MaskCAE: Masked Convolutional AutoEncoder via Sensor Data Reconstruction for Self-Supervised Human Activity Recognition.

Self-supervised Human Activity Recognition (HAR) has been gradually gaining a lot of attention in ubiquitous computing community. Its current focus primarily lies in how to overcome the challenge of manually labeling complicated and intricate sensor data from wearable devices, which is often hard to interpret. However, current self-supervised algorithms encounter three main challenges: performance variability caused by data augmentations in contrastive learning paradigm, limitations imposed by traditional self-supervised models, and the computational load deployed on wearable devices by current mainstream transformer encoders.

Comparison of nitrogen removal efficiency and microbial characteristics of modified two-stage A/O, A/O and SBR processes.

The low temperature of sewage in north China results in low performance of biological treatment at municipal wastewater treatment plants (MWTPs), especially in biological nitrogen removal. A modified two-stage A/O process with an embedded biofilm was proposed to enhance nitrogen removal. The operation performance of a pilot test was compared with an A/O and SBR process at existing MWTPs to investigate the resistance to low temperature.