BrainDistill: Implantable Motor Decoding with Task-Specific Knowledge Distillation

2026-01-24

AI Summary

This work addresses the challenge of deploying large-scale Transformer decoders in power-constrained implantable brain–computer interfaces (BCIs), despite their strong neural decoding performance. To this end, the authors propose BrainDistill, a framework that integrates an Implantable Neural Decoder (IND) with Task-Specific Knowledge Distillation (TSKD). The approach preserves critical decoding features through supervised projection and enables efficient integer-only inference via quantization-aware training combined with learned activation clipping ranges. Experimental results demonstrate that the IND outperforms existing decoders across multiple neural datasets, while TSKD significantly surpasses alternative distillation methods under few-shot calibration settings. Moreover, the quantized model incurs negligible performance degradation, thereby meeting the stringent power constraints of implantable BCI systems.

Technology Category

• Cognitive Modeling & Cognitive Systems: Neural Spike Coding
• Machine Learning: Hardware-aware ML
• Natural Language Processing: Learning & Optimization for NLP

Application Category

• User Modeling, Personalization and Recommendation: On-Device user modeling, personalization, and recommendation
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications

Abstract

Transformer-based neural decoders with large parameter counts, pre-trained on large-scale datasets, have recently outperformed classical machine learning models and small neural networks on brain-computer interface (BCI) tasks. However, their large parameter counts and high computational demands hinder deployment in power-constrained implantable systems. To address this challenge, we introduce BrainDistill, a novel implantable motor decoding pipeline that integrates an implantable neural decoder (IND) with a task-specific knowledge distillation (TSKD) framework. Unlike standard feature distillation methods that attempt to preserve teacher representations in full, TSKD explicitly prioritizes features critical for decoding through supervised projection. Across multiple neural datasets, IND consistently outperforms prior neural decoders on motor decoding tasks, while its TSKD-distilled variant further surpasses alternative distillation methods in few-shot calibration settings. Finally, we present a quantization-aware training scheme that enables integer-only inference with activation clipping ranges learned during training. The quantized IND enables deployment under the strict power constraints of implantable BCIs with minimal performance loss.

Problem

• implantable BCI
• motor decoding
• knowledge distillation
• computational efficiency
• power constraints

Innovation

• knowledge distillation
• implantable BCI
• motor decoding
• quantization-aware training
• neural decoder