Prompt-Driven Feature Diffusion for Open-World Semi-Supervised Learning

TL;DR

This paper introduces Prompt-Driven Feature Diffusion (PDFD), a novel semi-supervised learning method that uses class prompts and diffusion models to improve recognition of unseen classes in open-world settings.

Contribution

PDFD is the first to integrate class-specific prompts with feature diffusion for open-world semi-supervised learning, effectively handling unseen classes with limited labeled data.

Findings

PDFD outperforms state-of-the-art methods in open-world semi-supervised learning tasks.

The use of class prototypes as prompts enhances discriminative feature learning.

Incorporating adversarial loss improves the quality of generated features.

Abstract

In this paper, we present a novel approach termed Prompt-Driven Feature Diffusion (PDFD) within a semi-supervised learning framework for Open World Semi-Supervised Learning (OW-SSL). At its core, PDFD deploys an efficient feature-level diffusion model with the guidance of class-specific prompts to support discriminative feature representation learning and feature generation, tackling the challenge of the non-availability of labeled data for unseen classes in OW-SSL. In particular, PDFD utilizes class prototypes as prompts in the diffusion model, leveraging their class-discriminative and semantic generalization ability to condition and guide the diffusion process across all the seen and unseen classes. Furthermore, PDFD incorporates a class-conditional adversarial loss for diffusion model training, ensuring that the features generated via the diffusion process can be discriminatively aligned with the class-conditional features of the real data. Additionally, the class prototypes of the unseen classes are computed using only unlabeled instances with confident predictions within a semi-supervised learning framework. We conduct extensive experiments to evaluate the proposed PDFD. The empirical results show PDFD exhibits remarkable performance enhancements over many state-of-the-art existing methods.