研究项目: Multi-frequency Factor of RNN with Attention Mechanism Shenzhen, China 项目负责人 May. 2023 – Jun. 2023
 Established the GRU model for time series stock data within daily frequency and minute-frequency stock data as basic models and processed the attention mechanism framework to optimize the prediction.
 Combined with different frequency data, the parameter adjustment process was carried out to maximize the performance of the GRU model for the output of a single prediction, and the effect was significantly improved, with the portfolio return rate increasing by about 6% at least.

研究项目: Deep Learning Model for News Text Detection by AI or Human Shenzhen, China项目负责人 Apr. 2023 – May. 2023
 Used black box detection method to build RNN basic backtest, considering the control requirements of neural network for data analysis, with the backtest accuracy rate reaching 99.9% for similar texts.
 Used LSTM and GRU algorithms to optimize RNN, considering timing and test performance issues; GloVe was used to replace Word2vec for data vectorization analysis.
 Optimized the outcome using the BERT kernel to improve word recognition to satisfy text variety.

研究项目: Machine learning analysis of stock price prediction Shenzhen, China 项目负责人 Nov. 2022 – Dec. 2022
 Used tree prediction machine learning model techniques to analyze the accuracy compared with linear regression and CAPM model, demonstrating the effectiveness of SDT, RF, and Bagging tree machine learning models on stock price prediction.
 Adopted the CSI 500 stock as the stock pool and selected 10 stocks of different types. The model’s accuracy reached 98%, and the standard deviation of fluctuation control was 0.03, whi