Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

vEEGNet: learning latent representations to reconstruct EEG raw data via variational autoencoders

Alberto Zancanaro, Giulia Cisotto, Italo Zoppis, Sara Lucia Manzoni

TL;DR

vEEGNet introduces a two-branch framework that combines a variational autoencoder with EEGNet to jointly classify motor-imagery EEG and reconstruct raw signals. Two variants, vEEGNet1 and vEEGNet2, explore different encoder designs and latent capacities, achieving competitive classification while expanding reconstruction beyond ultra-low frequencies in the second variant. Evaluated on the IV BCI 2a dataset, the approach demonstrates MRCP-like low-frequency reconstruction with vEEGNet1 and middle-frequency reconstruction with vEEGNet2, though full-segment reconstruction remains challenging and highly subject-dependent. The work highlights the potential of aligned generative latent representations for EEG and points to future directions in synthetic data generation and cross-dataset generalization.

Abstract

Electroencephalografic (EEG) data are complex multi-dimensional time-series that are very useful in many applications, from diagnostics to driving brain-computer interface systems. Their classification is still a challenging task, due to the inherent within- and between-subject variability and their low signal-to-noise ratio. On the other hand, the reconstruction of raw EEG data is even more difficult because of the high temporal resolution of these signals. Recent literature has proposed numerous machine and deep learning models that could classify, e.g., different types of movements, with an accuracy in the range 70% to 80% (with 4 classes). On the other hand, a limited number of works targeted the reconstruction problem, with very limited results. In this work, we propose vEEGNet, a DL architecture with two modules, i.e., an unsupervised module based on variational autoencoders to extract a latent representation of the data, and a supervised module based on a feed-forward neural network to classify different movements. To build the encoder and the decoder of VAE we exploited the well-known EEGNet network. We implemented two slightly different architectures of vEEGNet, thus showing state-of-the-art classification performance, and the ability to reconstruct both low-frequency and middle-range components of the raw EEG. Although preliminary, this work is promising as we found out that the low-frequency reconstructed signals are consistent with the so-called motor-related cortical potentials, well-known motor-related EEG patterns and we could improve over previous literature by reconstructing faster EEG components, too. Further investigations are needed to explore the potentialities of vEEGNet in reconstructing the full EEG data, generating new samples, and studying the relationship between classification and reconstruction performance.

vEEGNet: learning latent representations to reconstruct EEG raw data via variational autoencoders

TL;DR

vEEGNet introduces a two-branch framework that combines a variational autoencoder with EEGNet to jointly classify motor-imagery EEG and reconstruct raw signals. Two variants, vEEGNet1 and vEEGNet2, explore different encoder designs and latent capacities, achieving competitive classification while expanding reconstruction beyond ultra-low frequencies in the second variant. Evaluated on the IV BCI 2a dataset, the approach demonstrates MRCP-like low-frequency reconstruction with vEEGNet1 and middle-frequency reconstruction with vEEGNet2, though full-segment reconstruction remains challenging and highly subject-dependent. The work highlights the potential of aligned generative latent representations for EEG and points to future directions in synthetic data generation and cross-dataset generalization.

Abstract

Electroencephalografic (EEG) data are complex multi-dimensional time-series that are very useful in many applications, from diagnostics to driving brain-computer interface systems. Their classification is still a challenging task, due to the inherent within- and between-subject variability and their low signal-to-noise ratio. On the other hand, the reconstruction of raw EEG data is even more difficult because of the high temporal resolution of these signals. Recent literature has proposed numerous machine and deep learning models that could classify, e.g., different types of movements, with an accuracy in the range 70% to 80% (with 4 classes). On the other hand, a limited number of works targeted the reconstruction problem, with very limited results. In this work, we propose vEEGNet, a DL architecture with two modules, i.e., an unsupervised module based on variational autoencoders to extract a latent representation of the data, and a supervised module based on a feed-forward neural network to classify different movements. To build the encoder and the decoder of VAE we exploited the well-known EEGNet network. We implemented two slightly different architectures of vEEGNet, thus showing state-of-the-art classification performance, and the ability to reconstruct both low-frequency and middle-range components of the raw EEG. Although preliminary, this work is promising as we found out that the low-frequency reconstructed signals are consistent with the so-called motor-related cortical potentials, well-known motor-related EEG patterns and we could improve over previous literature by reconstructing faster EEG components, too. Further investigations are needed to explore the potentialities of vEEGNet in reconstructing the full EEG data, generating new samples, and studying the relationship between classification and reconstruction performance.
Paper Structure (11 sections, 7 equations, 4 figures, 3 tables)

This paper contains 11 sections, 7 equations, 4 figures, 3 tables.

Table of Contents

  1. Introduction
  2. State of the art
  3. Methods
  4. Variational Autoencoder
  5. vEEGNet general architecture
  6. Unsupervised mechanism
  7. Supervised mechanism
  8. Results and discussion
  9. Dataset and vEEGNet implementations
  10. vEEGNet performance
  11. Conclusions

Figures (4)

  • Figure 1: The general vEEGNet architecture (modified from our previous work Zancanaro_2023_vEEGNet).
  • Figure 2: The vEEGNet1 architecture, with all implementation details (i.e., parameters and activation functions).
  • Figure 3: An example of an signal (channel C3, right-hand MI) as reconstructed by (a) vEEGNet1 and (b) vEEGNet2.
  • Figure 4: An example of 4 different signals from 4 target sensors (namely, C3, C4, Cz, and the average between FC3 and FC4) as reconstructed by (a) vEEGNet and (b) vEEGNet2. Panel (a) has been reported from our previous conference paper Zancanaro_2023_vEEGNet to show the difference with the results from vEEGNet2.