Research On Activity Status Classification And Activity Recognition Based On Personal Information

With the rise of 5G technology and the popularity of Wi Fi technology,more and more researchers have begun to study how to use Wi Fi signals for activity recognition.Because the traditional machine learning method needs to use a large number of manual methods to extract features and is easily affected by the environment,while a single neural network can only extract features in a specific field,it faces the bottleneck of recognition accuracy and poor robustness.This paper proposes a novel network architecture that combines the advantages of convolutional neural networks and bidirectional long shortterm memory networks to improve the performance of activity recognition.The main research contents of this paper are as follows.First,since daily activities are continuous,effective segmentation of activities becomes the key to activity recognition.The traditional activity segmentation technology is based on a certain feature affected by the activity to perform the optimal threshold segmentation,which is easily affected by the environment and has poor effect.In this paper,an algorithm using a two-layer decision tree is proposed to adjust the appropriate threshold according to the characteristics,which improves the accuracy of activity segmentation.Second,because a single neural network model can only extract features from a specific domain,the accuracy of activity recognition is limited.This paper combines the advantages of convolutional neural network and bidirectional long short-term memory neural network,and proposes a joint neural network CNN-Bi LSTM model,which improves the performance of activity recognition in terms of training time and recognition accuracy.Finally,the experiment of the activity identification model proposed in this article was experiment.The experiment found that the model of this article has a high recognition accuracy and the model of this article has a good robustness.Compare the model of this article with a single convolutional neural network and a single two-way long-term memory network,and use the accuracy and F1-SCORE evaluation indicators for model evaluation to verify the effectiveness of the model in this article in terms of activity recognition.