Shallow recurrent decoders for neural and behavioural dynamics.

Abstract

Machine learning algorithms are affording new opportunities for building bio-inspired and data-driven models characterizing neural activity. Critical to understanding decision-making and behaviour is quantifying the relationship between the activity of neuronal population codes and individual neurons. We leverage a SHallow REcurrent Decoder (SHRED) architecture for mapping the dynamics of population codes to individual neurons and other proxy measures of neural activity and behaviour. SHRED is constructed from a temporal sequence model, which encodes the temporal dynamics of limited sensor data in multiple scenarios, and a shallow decoder, which reconstructs the corresponding high-dimensional neuronal and/or behavioural states. It is a robust and flexible sensing strategy which allows for decoding the diversity of neural measurements with only a few sensor measurements. Thus, estimates of whole-brain activity, behaviour and individual neurons can be constructed with only a few neural time-series recordings. Several examples in this article further highlight the potential of leveraging non-invasive or minimally invasive measurements to estimate large-scale brain dynamics. We empirically demonstrate the capabilities of the method on a number of model organisms including Caenorhabditis elegans, mouse, zebrafish and human biolocomotion. This article is part of the discussion meeting issue 'Digital healthcare for the management of functional neurological disorders'.