Sensing human activity via WiFi Channel State Information (CSI) has considerable application prospects in future intelligent interaction scenarios such as virtual reality, intelligent games, metaverse, etc. Recently, many Deep Learning-based WiFi sensing schemes have been proposed in the literature, which gained high accuracy for a wide range of simple activities such as standing, squatting, and bending. However, the performance will be suffered when existing approaches are used to recognize drastic activities, such as actions in vigorous sports. This is mainly due to the reason that the spatiotemporal information of these actions is not well utilized. To overcome this drawback, we propose a novel DL-based WiFi sensing method for drastic activity recognition by combining the Convolutional Neural Network (CNN) and the Bidirectional Long Short-Term Memory (BiLSTM) network. The designed CNN-BiLSTM architecture is in parallel with feature extraction, which can simultaneously extract sufficient spatiotemporal features of action data and establish the mapping relationship between actions and CSI streams, thereby improving the accuracy of activity recognition. The CNN is used to extract information on the spatial dimension, while the BiLSTM extracts information on the time dimension. To verify the performance of the proposed scheme, we build a hardware experiment platform and constrain a dataset with 1400 pieces of records for 7 classes of basketball actions. After training over the dataset, the proposed CNN-BiLSTM scheme achieves 96% experimental accuracy on the test set, which is better than the benchmark methods.

Identifying driver pathway is a critical step to uncover the natural laws of the occurrence and progression of disease. Many studies show that multiple pathways often function cooperatively in carcinogenesis. However, how to computationally identify cooperative driver pathways of cancers is not well studied yet. Existing cooperative driver pathway identification methods either suffer from single type of genetic information source or computation difficulty. In this paper, we proposed a method (CDPLP) based on hierarchical clustering and link prediction. CDPLP firstly devises a new similarity metric to quantity the exclusivity and co-expression of two gene modules, and thus to obtain gene sets with exclusivity by hierarchical clustering. Next, it uses link prediction on the pathway-pathway interaction network to replenish the interactions between pathways. After that, CDPLP combines the gene sets and updated pathway network to discover the pathway pairs with high functional interaction and occurrence as cooperative pathways. CDPLP can make full use of multiple genetic information sources such as the mutation data, gene-gene interaction data and pathway-pathway network, and facilitate the optimization solution. We evaluated the performance of CDPLP on TCGA breast cancer (BRCA) dataset and compared it with other popular methods. The results show that cooperative driver pathways identified by CDPLP are highly associated with the target cancer, and are involved with carcinogenesis and several key biological processes.