<ul><li>SETransformer is a hybrid deep neural architecture designed for Human Activity Recognition (HAR) using wearable sensor data.</li><li>It combines Transformer-based temporal modeling with channel-wise squeeze-and-excitation (SE) attention and a learnable temporal attention pooling mechanism.</li><li>SETransformer outperforms traditional models such as LSTM, GRU, BiLSTM, and CNN baselines on the WISDM dataset, achieving a validation accuracy of 84.68% and a macro F1-score of 84.64%.</li><li>The model shows promising potential for deployment in mobile and ubiquitous sensing applications, offering a competitive and interpretable solution for real-world HAR tasks.</li></ul>

SETransformer: A Hybrid Attention-Based Architecture for Robust Human Activity Recognition

Discover more