Reproducibility of predictive networks for mouse visual cortex

TL;DR

This study examines the stability of neuronal embeddings derived from deep predictive models of mouse visual cortex activity, proposing regularization and pruning techniques to enhance interpretability and consistency.

Contribution

It introduces adaptive regularization and an iterative pruning strategy to improve the stability and interpretability of neuronal embeddings in deep models.

Findings

L1 regularization enhances structured embeddings.

Adaptive regularization improves clustering consistency.

Dimensionality reduction maintains performance while simplifying models.

Abstract

Deep predictive models of neuronal activity have recently enabled several new discoveries about the selectivity and invariance of neurons in the visual cortex. These models learn a shared set of nonlinear basis functions, which are linearly combined via a learned weight vector to represent a neuron's function. Such weight vectors, which can be thought as embeddings of neuronal function, have been proposed to define functional cell types via unsupervised clustering. However, as deep models are usually highly overparameterized, the learning problem is unlikely to have a unique solution, which raises the question if such embeddings can be used in a meaningful way for downstream analysis. In this paper, we investigate how stable neuronal embeddings are with respect to changes in model architecture and initialization. We find that $L_1$ regularization to be an important ingredient for structured embeddings and develop an adaptive regularization that adjusts the strength of regularization per neuron. This regularization improves both predictive performance and how consistently neuronal embeddings cluster across model fits compared to uniform regularization. To overcome overparametrization, we propose an iterative feature pruning strategy which reduces the dimensionality of performance-optimized models by half without loss of performance and improves the consistency of neuronal embeddings with respect to clustering neurons. This result suggests that to achieve an objective taxonomy of cell types or a compact representation of the functional landscape, we need novel architectures or learning techniques that improve identifiability. We will make our code available at publication time.