Problem
Research questions and friction points this paper is trying to address.
intracortical motor decoding
cross-session
neural variability
brain-computer interface
Association profile conditioning in a set-temporal transformer for cross-session intracortical motor decoding
🤖 AI Summary
This study addresses the cross-session performance degradation in cortical motor decoders caused by variations in recorded neural units. To this end, we propose the Association Profile-conditioned Set Temporal Transformer (APST). This method innovatively introduces a closed-form computed four-dimensional association profile to guide the attention mechanism, integrating a set attention encoder with a causal Transformer architecture. APST enables rapid adaptation to novel sessions under frozen network weights without requiring parameter updates. Experimental evaluations on the DANDI688 dataset demonstrate that APST achieves velocity decoding R² scores of 0.78 and 0.81, significantly outperforming existing baseline methods while matching the performance of fine-tuned recurrent neural networks. These results establish APST as an efficient new paradigm for training-free neural decoding.
📝 Abstract
Intracortical motor decoders degrade across sessions because the set of recorded units changes and persisting units can alter how their firing relates to behavior. Most existing methods update network weights on each new session or rely on unlabeled activity, which does not directly reveal such changes. We present APST, an Association Profile-conditioned Set-Temporal transformer that adapts to new sessions with all network weights frozen. From a few labeled calibration trials, APST summarizes how each unit's firing relates to behavior in a four-dimensional association profile computed in closed form. The profiles condition a set-attention encoder that accepts any number and order of units, followed by a causal transformer for streaming decoding. On held-out DANDI688 sessions from two monkeys, APST reaches velocity $R^2$ of $0.78$ and $0.81$, versus $0.40$ and $0.58$ for a variant that uses neural activity alone, and matches or exceeds an RNN fine-tuned on the same trials. On FALCON private held-out evaluation, it attains $R^2$ of $0.65$, $0.42$, and $0.44$ on M1, M2, and H1.
Innovation
Methods, ideas, or system contributions that make the work stand out.
cross-session decoding
association profile conditioning
set-temporal transformer
frozen weights adaptation
intracortical motor decoding
🔎 Similar Papers
No similar papers found.
💼 Related Jobs
No related jobs found.
