CSI: Contrastive Data Stratification for Interaction Prediction and its Application to Compound-Protein Interaction Prediction

TL;DR

This paper introduces CSI, a contrastive data stratification method that leverages multi-view data partitioning to improve interaction prediction, demonstrated on compound-protein interactions.

Contribution

It proposes a novel contrastive learning framework that uses data stratification to enhance interaction prediction accuracy.

Findings

Improved prediction performance on compound-protein interaction datasets

Effective use of multi-view data partitioning in contrastive learning

Enhanced object representations for interaction tasks

Abstract

Accurately predicting the likelihood of interaction between two objects (compound-protein sequence, user-item, author-paper, etc.) is a fundamental problem in Computer Science. Current deep-learning models rely on learning accurate representations of the interacting objects. Importantly, relationships between the interacting objects, or features of the interaction, offer an opportunity to partition the data to create multi-views of the interacting objects. The resulting congruent and non-congruent views can then be exploited via contrastive learning techniques to learn enhanced representations of the objects.