To avoid the over-fitting phenomenon that appeared in performing Graph Neural Networks (GNNs) on test examples, the feature encoding scheme of GNNs usually introduces the dropout procedure. However, after learning latent node representations under this scheme, Gaussian noise produced by the dropout operation is inevitably transmitted into the next neighborhood aggregation step, which necessarily hampers the unbiased aggregation ability of GNN models. To address this issue, we present a novel aggregator, De-Noising Aggregation (DNAG), which utilizes Principal Component Analysis (PCA) to preserve the aggregated real signals from neighboring features and simultaneously filter out the Gaussian noise. The idea is different from using PCA on traditional applications to reduce the feature dimension. We regard PCA as an aggregator to compress the neighboring node features to have better expressive de-noising power. We propose new training architectures to simplify the intensive computation of PCA in DNAG. Numerical experiments show the apparent superiority of the proposed DNAG models in gaining more de-noising capability and achieving the state-of-the-art for a set of predictive tasks on several graph-structured datasets.

The Sensor-based Human Activity Recognition (SHAR) task seeks to recognize signals collected by various sensors embedded in intelligent devices to assist people in their daily lives. Motivated by the success of deep learning, many researchers are studying combining deep learning with SHAR. The key to implementing sensor-based human activity recognition with deep learning lies in facilitating model performance and maintaining efficiency when the model is performed on resource-constrained devices. To address this challenge, we present an effective sensor signal representation method, termed the Temporal-Spatial Dynamic Convolutional (TS-DyConv) network, to recognize human activity. TS-DyConv aims to dynamically learn the convolutional kernels weighted with attention generated along the temporal and spatial kernel spaces. This way, the TS-DyConv can diversify the kernel to enhance the sensor signal’s recognition capabilities without raising complexity and maintaining efficiency. Extensive experiments conducted on three benchmark SHAR datasets, e.g., OPPORTUNITY, PAMAP2, and USC-HAD, demonstrate the superiority of our method over the deep learning baselines and existing SHAR works.

The Sensor-based Human Activity Recognition (SHAR) task aims to detect and analyze signals captured by various sensors embedded in intelligent devices to assist people’s daily lives, which inherently have variations, spatial inconsistency, and noise characteristics. Inspired by the success of deep learning, many researchers have designed various neural network architectures to improve the SHAR model’s performance. Albeit effective, these manners are not optimal due to ignoring the imbalance and difficulty of inter- and intra-class of sensor inherently characteristic, which proposes the challenge for resource-constrained devices. To address this challenge, we present a simple and effective optimization strategy for sensor signal processing, including Dynamic Weight Optimization (DWO) and Smoothness Constraint (SC) strategies, termed DWOSC, which provide a framework to optimize loss function and enhance activity recognition performance. The DWO aims to address the class imbalance and hard-to-recognize samples. The SC seeks to maintain the smoothness of decision boundaries for complex samples. This way, the proposed approach can achieve a more stable convergence of the model and further improve the SHAR task’s performance without raising complexity. Extensive experiments conducted on three benchmark SHAR datasets, e.g., OPPORTUNITY, USC-HAD, and UniMib, demonstrate the superiority of our method over the deep learning baselines and existing SHAR works.