IL-Net: Using Expert Knowledge to Guide the Design of Furcated Neural Networks

TL;DR

This paper introduces IL-Net, a furcated neural network that leverages domain knowledge to improve property prediction of ionic liquids, achieving 20-35% better accuracy than existing methods without extra labeled data.

Contribution

The paper presents a novel furcated neural network architecture guided by domain knowledge, demonstrating its effectiveness in a complex chemical prediction task and distilling general design principles.

Findings

Furcated networks improve accuracy by 20-35%.

IL-Net outperforms state-of-the-art methods.

Two key design principles for furcated networks are identified.

Abstract

Deep neural networks (DNN) excel at extracting patterns. Through representation learning and automated feature engineering on large datasets, such models have been highly successful in computer vision and natural language applications. Designing optimal network architectures from a principled or rational approach however has been less than successful, with the best successful approaches utilizing an additional machine learning algorithm to tune the network hyperparameters. However, in many technical fields, there exist established domain knowledge and understanding about the subject matter. In this work, we develop a novel furcated neural network architecture that utilizes domain knowledge as high-level design principles of the network. We demonstrate proof-of-concept by developing IL-Net, a furcated network for predicting the properties of ionic liquids, which is a class of complex multi-chemicals entities. Compared to existing state-of-the-art approaches, we show that furcated networks can improve model accuracy by approximately 20-35%, without using additional labeled data. Lastly, we distill two key design principles for furcated networks that can be adapted to other domains.