Temporally Local Weighting-Based Phase-Locked Time-Shift Data Augmentation Method for Fast-Calibration SSVEP-BCI DOI Creative Commons
Jiayang Huang, Yidan Lv, Zhiqiang Zhang

et al.

IEEE Transactions on Neural Systems and Rehabilitation Engineering, Journal Year: 2024, Volume and Issue: 32, P. 2376 - 2387

Published: Jan. 1, 2024

Various training-based spatial filtering methods have been proposed to decode steady-state visual evoked potentials (SSVEPs) efficiently. However, these require extensive calibration data obtain valid filters and temporal templates. The time-consuming collection process would reduce the practicality of SSVEP-based brain-computer interfaces (BCIs). Therefore, we propose a temporally local weighting-based phase-locked time-shift (TLW-PLTS) augmentation method augment training for calculating In this method, sliding window strategy using SSVEP response period as step is generate augmented data, time filter which maximises covariance between original template signal sine-cosine reference used suppress noise in data. For performance evaluation, TLW-PLTS was incorporated with state-of-the-art calculate classification accuracies information transfer rates (ITRs) three datasets. Compared other methods, demonstrates superior decoding fewer promising development fast-calibration BCIs.

Language: Английский

Representative-Based Cold Start for Adaptive SSVEP-BCI DOI Creative Commons
Nanlin Shi, Xiang Li, Bingchuan Liu

et al.

IEEE Transactions on Neural Systems and Rehabilitation Engineering, Journal Year: 2023, Volume and Issue: 31, P. 1521 - 1531

Published: Jan. 1, 2023

The tradeoff between calibration effort and model performance still hinders the user experience for steady-state visual evoked brain-computer interfaces (SSVEP-BCI). To address this issue improve generalizability, work investigated adaptation from cross-dataset to avoid training process, while maintaining high prediction ability.When a new subject enrolls, group of user-independent (UI) models is recommended as representative multi-source data pool. then augmented with online transfer learning techniques based on user-dependent (UD) data. proposed method validated both offline (N=55) (N=12) experiments.Compared UD adaptation, relieved approximately 160 trials efforts user. In experiment, time window decreased 2 s 0.56±0.2 s, accuracy 0.89-0.96. Finally, achieved average information rate (ITR) 243.49 bits/min, which highest ITR ever reported in complete calibration-free setting. results result were consistent experiment.Representatives can be even cross-subject/device/session situation. With help represented UI data, achieve sustained without process.This provides an adaptive approach transferable SSVEP-BCIs, enabling more generalized, plug-and-play high-performance BCI free calibrations.

Language: Английский

Citations

13
Cross-Subject Transfer Method Based on Domain Generalization for Facilitating Calibration of SSVEP-Based BCIs DOI Creative Commons
Jiayang Huang, Zhiqiang Zhang, Bang Xiong

et al.

IEEE Transactions on Neural Systems and Rehabilitation Engineering, Journal Year: 2023, Volume and Issue: 31, P. 3307 - 3319

Published: Jan. 1, 2023

In steady-state visual evoked potential (SSVEP)-based brain-computer interfaces (BCIs), various spatial filtering methods based on individual calibration data have been proposed to alleviate the interference of spontaneous activities in SSVEP signals for enhancing detection performance. However, time-consuming session would increase fatigue subjects and reduce usability BCI system. The key idea this study is propose a cross-subject transfer method domain generalization, which transfers domain-invariant filters templates learned from source target subject with no access EEG subject. transferred are obtained by maximizing intra- inter-subject correlations using corresponding its neighboring stimuli. For subject, four types correlation coefficients calculated construct feature vector. Experimental results estimated three datasets show that improves performance compared state-of-art methods. satisfactory demonstrate provides an effective learning strategy requiring tedious collection process new users, holding promoting practical applications SSVEP-based BCI.

Language: Английский

Citations

12
A Capsule Decision Neural Network Based on Transfer Learning for EEG Signal Classification DOI Creative Commons
Wei Zhang,

Xianlun Tang,

Xiaoyuan Dang

et al.

Biomimetics, Journal Year: 2025, Volume and Issue: 10(4), P. 225 - 225

Published: April 4, 2025

Transfer learning is the act of using data or knowledge in a problem to help solve different but related problems. In brain computer interface (BCI), it important deal with individual differences between topics and/or tasks. A kind capsule decision neural network (CDNN) based on transfer proposed. order feature distortion caused by EEG extraction algorithm, deep was constructed. The architecture includes multiple primary capsules form hidden layer, and connection advanced determined routing algorithm. Unlike dynamic algorithm that iteratively calculates similarity capsules, computes relationship each shallow layers probabilistic manner. At same time, distribution covariance matrix aligned Riemann space, regional adaptive method further introduced improve independent decoding ability for subject’s signals. Experiments two motor imagery datasets show CDNN outperforms several most methods.

Language: Английский

Citations

0
A dual alignment-based multi-source domain adaptation framework for motor imagery EEG classification DOI Open Access
Dongqin Xu, Mingai Li

Applied Intelligence, Journal Year: 2022, Volume and Issue: 53(9), P. 10766 - 10788

Published: Aug. 25, 2022

Language: Английский

Citations

17
Compact Artificial Neural Network Based on Task Attention for Individual SSVEP Recognition With Less Calibration DOI Creative Commons
Ze Wang, Chi Man Wong, Boyu Wang

et al.

IEEE Transactions on Neural Systems and Rehabilitation Engineering, Journal Year: 2023, Volume and Issue: 31, P. 2525 - 2534

Published: Jan. 1, 2023

Objective: Recently, artificial neural networks (ANNs) have been proven effective and promising for the steady-state visual evoked potential (SSVEP) target recognition. Nevertheless, they usually lots of trainable parameters thus require a significant amount calibration data, which becomes major obstacle due to costly EEG collection procedures. This paper aims design compact network that can avoid over-fitting ANNs in individual SSVEP Method: study integrates prior knowledge recognition tasks into attention design. First, benefiting from high model interpretability mechanism, layer is applied convert operations conventional spatial filtering algorithms ANN structure, reduces connections between layers. Then, signal models common weights shared across stimuli are adopted constraints, further condenses parameters. Results: A simulation on two widely-used datasets demonstrates proposed structure with constraints effectively eliminates redundant Compared existing prominent deep (DNN)-based correlation analysis (CA)-based algorithms, method by more than ${90}\%$ notation="LaTeX">${80}\%$ respectively, boosts performance at least notation="LaTeX">${57}\%$ notation="LaTeX">${7}\%$ respectively. Conclusion: Incorporating task make it efficient. The has less requires performance.

Language: Английский

Citations

10
Facilitating applications of SSVEP-BCI by effective Cross-Subject knowledge transfer DOI Creative Commons
Hui Li, Guanghua Xu, Chenghang Du

et al.

Expert Systems with Applications, Journal Year: 2024, Volume and Issue: 249, P. 123492 - 123492

Published: Feb. 18, 2024

In steady-state visual evoked potential (SSVEP)-based brain-computer interface (BCI), improving the recognition performance for new subjects without calibration data is key challenge practical application. Unsupervised transfer learning an effective way to overcome it. However, existing studies focus solely on what transfer, rather than how effectively resulting in unsatisfactory effectiveness or even negative transfer. this study, innovative unsupervised cross-subject method SSVEP-BCI was proposed, named SUTL. It involves that subject transferability estimation (STE) and a multi-domain alignment were proposed alleviate interference of differences SSVEP signal distribution among subjects. STE screens appropriate transferable from source pool, while domain directly makes all more similar. Then, SUTL sufficiently exploits information selected subjects, transferring both generalization knowledge subject-specific boost subject. The evaluated two public datasets (benchmark dataset BETA dataset) with 40 classes, extensive experimental results reveal markedly boosts dramatically outperforms state-of-art methods. significantly enhances facilitates its

Language: Английский

Citations

3
SSVEP-DAN: Cross-Domain Data Alignment for SSVEP-based Brain-Computer Interfaces DOI Creative Commons

Sung-Yu Chen,

Chi-Min Chang, Kuan-Jung Chiang

et al.

IEEE Transactions on Neural Systems and Rehabilitation Engineering, Journal Year: 2024, Volume and Issue: 32, P. 2027 - 2037

Published: Jan. 1, 2024

Steady-state visual-evoked potential (SSVEP)-based brain-computer interfaces (BCIs) offer a non-invasive means of communication through high-speed speller systems. However, their efficiency is highly dependent on individual training data acquired during time-consuming calibration sessions. To address the challenge insufficiency in SSVEP-based BCIs, we introduce SSVEP-DAN, first dedicated neural network model designed to align SSVEP across different domains, encompassing various sessions, subjects, or devices. Our experimental results demonstrate ability SSVEP-DAN transform existing source into supplementary data. This significant improvement decoding accuracy while reducing time. We envision playing crucial role future applications high-performance BCIs. The code for this work available at: https://github.com/CECNL/SSVEP-DAN.

Language: Английский

Citations

3
Enhancing SSVEP Identification With Less Individual Calibration Data Using Periodically Repeated Component Analysis DOI
Yufeng Ke, Shuang Liu, Dong Ming

et al.

IEEE Transactions on Biomedical Engineering, Journal Year: 2023, Volume and Issue: 71(4), P. 1319 - 1331

Published: Nov. 16, 2023

Spatial filtering and template matching-based steady-state visually evoked potentials (SSVEP) identification methods usually underperform in SSVEP with small-sample-size calibration data, especially when a single trial of data is available for each stimulation frequency.

Language: Английский

Citations

8
Unsupervised Domain Adaptation via Spatial Pattern Alignment for VEP-Based Identity Recognition DOI
Hongze Zhao, Yijun Wang, Xiaorong Gao

et al.

IEEE Internet of Things Journal, Journal Year: 2024, Volume and Issue: 11(20), P. 33722 - 33733

Published: July 25, 2024

Language: Английский

Citations

2
OS-SSVEP: One-shot SSVEP classification DOI
Yang Deng, Zhiwei Ji, Yijun Wang

et al.

Neural Networks, Journal Year: 2024, Volume and Issue: 180, P. 106734 - 106734

Published: Sept. 25, 2024

Language: Английский

Citations

2