Showing 1–2 of 2 results for author: Mirsalari, S A

Multi-level Binarized LSTM in EEG Classification for Wearable Devices

Authors: Najmeh Nazari, Seyed Ahmad Mirsalari, Sima Sinaei, Mostafa E. Salehi, Masoud Daneshtalab

Abstract: Long Short-Term Memory (LSTM) is widely used in various sequential applications. Complex LSTMs could be hardly deployed on wearable and resourced-limited devices due to the huge amount of computations and memory requirements. Binary LSTMs are introduced to cope with this problem, however, they lead to significant accuracy loss in some application such as EEG classification which is essential to be… ▽ More Long Short-Term Memory (LSTM) is widely used in various sequential applications. Complex LSTMs could be hardly deployed on wearable and resourced-limited devices due to the huge amount of computations and memory requirements. Binary LSTMs are introduced to cope with this problem, however, they lead to significant accuracy loss in some application such as EEG classification which is essential to be deployed in wearable devices. In this paper, we propose an efficient multi-level binarized LSTM which has significantly reduced computations whereas ensuring an accuracy pretty close to full precision LSTM. By deploying 5-level binarized weights and inputs, our method reduces area and delay of MAC operation about 31* and 27* in 65nm technology, respectively with less than 0.01% accuracy loss. In contrast to many compute-intensive deep-learning approaches, the proposed algorithm is lightweight, and therefore, brings performance efficiency with accurate LSTM-based EEG classification to real-time wearable devices. △ Less

Submitted 19 April, 2020; originally announced April 2020.

Comments: o appear in IEEE International Conference on Parallel, Distributed and Network-based Processing in 2020. arXiv admin note: text overlap with arXiv:1812.04818 by other authors

MSC Class: 03B05 ACM Class: I.2.6; B.8.2

MuBiNN: Multi-Level Binarized Recurrent Neural Network for EEG signal Classification

Authors: Seyed Ahmad Mirsalari, Sima Sinaei, Mostafa E. Salehi, Masoud Daneshtalab

Abstract: 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 method… ▽ More 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. △ Less

Submitted 19 April, 2020; originally announced April 2020.

Comments: To appear in IEEE International Symposium on Circuits & Systems in 2020. arXiv admin note: text overlap with arXiv:1807.04093 by other authors

MSC Class: 03B05 ACM Class: I.2.6; B.8.2