SSDL: Self-Supervised Dictionary Learning

TL;DR

This paper introduces SSDL, a self-supervised dictionary learning framework that uses a hypergraph-based pretext task to generate pseudo labels, improving dictionary learning in semi-supervised and unsupervised settings.

Contribution

The paper proposes a novel self-supervised approach for dictionary learning using a hypergraph-based pretext task to generate pseudo labels, enhancing performance without relying on labeled data.

Findings

SSDL outperforms state-of-the-art methods on human activity recognition datasets.

The hypergraph-based pretext task effectively generates useful pseudo labels.

SSDL demonstrates efficiency in semi-supervised and unsupervised learning scenarios.

Abstract

The label-embedded dictionary learning (DL) algorithms generate influential dictionaries by introducing discriminative information. However, there exists a limitation: All the label-embedded DL methods rely on the labels due that this way merely achieves ideal performances in supervised learning. While in semi-supervised and unsupervised learning, it is no longer sufficient to be effective. Inspired by the concept of self-supervised learning (e.g., setting the pretext task to generate a universal model for the downstream task), we propose a Self-Supervised Dictionary Learning (SSDL) framework to address this challenge. Specifically, we first design a $p$-Laplacian Attention Hypergraph Learning (pAHL) block as the pretext task to generate pseudo soft labels for DL. Then, we adopt the pseudo labels to train a dictionary from a primary label-embedded DL method. We evaluate our SSDL on two human activity recognition datasets. The comparison results with other state-of-the-art methods have demonstrated the efficiency of SSDL.