[Paper Review] MetaSort: An Accelerated Approach for Non-uniform Compression and Few-shot Classification of Neural Spike Waveforms

Luca M. Meyer, Majid Zamani|arXiv (Cornell University)|Mar 8, 2026

Neural dynamics and brain function0 citations

TL;DR

MetaSort jointly tackles non-uniform compression and few-shot classification of neural spike waveforms by using an adaptive level crossing compression and a latent feature representation, enhanced by meta-transfer learning that leverages geometric data structure.

ABSTRACT

Many previous works in spike sorting study spike classification and compression independently. In this paper, a novel algorithm is proposed called MetaSort to address these two problems. To deal with compression, a novel adaptive level crossing algorithm is proposed to approximate spike shapes with high fidelity. Meanwhile, the latent feature representation is used to handle the classification problem. Besides, to guarantee MetaSort is robust and discriminative, the geometric information of data is exploited simultaneously in the proposed framework by meta-transfer learning. Empirical experiments with in-vivo spike data demonstrate that MetaSort delivers promising performance, highlighting its potential and motivating continued development toward an ultra-low-power, on-chip implementation.

Motivation & Objective

• Address the coupled problems of compressing neural spike waveforms with high fidelity and classifying them with few labeled examples.
• Develop an adaptive level crossing algorithm to efficiently approximate spike shapes for compression.
• Leverage a latent feature representation for classification.
• Incorporate geometric information through meta-transfer learning to enhance robustness and discriminability.
• Demonstrate potential for ultra-low-power, on-chip implementation.

Proposed method

• Introduce an adaptive level crossing algorithm for high-fidelity, non-uniform spike compression.
• Use latent feature representations to perform spike classification.
• Incorporate geometric information of data via meta-transfer learning to improve discrimination.
• Ensure robustness and discriminability of the combined compression-classification framework.
• Evaluate on in-vivo spike data to validate performance and feasibility for on-chip deployment.

Experimental results

Research questions

• RQ1 Can adaptive level crossing achieve high-fidelity non-uniform compression of neural spike waveforms?
• RQ2 Can latent feature representations enable effective few-shot classification of spikes?
• RQ3 Does meta-transfer learning leveraging geometric data improve robustness and discriminability of the method?
• RQ4 Is the approach suitable for ultra-low-power, on-chip implementation in practical settings?

Key findings

• Empirical experiments on in-vivo spike data show that MetaSort delivers promising performance in both compression and classification tasks.
• The framework exploits both the latent feature space and geometric information to enhance discrimination.
• MetaSort demonstrates potential for robust performance under few-shot conditions.
• The approach motivates the development of ultra-low-power, on-chip implementations for spike sorting tasks.