mlsanityData to 14 Oct 2026

LUNA: Efficient and Topology-Agnostic Foundation Model for EEG Signal Analysis

arXiv 2510.22257 ↗First posted 25 Oct 2025cs.LG cs.AIcited by 26

Not on the map. Why?

Abstract (arXiv)

Electroencephalography (EEG) offers a non-invasive lens into human brain activity, but building large-scale models is hampered by topological heterogeneity: each public EEG data defines its own electrode layout, limiting generalization. We introduce LUNA (Latent Unified Network Architecture), a self-supervised foundation model that reconciles disparate electrode geometries while scaling linearly -- not quadratically -- with channel count. LUNA compresses multi-channel EEG into a fixed-size, topology-agnostic latent space via learned queries and cross-attention. Downstream transformer blocks then operate exclusively on this latent representation using patch-wise temporal self-attention, decoupling computation from electrode count. Pre-trained on TUEG and Siena (over 21,000 hours of raw EEG across diverse montages) using a masked-patch reconstruction objective, LUNA transfers effectively to four downstream tasks: abnormality detection, artifact rejection, slowing classification, and emotion recognition. It demonstrates highly competitive performance across several benchmarks, achieving state-of-the-art results on TUAR and TUSL, e.g., 0.921 AUROC on TUAR, while reducing FLOPs by 300x and trimming GPU memory use by up to 10x. Critically, these gains are consistent across all evaluated electrode configurations. Code is available at https://github.com/pulp-bio/BioFoundation

Key facts. Cited by 26 papers in the map (data through 14 Oct 2026). Builds on 22 of its 39 references that are in the map. First posted on arXiv 25 Oct 2025.

Builds on

22 of 39 references.

References · 22

  1. 2021RoFormer: Enhanced Transformer with Rotary Position Embeddingcited by 4,703
  2. 2016Gaussian Error Linear Units (GELUs)cited by 3,502
  3. 2018Group Normalizationcited by 1,524
  4. 2021Perceiver IO: A General Architecture for Structured Inputs & Outputscited by 629
  5. 2016EEGNet: A Compact Convolutional Network for EEG-based Brain-Computer Interfacescited by 628
  6. 2022Transformers in Time Series: A Surveycited by 497
  7. 2024Large Brain Model for Learning Generic Representations with Tremendous EEG Data in BCIcited by 327
  8. 2021BENDR: using transformers and a contrastive self-supervised learning task to learn from massive amounts of EEG datacited by 199
  9. 2024CBraMod: A Criss-Cross Brain Foundation Model for EEG Decodingcited by 199
  10. 2021Transformer-based Spatial-Temporal Feature Learning for EEG Decodingcited by 97
  11. 2021Self-Supervised Graph Neural Networks for Improved Electroencephalographic Seizure Analysiscited by 76
  12. 2022Transformer Convolutional Neural Networks for Automated Artifact Detection in Scalp EEGcited by 55
10 more
  1. 2024EEGFormer: Towards Transferable and Interpretable Large-Scale EEG Foundation Modelcited by 52
  2. 2024EEG2Rep: Enhancing Self-supervised EEG Representation Through Informative Masked Inputscited by 50
  3. 2021Self-supervised EEG Representation Learning for Automatic Sleep Stagingcited by 32
  4. 2022Modeling Multivariate Biosignals With Graph Neural Networks and Structured State Space Modelscited by 29
  5. 2024Population Transformer: Learning Population-level Representations of Neural Activitycited by 26
  6. 2024Neural decoding from stereotactic EEG: accounting for electrode variability across subjectscited by 17
  7. 2020Learning from Heterogeneous EEG Signals with Differentiable Channel Reorderingcited by 16
  8. 2025FEMBA: Efficient and Scalable EEG Analysis with a Bidirectional Mamba Foundation Modelcited by 16
  9. 2025CEReBrO: Compact Encoder for Representations of Brain Oscillations Using Efficient Alternating Attentioncited by 14
  10. 2022Energy-Efficient Tree-Based EEG Artifact Detectioncited by 6

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2025: 2; 2026: 24.

26 papers.

Sort:

By area:

  1. 2026NeuralBench: A Unifying Framework to Benchmark NeuroAI Modelscited by 16
  2. 2026DeeperBrain: A Neuro-Grounded EEG Foundation Model Towards Universal BCIcited by 9
  3. 2025TinyMyo: a Tiny Foundation Model for Flexible EMG Signal Processing at the Edgecited by 7
  4. 2026Neural Signals Generate Clinical Notes in the Wildcited by 4
  5. 2026NeuroWeaver: An Autonomous Evolutionary Agent for Exploring the Programmatic Space of EEG Analysis Pipelinescited by 4
  6. 2026LuMamba: Latent Unified Mamba for Electrode Topology-Invariant and Efficient EEG Modelingcited by 4
  7. 2025Neural Brain Fields: A NeRF-Inspired Approach for Generating Nonexistent EEG Electrodescited by 2
  8. 2026FEMBA on the Edge: Physiologically-Aware Pre-Training, Quantization, and Deployment of a Bidirectional Mamba EEG Foundation Model on an Ultra-low Power Microcontrollercited by 1
  9. 2026NAKUL-Med: Spectral-Graph State Space Models with Dynamics Kernels for Medical Signalscited by 1
  10. 2026Aperiodic and Low-Frequency Spectral Bias in Reconstruction based EEG Foundation Modelscited by 1
  11. 2026A Multi-dimensional Framework for Evaluating Generalization in EEG Foundation Modelscited by 1
  12. 2026CaMBRAIN: Real-time, Continuous EEG Inference with Causal State Space Modelscited by 1
14 more
  1. 2026Stacked LoRA for Subject-Adaptive EEG Foundation Models in Motor Imagery Decodingcited by 1
  2. 2026PanLUNA: An Efficient and Robust Query-Unified Multimodal Model for Edge Biosignal Intelligencecited by 0
  3. 2026Evaluation of EEG Foundation Models for Event-Based Burst-Suppression Detection in ICUcited by 0
  4. 2026CoCoT-EEG: Contrastive-Pretrained Multiscale Convolutional Transformer for EEG Decodingcited by 0
  5. 2026Differentiable Logic Gate Networks for Low-Latency EEG Classification on Edge Devicescited by 0
  6. 2026S-CEReBrO: Breaking the Memory Barrier in Continuous EEG Monitoringcited by 0
  7. 2026EEG-PRISM: Physiologically-Grounded Interpretability of Predictions by EEG Foundation Modelscited by 0
  8. 2026ReMAP: Self-supervised learning to unveil brain representations and vulnerabilitycited by 0
  9. 2026Neural State Prediction: Obstructing Shortcut Learning in EEG Foundation Modelscited by 0
  10. 2026What masking geometry works best for EEG foundation models?cited by 0
  11. 2026Representational and Functional Robustness to Electrode Montages in EEG Foundation Modelscited by 0
  12. 2026NeurDuo-EEG: A Long-Sequence EEG Foundation Model with Persistent State and Explicit Memorycited by 0
  13. 2026Lightweight Semantic EEG Foundation Model for Frozen Cross-Disorder Transfercited by 0
  14. 2026Brain2Face: A Study on the Relation of EEG Signals and Spatio-temporal Head Avatarscited by 0