A new stock prediction method called AFG is proposed based on the application of artificial intelligence in financial data.AFG uses position encoding and time encoding to obtain the position and time information of stock data,and then extracts features from the stock data separately through gated loop units and multi head self attention mechanisms.After fusing the features of the two types of stocks,the final stock prediction curve is derived from the fully connected layer.
关键词
股票预测/门控循环单元/多头自注意力机制/位置编码/时间编码
Key words
stock prediction/gate control loop unit/attention mechanism/location coding/time coding
维普
万方数据