Multi-Feature Riemannian Hypergraph for Online Test-Time Adaptation of Motor Imagery Brain-Computer Interface

Siqi Li; Zhi Li; Tong Liu; Shuai Zhang; Yanfei Jia; Zhiqiang Yi; Jue Xie; Ni Ji

cs.LGAug 17, 2026

Multi-Feature Riemannian Hypergraph for Online Test-Time Adaptation of Motor Imagery Brain-Computer Interface

Authors: Siqi Li, Zhi Li, Tong Liu, Shuai Zhang, Yanfei Jia, Zhiqiang Yi, Jue Xie, Ni Ji

Organizations: Peking University · Chinese Institute for Brain Research, Beijing · NeuCyber Neurotech · Beijing Medical University ·

Abstract

In clinical motor imagery brain-computer interface (MI-BCI) decoding, cross-day transferability and online operation remain two critical challenges. Hypergraphs can improve transferability by capturing higher-order sample relationships, yet existing hypergraph-based methods for online emotion recognition neglect the cross-day benefits of Riemannian geometry widely adopted in EEG transfer learning. To bridge this gap, we propose the Multi-feature Riemannian Hypergraph (MRieHy), a framework tailored for online test-time adaptation in MI-BCI decoding that leverages Riemannian geometry to strengthen cross-day transferability. MRieHy first computes Riemannian means of covariance matrices from cross-day training data to align multi-day distributions. It then constructs a hypergraph over covariance matrices using Riemannian distance, complemented by a second hypergraph over deep features built with cosine similarity. The two hypergraphs are fused via adaptively learned combination weights, jointly optimized with the label projection matrices. During online testing, MRieHy maintains a first-in-first-out buffer of recent samples, performs Riemannian alignment on the buffered data, and decodes with the learned hypergraph. Extensive experiments on a private four-class ECoG dataset and two public four-class EEG datasets validate that MRieHy achieves notable performance gains over state-of-the-art baselines.

Figures & tables

Figure 1: Schematic illustration of the Multi-feature Riemannian Hypergraph (MRieHy) framework.

Table 1: Comparison between MRieHy and the baseline methods on three datasets

Align	Co	Sim	Deep	ECoG128	BCI Comp	Stieger2021
Rie	✓	Cos	✓	59.9 $\pm$ 2.2	45.1 $\pm$ 1.2	52.0 $\pm$ 0.3
Rie	✓	TanCos	✓	64.1 $\pm$ 1.0	71.2 $\pm$ 0.9	54.1 $\pm$ 0.3
Rie	✓	GauRie	✓	63.3 $\pm$ 0.9	68.8 $\pm$ 1.5	52.5 $\pm$ 0.4
Rie	✓	EuDM	✓	59.6 $\pm$ 2.1	44.8 $\pm$ 1.1	52.0 $\pm$ 0.2
Rie	✓	RieDM	✓	63.3 $\pm$ 0.9	68.8 $\pm$ 1.5	52.5 $\pm$ 0.3
Rie	✓	TanDM	✓	64.1 $\pm$ 0.9	71.2 $\pm$ 0.9	54.1 $\pm$ 0.4

Table 2: Ablation study of different alignment methods, features (Co&Deep), and similarity computation (Sim) for covariance matrix

Panel of Figure 2 — Figure 2: Percentage of distinct days per hyperedge in the covariance - based hypergraph component

Panel of Figure 3 — Figure 3: Neighbor vertices and incident hyperedges for a sample from Stieger2021 dataset Subject 1 in the covariance-based hypergraph component built with cosine similarity (left) and with TanDM (right). Numbers indicate the day each vertex is from.

Panel of Figure 4 — Figure 4: Impact of training day count on MRieHy and component hypergraph performance.

Panel of Figure 5 — Figure 5: Effect of $k$ , $\eta$ , and buffer size on MRieHy’s average test accuracy across 9 subjects on BCI Competition IV 2a (upper row) and 7 subjects on Stieger2021 (bottom row).

Appendix figures & tables1 asset

Supplementary material from the paper’s appendix.

Appendix

Panel of Figure 8 — Figure D.1: Percentage of distinct days per hyperedge in the covariance - based hypergraph component

Cross-Session EEG Decoding

See all 9 papers

Time Window Optimization for Motor Imagery Classification via Temporal Structure Embedded Reinforcement Learning
Oct 12, 2026Huan Luo, Na Lu, Xiaopeng Wang +1MI Classification EEG Decoding
School of Future Technology, Xi’an Jiaotong University, Xi’an, 710049, China · School of Automation Science and Engineering, Xi’an Jiaotong University, Xi’an, 710049, China
ATCNet-CIAM for Multi-Session Motor Imagery EEG Signal Classification
Jul 26, 2026Le Huu Son Hai, Nguyen Chi Hai, Truong Viet Vu +3MI Classification Brain-Computer Interfaces
Faculty of Information Technology, Van Lang School of Technology, Van Lang University, Ho Chi Minh City 70000, Vietnam
Memory-Augmented Online Test-Time Adaptation for EEG Motor Imagery via Perturbation Consistency and Confidence Calibration
Oct 21, 2026Kunkun Zhang, Kehao Wu, Zhitao Chen +2EEG Decoding Brain-Computer Interfaces
College of Computer Science and Technology, Zhejiang University of Technology, Hangzhou 310023, China · Zhejiang Key Laboratory of Visual Information Intelligent Processing, Hangzhou 310023, China

Oct 12, 2026J/K move · Enter open · S save

Time Window Optimization for Motor Imagery Classification via Temporal Structure Embedded Reinforcement Learning

Huan Luo, Na Lu, Xiaopeng Wang, Ziyang Shang

School of Future Technology, Xi’an Jiaotong University, Xi’an, 710049, China · School of Automation Science and Engineering, Xi’an Jiaotong University, Xi’an, 710049, China

MI Classification EEG Decoding

Open paper AskPDF arXiv

Online motor imagery-based brain-computer interfaces (MI-BCIs) typically decode continuous EEG using fixed-length sliding windows that may be misaligned with neural events. Existing window optimization methods often evaluate windows independently, overlooking temporal dependencies and potentially producing fragmented or physiologically implausible selections. We propose Temporal Structure Embedded Reinforcement Learning (TSE-RL), a framework that incorporates prior knowledge of MI event structure into a constrained Markov decision process for adaptive window optimization. First, TSE-RL assigns each window an explicit semantic role through Discard, Onset, Sustain, or Offset actions, extending binary retain/discard decisions to event-level alignment. Second, a finite-state machine masks invalid actions to enforce the temporal order of MI phases and promote physiologically plausible window sequences. Third, a Transformer-based history encoder captures dependencies across windows, improving training stability and computational efficiency compared with recurrent architectures. TSE-RL is plug-and-play and can be integrated with diverse MI classifiers. Experiments on multiple benchmark datasets demonstrate consistent improvements in decoding performance in both cross-session and cross-subject settings. Visualization analyses further suggest that the learned window selection policy aligns with neurophysiological principles.

Abstract

Cross-Session EEG Decoding

Explore similar work

Time Window Optimization for Motor Imagery Classification via Temporal Structure Embedded Reinforcement Learning

ATCNet-CIAM for Multi-Session Motor Imagery EEG Signal Classification

Memory-Augmented Online Test-Time Adaptation for EEG Motor Imagery via Perturbation Consistency and Confidence Calibration