Activity Recognition and Classification via Deep Neural Networks
Published in TridentCom (International Conference on Testbeds and Research Infrastructures), 2019
Summary: We present deep neural models for multi-class activity recognition and evaluate them on controlled datasets.
Abstract
The paper examines feedforward and temporal DNNs for recognizing human activities from sensor sequences. We analyze design trade-offs in depth, temporal context, and regularization, and report consistent performance gains over classical baselines. The results inform practical architectures for deployment-ready recognition systems. 