graph convolution sorter models the representation of a user and a commodity according to different knowledge relationship domains in a knowledge graph by using a graph convolution network. The query perception multi-path reasoner is responsible for exploring a query-specific multi-path from the knowledge graph to meet the search intention of the user. The two components share basic parameters and are collaboratively trained to constitute a complete network. The method can significantly improve the retrieval performance and generate better explanations for the search result.

The application belongs to the technical field of artificial intelligence, and discloses a multi-behavior sequence recommendation system and method for capturing user multi-aspect preferences and intentions, which comprises a preprocessing module, a multi-aspect preference modeling module, a multi-aspect intention modeling module, a preference and intention fusion module and a probability output module. The preprocessing module pre-processes multi-type behavior data; the multi-aspect preference modeling module models user multi-aspect preferences through a multi-aspect projection mechanism; the multi-aspect intention modeling module models user multi-aspect intentions through a behavior detail perception LSTM and a preference-guided multi-aspect attention mechanism; the preference and intention fusion module adaptively integrates user preference and intention information through a multi-aspect preference and intention fusion mechanism; and the probability output module, which is connected to the preference and intention fusion module, predicts the next item and category that the user is interested in. The system can capture user multi-aspect preferences and intentions and obtain accurate prediction results.

The application discloses a zero-shot stance detection method and system based on multi-expert collaborative learning, which introduces multiple experts to learn the underlying multi-aspect features of text semantic expression, so as to improve the zero-shot stance classification accuracy of unknown targets. First, an expert model is constructed based on BERT, and different experts use different underlying outputs of BERT as text encoding to obtain different aspects of text semantic representation. Second, a gating mechanism is used to filter the effective feature aspects of the stance, so as to improve the feature quality of the stance prediction. Finally, the filtered multi-aspect semantic deconstruction features and high-level semantic features are comprehensively used to realize the detection of the text stance. The application is based on the deconstruction of the text semantic expression features, and the text semantic features are modeled in a fine-grained manner, so as to improve the feature transferability, obtain more effective features, and improve the zero-shot stance classification accuracy of unknown targets.

The application discloses a small sample target detection method based on contrast learning and multi-aspect distillation, constructs a small sample target detection network model based on contrast learning and multi-aspect distillation, carries out basic class basic training on the small sample target detection network model, obtains a small sample target detection basic class model, carries out new class fine-tuning training on the small sample target detection basic class model, obtains a final small sample target detection model, and finally detects small sample targets according to the final small sample target detection model. The method solves the problems that the traditional target detection method is prone to class confusion and weak information capturing capacity, greatly improves the capturing capacity of the detection model on target features, simultaneously strengthens foreground representation and reduces the interference of background information, and improves target detection performance.