Few-Shot Concept Unlearning with Low Rank Adaptation

TL;DR

This paper introduces a fast method for concept unlearning in diffusion models by updating the final layers of text encoders using a weighted loss that incorporates Textual Inversion and Low-Rank Adaptation, significantly reducing unlearning time.

Contribution

It proposes a novel, efficient algorithm for concept unlearning in diffusion models that updates only the final layers of text encoders with a weighted loss function.

Findings

Unlearning takes approximately 50 seconds per concept.

The method effectively removes concept influence from the model.

It leverages Textual Inversion and Low-Rank Adaptation techniques.

Abstract

Image Generation models are a trending topic nowadays, with many people utilizing Artificial Intelligence models in order to generate images. There are many such models which, given a prompt of a text, will generate an image which depicts said prompt. There are many image generation models, such as Latent Diffusion Models, Denoising Diffusion Probabilistic Models, Generative Adversarial Networks and many more. When generating images, these models can generate sensitive image data, which can be threatening to privacy or may violate copyright laws of private entities. Machine unlearning aims at removing the influence of specific data subsets from the trained models and in the case of image generation models, remove the influence of a concept such that the model is unable to generate said images of the concept when prompted. Conventional retraining of the model can take upto days, hence fast algorithms are the need of the hour. In this paper we propose an algorithm that aims to remove the influence of concepts in diffusion models through updating the gradients of the final layers of the text encoders. Using a weighted loss function, we utilize backpropagation in order to update the weights of the final layers of the Text Encoder componet of the Stable Diffusion Model, removing influence of the concept from the text-image embedding space, such that when prompted, the result is an image not containing the concept. The weighted loss function makes use of Textual Inversion and Low-Rank Adaptation.We perform our experiments on Latent Diffusion Models, namely the Stable Diffusion v2 model, with an average concept unlearning runtime of 50 seconds using 4-5 images.