TRACE: Contrastive learning for multi-trial time series data in neuroscience

Lisa Schmors, Dominic Gonschorek, Jan Niklas Böhm, Yongrong Qiu, Na Zhou, Dmitry Kobak, Andreas S. Tolias, Fabian H. Sinz, Jacob Reimer, Katrin Franke, Sebastian Damrich, Philipp Berens

Keywords: Neural cell types, Dimensionality reduction, Contrastive learning, Two-photon recordings, Time series data

TL;DR: TRACE is a contrastive learning framework that uses averaging of multi-trial neural activity to create interpretable 2D embeddings of large-scale neural recordings, revealing both continuous biological variation and discrete cell-type structures.

Abstract: Modern neural recording techniques such as two-photon imaging or Neuropixel probes allow to acquire vast time-series datasets with responses of hundreds or thousands of neurons. Contrastive learning is a powerful self-supervised framework for learning representations of complex datasets. Existing applications for neural time series rely on generic data augmentations and do not exploit the multi-trial data structure inherent in many neural datasets. Here we present TRACE, a new contrastive learning framework that averages across different subsets of trials to generate positive pairs. TRACE allows to directly learn a two-dimensional embedding, combining ideas from contrastive learning and neighbor embeddings. We show that TRACE outperforms other methods, resolving fine response differences in simulated data. Further, using in vivo recordings, we show that the representations learned by TRACE capture both biologically relevant continuous variation, cell-type-related cluster structure, and can assist data quality control.

Primary Area: Neuroscience and cognitive science (e.g., neural coding, brain-computer interfaces)

Submission Number: 12516