金融工程：AI识图关注化工、非银和卫星

• The report introduces a quantitative model based on Convolutional Neural Networks (CNN) to analyze price-volume data and predict future stock prices. The model maps learned features to industry theme indices, including chemical, non-bank financial, and satellite sectors[77][79][80] - The construction process involves standardizing price-volume data into graphical formats for each stock within a specific window period. These standardized charts are then used as input for the CNN model to identify patterns and predict future trends[77][78][79] - The model's evaluation highlights its ability to capture complex relationships in price-volume data and its application in thematic industry allocation. It emphasizes the importance of deep learning techniques in quantitative finance[77][80] - Backtesting results show the latest thematic allocations include indices such as CSI Subdivision Chemical Industry Theme Index, CSI Satellite Industry Index, CSI All Share Dividend Quality Index, and others, reflecting the model's predictions[79][80]