# ConTraNet: A single end-to-end hybrid network for EEG-based and EMG-based human machine interfaces

**Authors:** Omair Ali, Muhammad Saif-ur-Rehman, Tobias Glasmachers, Ioannis Iossifidis, Christian Klaes  
**arXiv:** [2206.10677](https://arxiv.org/abs/2206.10677)  
**Date:** 2022-06-21  
**Summary page:** https://www.summarizepaper.com/en/paper/2206.10677v1/

## Key points

- Objective: Improve decoding process of EEG and EMG signals for human-machine interface (HMI) technologies
- Recent advancements: CNNs effectively map raw time-series signals into decision space
- Limitation: Existing networks are often limited to a single paradigm
- Novel hybrid model ConTraNet introduced, combining CNN and Transformer architectures
- ConTraNet proven effective for both EEG-HMI and EMG-HMI paradigms by leveraging self-attention mechanisms
- Evaluation against state-of-the-art methods using three publicly available datasets showed superior performance across various category tasks
- Potential impact of ConTraNet as a versatile solution for decoding bio-signals within HMI technologies, enhancing rehabilitation processes for individuals with physical disabilities

## Summary

Objective: The objective of this study is to improve the decoding process of EEG and EMG signals into control commands for human-machine interface (HMI) technologies, specifically within the EEG-HMI and EMG-HMI paradigms. Recent advancements have introduced Convolutional Neural Networks (CNNs) based architectures that can effectively map raw time-series signals into decision space. However, these networks are often limited to a single paradigm. To address this limitation, a novel hybrid model named ConTraNet has been introduced in this study, which combines elements from CNN and Transformer architectures. This model has proven to be equally effective for both EEG-HMI and EMG-HMI paradigms by leveraging self-attention mechanisms to capture long-range dependencies crucial for accurate classification of EEG and EMG signals. The efficacy of ConTraNet was evaluated against state-of-the-art methods using three publicly available datasets representing different HMI paradigms, showing superior performance across various category tasks. These results highlight the potential impact of ConTraNet as a versatile solution for decoding bio-signals within HMI technologies, making it a significant advancement towards enhancing rehabilitation processes for individuals with physical disabilities. Keywords: , , , , .