neuro2voc: Decoding Vocalizations from Neural Activity

TL;DR

This paper presents new methods for decoding zebra finch vocalizations from neural activity, including machine learning models and generative frameworks, advancing neural decoding techniques for complex motor outputs.

Contribution

Introduces novel neural decoding approaches using GPT2 tokenization, contrastive learning, and VAE, specifically applied to zebra finch neural data.

Findings

XGBoost with SHAP identified key neuronal interactions.

GPT2-based tokenization improved syllable decoding.

Contrastive learning-VAE generated spectrograms from neural data.

Abstract

Accurate decoding of neural spike trains and relating them to motor output is a challenging task due to the inherent sparsity and length in neural spikes and the complexity of brain circuits. This master project investigates experimental methods for decoding zebra finch motor outputs (in both discrete syllables and continuous spectrograms), from invasive neural recordings obtained from Neuropixels. There are three major achievements: (1) XGBoost with SHAP analysis trained on spike rates revealed neuronal interaction patterns crucial for syllable classification. (2) Novel method (tokenizing neural data with GPT2) and architecture (Mamba2) demonstrated potential for decoding of syllables using spikes. (3) A combined contrastive learning-VAE framework successfully generated spectrograms from binned neural data. This work establishes a promising foundation for neural decoding of complex motor outputs and offers several novel methodological approaches for processing sparse neural data.