# Learning aligned EEG representations with subject-specific encoders

arXiv: https://arxiv.org/abs/2606.16462 · HTML: https://mlsanity.com/paper/2606.16462/

## Key facts

- Cited by 1 papers in the map (data through 14 Oct 2026).
- Builds on 22 of its 54 references that are in the map.
- First posted on arXiv 15 Jun 2026.

## Builds on (most cited first, 22 of 22)

- [A Comprehensive Survey on Transfer Learning](https://mlsanity.com/paper/1911.02685/) (2019)
- [EEGNet: A Compact Convolutional Network for EEG-based Brain-Computer Interfaces](https://mlsanity.com/paper/1611.08024/) (2016)
- [Large Brain Model for Learning Generic Representations with Tremendous EEG Data in BCI](https://mlsanity.com/paper/2405.18765/) (2024)
- [Characterizing and Avoiding Negative Transfer](https://mlsanity.com/paper/1811.09751/) (2018)
- [Deep learning-based electroencephalography analysis: a systematic review](https://mlsanity.com/paper/1901.05498/) (2019)
- [Decoding speech perception from non-invasive brain recordings](https://mlsanity.com/paper/2208.12266/) (2022)
- [Transfer Learning for Brain-Computer Interfaces: A Euclidean Space Data Alignment Approach](https://mlsanity.com/paper/1808.05464/) (2018)
- [Self-supervised representation learning from electroencephalography signals](https://mlsanity.com/paper/1911.05419/) (2019)
- [Scaling laws for decoding images from brain activity](https://mlsanity.com/paper/2501.15322/) (2025)
- [Toward reliable signals decoding for electroencephalogram: A benchmark study to EEGNeX](https://mlsanity.com/paper/2207.12369/) (2022)
- [Decoding individual words from non-invasive brain recordings across 723 participants](https://mlsanity.com/paper/2412.17829/) (2024)
- [Inter-subject Deep Transfer Learning for Motor Imagery EEG Decoding](https://mlsanity.com/paper/2103.05351/) (2021)
- [Calibration-free online test-time adaptation for electroencephalography motor imagery decoding](https://mlsanity.com/paper/2311.18520/) (2023)
- [A Systematic Evaluation of Euclidean Alignment with Deep Learning for EEG Decoding](https://mlsanity.com/paper/2401.10746/) (2024)
- [EEG motor imagery decoding: A framework for comparative analysis with channel attention mechanisms](https://mlsanity.com/paper/2310.11198/) (2023)
- [Evaluating the structure of cognitive tasks with transfer learning](https://mlsanity.com/paper/2308.02408/) (2023)
- [Minimizing subject-dependent calibration for BCI with Riemannian transfer learning](https://mlsanity.com/paper/2111.12071/) (2021)
- [EEG Decoding for Datasets with Heterogenous Electrode Configurations using Transfer Learning Graph Neural Networks](https://mlsanity.com/paper/2306.13109/) (2023)
- [EEGminer: Discovering Interpretable Features of Brain Activity with Learnable Filters](https://mlsanity.com/paper/2110.10009/) (2021)
- [Transfer Learning between Motor Imagery Datasets using Deep Learning -- Validation of Framework and Comparison of Datasets](https://mlsanity.com/paper/2311.16109/) (2023)
- [Mitigating Subject Dependency in EEG Decoding with Subject-Specific Low-Rank Adapters](https://mlsanity.com/paper/2510.08059/) (2025)
- [Physics-informed and Unsupervised Riemannian Domain Adaptation for Machine Learning on Heterogeneous EEG Datasets](https://mlsanity.com/paper/2403.15415/) (2024)

## Built on by (most cited first, 1 of 1)

- [Separating personal from population gains when calibrating EEG foundation models for new users](https://mlsanity.com/paper/2609.34801/) (2026)

## Abstract (arXiv)

Cross-subject EEG decoding promises more training data, but it also exposes neural networks to strong inter-subject distribution shifts. We study whether task supervision and architecture alone can learn subject-aligned representations. We replace a shared EEG encoder with subject-specific encoders followed by a common classifier, and compare this hybrid model with standard EEGNet, AttentionBaseNet, and CTNet baselines with Euclidean Alignment (EA) on three motor-imagery datasets and one motor-execution dataset. EA improves shared encoders by recentering subject covariances, whereas the hybrid encoder reduces reliance on EA: removing EA has little effect on validation-loss dynamics or latent-space organization, and both hybrid variants consistently outperform non-aligned shared baselines. Subject-specific heads increase class distinctiveness and place each subject close to its own latent manifold while improving within-subject class separation. However, on cross-subject classification, subject-specific heads hinder direct parameter transfer to unseen subjects, motivating quantitative head selection and a brief calibration session. Although decoding gains depend on the dataset and backbone, our main findings concern that the sole use of architecture pressure promotes representation learning and alignment in a direction complementary to domain adaptation methods such as Euclidean Alignment. A per-subject low-rank adapter of only 2Cr parameters recover the full encoder's accuracy across five backbones and ranks r=1 to 16, so the per-subject module can be compressed by two to three orders of magnitude.