基于注意力机制的双向LSTM人体活动识别

Human activity recognition based on attention mechanism bidirectional LSTM

针对目前大多数的信道状态信息(CSI)活动识别需要通过人工手动来提取特征,耗时且繁琐,准确率不高的问题,提出了基于双向长短期记忆(Bi-LSTM)神经网络加注意力机制的识别方法。将传统的LSTM网络扩展为双向LSTM并加入注意力机制,实现有选择性地提取序列的前后信息,从而提升了人体活动识别的准确率。实验结果表明,所提出的方法具有更好的识别效果。

Aiming at the problem that at present,most channel state information(CSI)activity recognition requires manual extraction features,which is time-consuming and cumbersome,and has low accuracy,a recognition method based on bidirection long short-term memory(Bi-LSTM)neural network and attention mechanism is proposed.Traditional LSTM network is extended to Bi-LSTM and added attention mechanism,and selectively extract information of sequence,so as to improve accuracy of human activity recognition.Experimental results show that the proposed method has better recognition effect.