Critical Percolation as a Framework to Analyze the Training of Deep Networks

TL;DR

This paper models the topology of maze data using a CNN-based architecture, analyzing the cost function landscape and learning obstacles, providing insights into deep network training on graph-structured data.

Contribution

It introduces a topological classification approach for maze data, deriving the cost function behavior and identifying learning challenges like local minima.

Findings

Cost function shape is independent of maze size in large limits.

Rare dataset events influence the cost landscape.

Deep networks up to 128 layers effectively trained on maze classification.

Abstract

In this paper we approach two relevant deep learning topics: i) tackling of graph structured input data and ii) a better understanding and analysis of deep networks and related learning algorithms. With this in mind we focus on the topological classification of reachability in a particular subset of planar graphs (Mazes). Doing so, we are able to model the topology of data while staying in Euclidean space, thus allowing its processing with standard CNN architectures. We suggest a suitable architecture for this problem and show that it can express a perfect solution to the classification task. The shape of the cost function around this solution is also derived and, remarkably, does not depend on the size of the maze in the large maze limit. Responsible for this behavior are rare events in the dataset which strongly regulate the shape of the cost function near this global minimum. We further identify an obstacle to learning in the form of poorly performing local minima in which the network chooses to ignore some of the inputs. We further support our claims with training experiments and numerical analysis of the cost function on networks with up to $128$ layers.