2026-09-09 00:00 Papers Foundations & Methods Translated from EN

Preprint: Federated NEXUS-MI Cuts Motor-Imagery BCI Backbone Traffic by About 42%

Summary Motor-imagery brain-computer interfaces vary widely between users and have little calibration data, and federated learning lets them share a model without uploading raw EEG. NEXUS-MI treats gateway synchronization as a joint learning-and-communication control problem: raw EEG and classifier heads stay local, while an edge coordinator maintains the shared backbone network. On the BCICIV-2a and OpenBMI datasets, communication-aware coordination cut server-to-client backbone traffic by about 42%, with small, implementation-dependent differences in cohort-average accuracy. Those averages masked individual vulnerability: on BCICIV-2a, losses relative to an ideal-link reference reached about 12 percentage points.

Why it matters Federated learning is usually pitched as a privacy fix, but this preprint treats it as a communication-scheduling problem in which gateway synchronization, stale-update admission and backbone download timing all shape individual accuracy; the 42% saving comes from backbone traffic rather than local EEG, and the more telling result is that cohort averages barely moved while some individuals lost up to 12 percentage points, so federated BCI evaluations that report only averages fall short and reliability must be judged per user.

BCIwiki (bciwiki.com) — A gateway-coordinated federated personalization framework cut server-to-client backbone traffic by about 42% in motor-imagery brain-computer interfaces. The framework, called NEXUS-MI, was posted to the arXiv preprint server on September 9, 2026 by Daniel Adu Worae and Aarthy Nagarajan and has not been peer reviewed. It treats synchronization as a coupled learning-and-communication control problem: raw EEG and classifier heads stay local while an edge coordinator maintains the shared backbone, so raw EEG is never centralized.

The authors evaluated NEXUS-MI through offline replay using BCI Competition IV Dataset 2a (BCICIV-2a; 9 subjects, 4 classes) and OpenBMI (54 subjects, 2 classes). Session 1 supports backbone learning, while Session 2 provides limited-calibration personalization and held-out testing. Beyond an ideal-link reference, 6 heterogeneous-link policies characterize gateway participation, buffering, stale-update admission and backbone-download control. The principal comparison holds delayed-update handling fixed while contrasting non-adaptive and communication-aware synchronization; paired subject-level comparisons use Holm adjustment, and robustness across 5 matched realizations is assessed by hierarchical bootstrap.

Communication-aware coordination reduced server-to-client backbone traffic by approximately 42% on both datasets, while cohort-level accuracy differences were small and realization-dependent. Cohort averages also concealed subject-level vulnerability, with losses reaching approximately 12 percentage points on BCICIV-2a relative to the ideal-link reference. The authors conclude that gateway synchronization should be an explicit design variable in federated motor-imagery personalization, motivating joint evaluation of personalized accuracy, communication cost, update freshness and subject-level reliability.

Compiled by BCIwiki from public sources

Sources · 1

arxiv.org 2026-09-09

EEG Motor Decoding Non-invasive BCI

Read original ↗

Suggest a correction Revisions · none 中 / EN