deepstruct -- linking deep learning and graph theory

TL;DR

Deepstruct bridges deep learning and graph theory by enabling the imposition and extraction of graph structures in neural networks, facilitating research in network design, pruning, and analysis.

Contribution

It introduces methods to impose graph-based restrictions on neural networks and tools to extract graph structures from trained models, advancing neural architecture research.

Findings

Supports various graph restrictions on neural networks

Provides tools for graph extraction from trained models

Facilitates research in pruning and neural architecture search

Abstract

deepstruct connects deep learning models and graph theory such that different graph structures can be imposed on neural networks or graph structures can be extracted from trained neural network models. For this, deepstruct provides deep neural network models with different restrictions which can be created based on an initial graph. Further, tools to extract graph structures from trained models are available. This step of extracting graphs can be computationally expensive even for models of just a few dozen thousand parameters and poses a challenging problem. deepstruct supports research in pruning, neural architecture search, automated network design and structure analysis of neural networks.