Selective, Controlled and Domain-Agnostic Unlearning in Pretrained CLIP: A Training- and Data-Free Approach

TL;DR

This paper introduces a novel, training- and data-free method for unlearning specific object classes in CLIP models, enabling flexible, domain-specific, and global forgetting without retraining or additional data.

Contribution

It proposes a new unlearning framework that leverages multimodal nullspace and synthesized prototypes for controlled, domain-agnostic forgetting in pretrained CLIP models.

Findings

Effectively removes undesired classes across domains

Preserves model performance on unrelated tasks

Operates without retraining or extra data

Abstract

Pretrained models like CLIP have demonstrated impressive zero-shot classification capabilities across diverse visual domains, spanning natural images, artistic renderings, and abstract representations. However, real-world applications often demand the removal (or "unlearning") of specific object classes without requiring additional data or retraining, or affecting the model's performance on unrelated tasks. In this paper, we propose a novel training- and data-free unlearning framework that enables three distinct forgetting paradigms: (1) global unlearning of selected objects across all domains, (2) domain-specific knowledge removal (e.g., eliminating sketch representations while preserving photo recognition), and (3) complete unlearning in selective domains. By leveraging a multimodal nullspace through synergistic integration of text prompts and synthesized visual prototypes derived from CLIP's joint embedding space, our method efficiently removes undesired class information while preserving the remaining knowledge. This approach overcomes the limitations of existing retraining-based methods and offers a flexible and computationally efficient solution for controlled model forgetting.