D2CAN: Domain-Guided Contrastive Adversarial Network for EEG-Based Cross-Subject Cognitive Workload Decoding

Communications in computer and information science, Journal Year: 2025, Volume and Issue: unknown, P. 250 - 264

Published: Jan. 1, 2025

Language: Английский

---

MAML-EEG: A Meta-learning Strategy Based Domain Generalization Framework for Unseen Subject Motor Imagery Classification

Journal of Physics Conference Series, Journal Year: 2025, Volume and Issue: 3004(1), P. 012037 - 012037

Published: May 1, 2025

Abstract Since the acquisition of electroencephalogram (EEG) is time-consuming and costly, target subjects usually unknown in practical application motor imagery (MI) based brain-computer interface (BCI) system. Due to heterogeneity individuals, distribution MI signals among quite different, which makes it difficult for traditional deep learning methods achieve cross-unknown applications. We propose a meta-learning (ML) method alleviates this problem by designing models that generalize well new test subjects. Specifically, our framwork simulates train/test subject shift during training stage randomly sampling partial samples synthesize virtual meta-task set within each batch. The final meta-optimization improves performance multiple while reducing meta-training loss subject. evaluated on classification, as obtaining its advanced performances two calssic available datasets.

Language: Английский

---

Meta-Learning-based Cross-Dataset Motor Imagery Brain-Computer Interface

Published: Feb. 26, 2024

Motor Imagery Brain-Computer Interface (MI-BCI) facilitates human to communicate with computers or machines using brain signals, such as electroencephalography (EEG), induced by the imagination of body movements. However, acquiring sufficient data for training reliable classification model is often time-consuming and impractical. Consequently, recent studies have shifted focus subject-independent EEG classification, leveraging from other subjects methodologies like transfer learning meta-learning. most exploit within same dataset, which might raise challenges especially when are scarce inaccessible. To address this issue, we propose a meta-learning-based cross-dataset MI classification. We first extract informative knowledge source dataset based on meta-learning framework. then leverage extracted (or meta-parameters) enhance performance target dataset. This method leverages BCI Competition IV-2a KU GIST datasets respectively. Our experimental results indicate that proposed enhances compared conventional subject-dependent scenarios.

Language: Английский

---

Harnessing Few-Shot Learning for EEG signal classification: a survey of state-of-the-art techniques and future directions

Frontiers in Human Neuroscience, Journal Year: 2024, Volume and Issue: 18

Published: July 10, 2024

This paper presents a systematic literature review, providing comprehensive taxonomy of Data Augmentation (DA), Transfer Learning (TL), and Self-Supervised (SSL) techniques within the context Few-Shot (FSL) for EEG signal classification. signals have shown significant potential in various paradigms, including Motor Imagery, Emotion Recognition, Visual Evoked Potentials, Steady-State Visually Rapid Serial Presentation, Event-Related Mental Workload. However, challenges such as limited labeled data, noise, inter/intra-subject variability impeded effectiveness traditional machine learning (ML) deep (DL) models. review methodically explores how FSL approaches, incorporating DA, TL, SSL, can address these enhance classification performance specific paradigms. It also delves into open research related to Specifically, examines identification DA strategies tailored creation TL architectures efficient knowledge transfer, formulation SSL methods unsupervised representation from data. Addressing is crucial enhancing efficacy robustness FSL-based By presenting structured discussing associated challenges, this offers valuable insights future investigations The findings aim guide inspire researchers, promoting advancements applying methodologies improved analysis real-world settings.

Language: Английский

---

Multi-layer prototype learning with Dirichlet mixup for open-set EEG recognition

Expert Systems with Applications, Journal Year: 2024, Volume and Issue: 266, P. 126047 - 126047

Published: Dec. 19, 2024

Language: Английский

---

Population-Based Evolutionary Search for Joint Hyperparameter and Architecture Optimization in Brain-Computer Interface

Published: Jan. 1, 2024

Language: Английский

---

Population-based evolutionary search for joint hyperparameter and architecture optimization in brain-computer interface

Expert Systems with Applications, Journal Year: 2024, Volume and Issue: unknown, P. 125832 - 125832

Published: Nov. 1, 2024

Language: Английский

---