DeepVisualInsight: Time-Travelling Visualization for Spatio-Temporal Causality of Deep Classification Training

TL;DR

DeepVisualInsight (DVI) is a visualization tool that reveals the spatio-temporal causality in deep learning training, helping to understand and debug model behavior during training by visualizing how training strategies influence learned representations over time.

Contribution

We introduce DVI, a novel visualization approach that captures and displays the spatio-temporal causality in deep classifier training, enabling better analysis and debugging of training processes.

Findings

DVI outperforms baseline methods in visualization quality and efficiency.

DVI effectively reflects different training scenarios and strategies.

The tool aids in diagnosing training issues and understanding model evolution.

Abstract

Understanding how the predictions of deep learning models are formed during the training process is crucial to improve model performance and fix model defects, especially when we need to investigate nontrivial training strategies such as active learning, and track the root cause of unexpected training results such as performance degeneration. In this work, we propose a time-travelling visual solution DeepVisualInsight (DVI), aiming to manifest the spatio-temporal causality while training a deep learning image classifier. The spatio-temporal causality demonstrates how the gradient-descent algorithm and various training data sampling techniques can influence and reshape the layout of learnt input representation and the classification boundaries in consecutive epochs. Such causality allows us to observe and analyze the whole learning process in the visible low dimensional space. Technically, we propose four spatial and temporal properties and design our visualization solution to satisfy them. These properties preserve the most important information when inverse-)projecting input samples between the visible low-dimensional and the invisible high-dimensional space, for causal analyses. Our extensive experiments show that, comparing to baseline approaches, we achieve the best visualization performance regarding the spatial/temporal properties and visualization efficiency. Moreover, our case study shows that our visual solution can well reflect the characteristics of various training scenarios, showing good potential of DVI as a debugging tool for analyzing deep learning training processes.