Newest papers

Scaling Parameter-Efficient Fine-Tuning for Brain Encoding

Zhuofan Josh Ying, Guanglong Sun, Nolan Ho, Kieran Liu, Rui Yang, Andrew Chen, Nikolaus Kriegeskorte, PONY MA, Hossein Adeli · 2026 · arXiv (Cornell University)

DOI: 10.48550/arxiv.2610.26802 · cited by 0 works (OpenAlex) · open-access PDF

Portuguese

Este conjunto de modelos de codificação do cérebro visa prever respostas neurais a estímulos. A abordagem mais amplamente usada combina leituras de costas pré-treinadas congeladas. Usando dados neurais para treinar modelos grandes a partir do zero ou perfeitamente ajustá-los após o pré-treinamento de tarefas provou ser difícil porque os dados neurais são caros e limitados. Mostramos que o ajustamento fino de parâmetros eficiente em uma única passagem (PEFT) nos permite adaptar de forma eficaz os costos de extração de características muito grandes para a codificação do cérebro público com um pequeno número de parâmetros treináveis. Para torná-los computacionalmente práticos, exploramos a estrutura causal do backbone para prever múltiplos pontos de tempo em uma única passagem em vez de usar janelas

Machine translation (Meta M2M100); the English original follows.

Abstract

Brain encoding models aim to predict neural responses to stimuli. The most widely used approach fits readouts from frozen pretrained backbones. Using neural data to train large models from scratch or fully fine-tune them after task-pretraining has proven difficult because neural data is costly and limited. We show that parameter-efficient fine-tuning (PEFT) enables us to effectively adapt very large pretrained backbones for brain encoding with a small number of trainable parameters. To make it computationally practical, we exploit the backbone's causal structure to predict multiple timepoints in a single pass instead of using sliding windows, significantly reducing feature-extraction cost. This enables efficient fine-tuning of multimodal backbones with up to 104B parameters, the largest model adapted for brain encoding to our knowledge. Trained with a low-rank adapter (LoRA) and a lightweight readout, our approach consistently improves brain encoding performance across nine diverse multimodal backbones. On the same Qwen3-Omni backbone as the prior state-of-the-art (SoTA), our method BrainLoRA achieves the best published performance on the Algonauts 2025 Challenge. We also show the adapted multimodal backbones remain interpretable: modality attribution recovers the canonical sensory topography that prior work often obtains by construction from separate unimodal encoders, and neural adaptation increases the contribution of audio inputs and cross-modal mixing within the backbone. To scale out the number of adapted models, we develop a multi-LoRA kernel that accelerates joint training of multiple LoRAs on a shared backbone. This enables an ensemble of 200+ adapted models that ranks first on the public Algonauts post-challenge leaderboard at the time of writing. Together, our results highlight PEFT as an efficient and powerful paradigm for brain encoding.

Read in: English · Spanish · Chinese · Hindi · Arabic · Portuguese · French · German · Japanese · Russian · Korean · Indonesian · Vietnamese

Topics: Computer science, Encoding (memory), Artificial intelligence, Exploit, Artificial neural network, Modality (human–computer interaction)

Sources: OpenAlex (CC0), arXiv, Crossref. Agent Exchange