2020/04/19 by Seyed Ahmad Mirsalari, Sima Sinaei, Mirsalari, Seyed Ahmad +5
Computer Science · Engineering · Neuroscience · #03B05 #Advanced Memory and Neural Computing #B.8.2 #Blind Source Separation Techniques #EEG and Brain-Computer Interfaces #FOS: Computer and information sciences #FOS: Electrical engineering #I.2.6 #Machine Learning (cs.LG) #Neural and Evolutionary Computing (cs.NE) #Signal Processing (eess.SP) #acm:03B05 #cs.LG #cs.NE #eess.SP #electronic engineering #information engineering #msc:03B05
paper · pdf · doi:10.48550/arxiv.2004.08914
To appear in IEEE International Symposium on Circuits & Systems in 2020. arXiv admin note: text overlap with arXiv:1807.04093 by other authors
arxiv created 2020/04/19 · openalex publication_date 2020/04/19 · arxiv updated 2020/04/21 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28
Recurrent Neural Networks (RNN) are widely used for learning sequences in applications such as EEG classification. Complex RNNs could be hardly deployed on wearable devices due to their computation and memory-intensive processing patterns. Generally, reduction in precision leads much more efficiency and binarized RNNs are introduced as energy-efficient solutions. However, naive binarization methods lead to significant accuracy loss in EEG classification. In this paper, we propose a multi-level binarized LSTM, which significantly reduces computations whereas ensuring an accuracy pretty close to the full precision LSTM. Our method reduces the delay of the 3-bit LSTM cell operation 47* with less than 0.01% accuracy loss.