SimPer: Simple Self-Supervised Learning of Periodic Targets

TL;DR

SimPer is a straightforward self-supervised learning method that effectively captures periodic features in data, improving representation quality for tasks across various real-world domains with limited supervision.

Contribution

It introduces a novel contrastive SSL framework with customized augmentations and loss functions specifically designed for learning periodic information.

Findings

Outperforms state-of-the-art SSL methods in real-world tasks

Demonstrates better data efficiency and robustness

Generalizes well to distribution shifts

Abstract

From human physiology to environmental evolution, important processes in nature often exhibit meaningful and strong periodic or quasi-periodic changes. Due to their inherent label scarcity, learning useful representations for periodic tasks with limited or no supervision is of great benefit. Yet, existing self-supervised learning (SSL) methods overlook the intrinsic periodicity in data, and fail to learn representations that capture periodic or frequency attributes. In this paper, we present SimPer, a simple contrastive SSL regime for learning periodic information in data. To exploit the periodic inductive bias, SimPer introduces customized augmentations, feature similarity measures, and a generalized contrastive loss for learning efficient and robust periodic representations. Extensive experiments on common real-world tasks in human behavior analysis, environmental sensing, and healthcare domains verify the superior performance of SimPer compared to state-of-the-art SSL methods, highlighting its intriguing properties including better data efficiency, robustness to spurious correlations, and generalization to distribution shifts. Code and data are available at: https://github.com/YyzHarry/SimPer.