Abstract:Abstract With the continuous expansion of the scale of the network, modeling the complex graph structure is a major challenge for the recommendation task. There are conflicts between these complex information, which will directly affect the recommendation results. For this reason, a graph neural network recommendation method based on multi-branch decision is proposed. The algorithm models the social network graph and the user project graph through the neural network, connects the two graphs intrinsically, and learns the feature vectors of the target users in the social space and the project space. Then two feature vectors are connected in series by MLP to extract the potential feature vector of the user. Finally, the prediction score is generated by integrating the probability matrix decomposition model. In this method, the graph neural network is used to model the program data and dependence graph, and the effective program features are automatically extracted from the source code, and then the extracted features are input into the downstream model for loop vectorization parameter prediction. Finally, the attention mechanism is used for information fusion to get the final session representation and predict the next interactive item. Comparative experiments are carried out under the two scenarios of e-commerce and civil aviation respectively. The experimental results show that, compared with the optimal benchmark model, the improvement of MGSP model on each index of e-commerce data set is more than 1%, and that of civil aviation data set is about 3%, which verifies the effectiveness of MGSP model. On the LLVM cyclic vector test set, the proposed method achieves a speedup of 2.08 times and improves the performance by 12% compared with the existing methods. In order to solve the problem that the existing deep learning model considers the program code as a serial sequence and misses a large performance optimization space, a new program heuristic method based on depth map network is proposed to achieve optimization.

A Fine-grained Technology Convergence Prediction Method Based on Graph Neu-ral Networks

From technology opportunities to solutions generation via patent analysis: Application of machine learning-based link prediction

Convergence Analysis of Graph Regularized Non-Negative Matrix Factorization

Technology opportunity discovery linking artificial intelligence and construction technologies: A graph convolution network-based approach

Lane-Level Traffic Flow Prediction with Heterogeneous Data and Dynamic Graphs

Predictive modeling for technology convergence: A patent data-driven approach through technology topic networks

A Market Convergence Prediction Framework Based on A Supply Chain Knowledge Graph

Self-adaptive Graph Convolution Networks with Application to Industrial Soft Sensor Modeling

Research on Machine Vision Effect Based on Graph Neural Network Decision

Convergence Technology Opportunity Discovery for Firms Based on Technology Portfolio Using the Stacked Denoising AutoEncoder (SDAE)

Long-term Multivariate Time Series Forecasting in Data Centers Based on Multi-Factor Separation Evolutionary Spatial–temporal Graph Neural Networks

Research topic trend prediction of scientific papers based on spatial enhancement and dynamic graph convolution network

The prediction of the quality of results in Logic Synthesis using Transformer and Graph Neural Networks

Automated Spatio-Temporal Synchronous Modeling with Multiple Graphs for Traffic Prediction

Adaptive spatio-temporal graph convolutional network with attention mechanism for mobile edge network traffic prediction

Optimized Graph Convolution Recurrent Neural Network for Traffic Prediction

Spatiotemporal fuzzy-graph convolutional network model with dynamic feature encoding for traffic forecasting

Time series prediction for production quality in a machining system using spatial-temporal multi-task graph learning

A Graph Neural Network Approach for Product Relationship Prediction

Implementing link prediction in protein networks via feature fusion models based on graph neural networks

A short-term wind power prediction method via self-adaptive adjacency matrix and spatiotemporal graph neural networks