Chao Zhang

This work proposed a multiplexer-based neural network (MUXnet), a multiplexer-based, multiplier-free neural network (NN) structure applicable to the implementation of all inner product-based NN layers. An on-chip MUXnet-based neural signal processing unit (NSPU) was designed, achieving a state-of-the-art accuracy of 82.4% on a public human sleep staging dataset, with the lowest reported energy consumption of $0.2~\mu $ J per classification. Building on MUXnet, this work introduced the first electrophysiology-optogenetics closed-loop bi-directional neural interface integrating neural signal recording, on-chip deep NN (DNN)-based sleep staging, and optogenetic stimulation. An in vivo sleep regulation experiment was conducted on animal subjects, demonstrating high efficiency in waking subjects from the non-rapid eye movement (NREM) sleep stage.

This paper proposes an efficient structure for electroencephalogram (EEG) signal recording. A single-sample strategy is proposed to mitigate metastability issues in Voltage-Controlled Oscillator (VCO) Analog Front Ends (AFEs), offering timing margins of 14 phase cycles and simplifying the result arbitration logic to 1-bit multiplexing. Additionally, an analysis of existing EEG sub-band filter designs is presented, followed by an efficient band multiplexing serial multiplier structure that capitalizes on timing slacks. This design features a reduction in both the number of multipliers and the complexity of the multiplexing network. The proposed design was implemented using 40nm CMOS technology. The VCO-AFE demonstrates stable, error-free recordings with an input-referred noise (IRN) of 0.66μVrms within 0.5–60Hz, according to the measurement results. The proposed sub-band filter exhibits substantial savings of 11% and 51% in area and power, respectively, compared to prior work when scaled to the same technology node.

This paper proposes to use cepstrum for artifact detection, recognition and removal in prefrontal EEG. This work focuses on the artifact caused by eye movement. A database containing artifact-free EEG and eye movement contaminated EEG from different subjects is established. A cepstral analysis-based feature extraction with support vector machine (SVM) based classifier is designed to identify the artifacts from the target EEG signals. The proposed method achieves an accuracy of 99.62% on the artifact detection task and a 82.79% accuracy on the 6-category eye movement classification task. A statistical value-based artifact removal method is proposed and evaluated on a public EEG database, where an accuracy improvement of 3.46% is obtained on the 3-category emotion classification task. In order to make a confident decision of each 5s EEG segment, the algorithm requires only 0.66M multiplication operations. Compared to the state-of-the-art approaches in artifact detection and removal, the proposed method features higher detection accuracy and lower computational cost, which makes it a more suitable solution to be integrated into a real-time and artifact robust Brain-Machine Interface (BMI).

This paper proposed a Multi-Channel Multi-Domain (MCMD) based knowledge distillation algorithm for sleep staging using single-channel EEG. Both knowledge from different domains and different channels are learnt in the proposed algorithm, simultaneously. A multi-channel pre-training and single-channel fine-tuning scheme is used in the proposed work. The knowledge from different channels in the source domain is transferred to the single-channel model in the target domain. A pre-trained teacher-student model scheme is used to distill knowledge from the multi-channel teacher model to the single-channel student model combining with output transfer and intermediate feature transfer in the target domain. The proposed algorithm achieves a state-of-the-art single-channel sleep staging accuracy of 86.5%, with only 0.6% deterioration from the state-of-the-art multi-channel model. There is an improvement of 2% compared to the baseline model. The experimental results show that knowledge from multiple domains (different datasets) and multiple channels (e.g. EMG, EOG) could be transferred to single-channel sleep staging.

In deep-learning-based speech enhancement (SE), an audio-knowledge-ignorant approach is often used, which estimates a denoising model to transform the noisy input speech into clean output speech without understanding the audio events that constitute the background noises. In this paper, an audio-knowledge-aware approach is proposed to improve SE, which explicitly leverages the knowledge from audio taggings to understand the background noises. Based on the recent progress in audio pattern analysis, the audio tagging knowledge is obtained using either additional input representations extracted by pre-trained audio tagging models, or from multi-task learning with extra audio event classification or regression tasks. Experimental results based on the DNS-2020 dataset and the pre-trained Wavegram-Logmel-CNN audio tagging model show that the proposed approach leads to considerable improvements in the STOI, SDR, and SI-SNR metrics.

Neural interfaces play an important role in vital sign monitoring [1] , clinical practice [2] , and neuroscience frontier research [3] using animal subjects, such as rodents and nonhuman primates [4] , [5] . Recently, neuroscientists have shown increasing interest in behavioral analysis of social situations with a single or multiple freely moving animals, which has been listed as a key investment by the NIH in 2023 [6] . Among those popular laboratory animals, marmosets are increasingly favored by researchers because of their small size, short sexual maturity, complex social patterns, and similarity to human beings [7] , [8] . A number of marmoset social behaviors, such as gaze following [9] , antiphonal calling [10] , and scent marking [11] , have been studied with neural signal acquisition and/or stimulation to affect the behavior of the freely moving marmoset at the same time.

This work proposed a decoding algorithm of the bioelectrical signal acquired from a head-fixed rabbit subject’s buccal nerves related to the corresponding whisker movements. The nerve-whisker matching algorithm is fulfilled by the signal pre-processing and the neural networks consisting of a Long Short-Term Memory Recurrent Neural Network (LSTM) layer and a fully-connected layer. Concordance Correlation Coefficient Loss is applied for training to capture the intricate relationships between nerve signals and corresponding whisker movements. The algorithm is also able to decode both amplitude and phase information of the recorded signal. A decoding accuracy of 80% is achieved according to in-vivo experimental results.

This paper proposes to use cepstrum for artifact detection, recognition and removal in prefrontal EEG. This work focuses on the artifact caused by eye movement. A database containing artifact-free EEG and eye movement contaminated EEG from different subjects is established. A cepstral analysis-based feature extraction with support vector machine (SVM) based classifier is designed to identify the artifacts from the target EEG signals. The proposed method achieves an accuracy of 99.62% on the artifact detection task and a 82.79% accuracy on the 6-category eye movement classification task. A statistical value-based artifact removal method is proposed and evaluated on a public EEG database, where an accuracy improvement of 3.46% is obtained on the 3-category emotion classification task. In order to make a confident decision of each 5s EEG segment, the algorithm requires only 0.66M multiplication operations. Compared to the state-of-the-art approaches in artifact detection and removal, the proposed method features higher detection accuracy and lower computational cost, which makes it a more suitable solution to be integrated into a real-time and artifact robust Brain-Machine Interface (BMI).

It is easy for the electroencephalogram (EEG) signal to be incomplete due to packet loss, electrode falling off, etc. This paper proposed a Cascade Transformer architecture and a loss weighting method for the single-channel EEG completion, which reduced the Normalized Root Mean Square Error (NRMSE) by 2.8% and 8.5%, respectively. With the percentage of the missing points ranging from 1% to 50%, the proposed method achieved a NRMSE from 0.026 to 0.063, which aligned with the state-of-the-art multi-channel completion solution. The proposed work shows it's feasible to perform the EEG completion with only single-channel EEG.

This paper proposes SaleNet - an end-to-end convolutional neural network (CNN) for sustained attention level evaluation using prefrontal electroencephalogram (EEG). A bias-driven pruning method is proposed together with group convolution, global average pooling (GAP), near-zero pruning, weight clustering and quantization for the model compression, achieving a total compression ratio of 183.11x. The compressed SaleNet obtains a state-of-the-art subject-independent sustained attention level classification accuracy of 84.2% on the recorded 6-subject EEG database in this work. The SaleNet is implemented on a Artix-7 FPGA with a competitive power consumption of 0.11 W and an energy-efficiency of 8.19 GOps/W.