Abstract:With the constant evolution of information technology, the paradigm of human-computer interaction has progressively transitioned from emphasizing command behavior to a novel approach that prioritizes natural language and emotional communication. In response to this trend, the development of an effective model for emotion recognition has become imperative. Consequently, this article introduces a novel multi-axis EEG channel attention model (MECAM) designed for emotion recognition. To enhance the model's capabilities, we employ a split depth-wise convolution with a larger convolutional kernel, resulting in the creation of four parallel branches. This not only reduces the computational complexity but also expands the receptive field of the model. The incorporation of a multi-axis attention mechanism, capturing both global and local information from the output features of parallel branches, further elevates the network's receptive field through the MBConv block. The primary objective of MECAM is to adeptly extract discriminative features from electroencephalogram (EEG) data, thereby enhancing the performance of emotion recognition. The model underwent rigorous validation using multiple datasets, including database for emission analysis using physical signals (DEAPs), SJTU emission EEG dataset (SEED), and SJTU emission EEG dataset with four emission (SEED-IV). In subject-dependent experiments, the accuracy of binary and ternary emotion classification tasks exceeded 95%, while the accuracy of the quaternary classification task surpassed 93%. In subject-independent experiments, the accuracy of most emotion classification tasks also exceeded 90%, with the exception of quaternary classification on DEAP, achieving an accuracy of 86.64%. The experimental results unequivocally underscore the superior performance of MECAM in EEG-based emotion recognition tasks.

Operation Mode Recognition With Multiple Sensing Data for Electro-Mechanical Actuator Based on Deep-Shallow Fusion Network

Gesture Classification in Electromyography Signals for Real-Time Prosthetic Hand Control Using a Convolutional Neural Network-Enhanced Channel Attention Model

Recognition of Emotions Using Multimodal Physiological Signals and an Ensemble Deep Learning Model

MF-Net: a multimodal fusion network for emotion recognition based on multiple physiological signals

Multi-Scale Feature Fusion Convolutional Neural Networks for Fault Diagnosis of Electromechanical Actuator

A Multimodal Multilevel Converged Attention Network for Hand Gesture Recognition With Hybrid sEMG and A-Mode Ultrasound Sensing

Multimodal Regression and Mode Recognition Via an Integrated Deep Neural Network

A multi-stage dynamical fusion network for multimodal emotion recognition

Multimodal Physiological Signal Emotion Recognition Based on Convolutional Recurrent Neural Network

Multimodal Emotion Recognition Model using Physiological Signals

Operation Mode Recognition of Airborne Radar Based on Multi-Feature Fusion RS-ConvNet

Multimodal Emotion Recognition From EEG Signals and Facial Expressions

Sensor Fusion for Myoelectric Control Based on Deep Learning With Recurrent Convolutional Neural Networks

MSFF-Net: Multi-Stream Feature Fusion Network for surface electromyography gesture recognition

MECAM: A Novel Multi-Axis EEG Channel Attention Model for Emotion Recognition

Emotion recognition using multimodal deep learning in multiple psychophysiological signals and video

Learning-Based Multimodal Information Fusion and Behavior Recognition of Vascular Interventionists' Operating Skills

Multi-level Attention Fusion for Multimodal Driving Maneuver Recognition

Multimodal Emotion Recognition Using a Hierarchical Fusion Convolutional Neural Network

A novel feature fusion network for multimodal emotion recognition from EEG and eye movement signals

A novel signal channel attention network for multi-modal emotion recognition