Concept Steerers: Leveraging K-Sparse Autoencoders for Controllable Generations

• URL: http://arxiv.org/abs/2501.19066v1
• Date: Fri, 31 Jan 2025 11:52:47 GMT
• Title: Concept Steerers: Leveraging K-Sparse Autoencoders for Controllable Generations
• Authors: Dahye Kim, Deepti Ghadiyaram,
• Abstract summary: Text-to-image generative models are prone to adversarial attacks and inadvertently generate unsafe, unethical content.<n>We propose a novel framework leveraging k-sparse autoencoders (k-SAEs) to enable efficient and interpretable concept manipulation.<n>Our method yields an improvement of $mathbf20.01%$ in unsafe concept removal, is effective in style manipulation, and is $mathbfsim5$x faster than current state-of-the-art.
• Score: 10.86252546314626
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Despite the remarkable progress in text-to-image generative models, they are prone to adversarial attacks and inadvertently generate unsafe, unethical content. Existing approaches often rely on fine-tuning models to remove specific concepts, which is computationally expensive, lack scalability, and/or compromise generation quality. In this work, we propose a novel framework leveraging k-sparse autoencoders (k-SAEs) to enable efficient and interpretable concept manipulation in diffusion models. Specifically, we first identify interpretable monosemantic concepts in the latent space of text embeddings and leverage them to precisely steer the generation away or towards a given concept (e.g., nudity) or to introduce a new concept (e.g., photographic style). Through extensive experiments, we demonstrate that our approach is very simple, requires no retraining of the base model nor LoRA adapters, does not compromise the generation quality, and is robust to adversarial prompt manipulations. Our method yields an improvement of $\mathbf{20.01\%}$ in unsafe concept removal, is effective in style manipulation, and is $\mathbf{\sim5}$x faster than current state-of-the-art.

Related papers

• ACE: Attentional Concept Erasure in Diffusion Models [0.0]
  Attentional Concept Erasure integrates a closed-form attention manipulation with lightweight fine-tuning. We show that ACE achieves state-of-the-art concept removal efficacy and robustness. Compared to prior methods, ACE better balances generality (erasing concept and related terms) and specificity (preserving unrelated content)
  arXiv  Detail & Related papers  (2025-04-16T08:16:28Z)
• Sparse Autoencoder as a Zero-Shot Classifier for Concept Erasing in Text-to-Image Diffusion Models [24.15603438969762]
  Interpret then Deactivate (ItD) is a novel framework to enable precise concept removal in T2I diffusion models. ItD uses a sparse autoencoder to interpret each concept as a combination of multiple features. It can be easily extended to erase multiple concepts without requiring further training.
  arXiv  Detail & Related papers  (2025-03-12T14:46:40Z)
• Continuous Concepts Removal in Text-to-image Diffusion Models [27.262721132177845]
  Concerns have been raised about the potential for text-to-image models to create content that infringes on copyrights or depicts disturbing subject matter.<n>We propose a novel approach called CCRT that includes a designed knowledge distillation paradigm.<n>It constrains the text-image alignment behavior during the continuous concept removal process by using a set of text prompts.
  arXiv  Detail & Related papers  (2024-11-30T20:40:10Z)
• Safety Without Semantic Disruptions: Editing-free Safe Image Generation via Context-preserving Dual Latent Reconstruction [88.18235230849554]
  Training multimodal generative models on large, uncurated datasets can result in users being exposed to harmful, unsafe and controversial or culturally-inappropriate outputs. We leverage safe embeddings and a modified diffusion process with weighted tunable summation in the latent space to generate safer images. We identify trade-offs between safety and censorship, which presents a necessary perspective in the development of ethical AI models.
  arXiv  Detail & Related papers  (2024-11-21T09:47:13Z)
• Scaling Concept With Text-Guided Diffusion Models [53.80799139331966]
  Instead of replacing a concept, can we enhance or suppress the concept itself? We introduce ScalingConcept, a simple yet effective method to scale decomposed concepts up or down in real input without introducing new elements. More importantly, ScalingConcept enables a variety of novel zero-shot applications across image and audio domains.
  arXiv  Detail & Related papers  (2024-10-31T17:09:55Z)
• Adversarial Robustification via Text-to-Image Diffusion Models [56.37291240867549]
  Adrial robustness has been conventionally believed as a challenging property to encode for neural networks. We develop a scalable and model-agnostic solution to achieve adversarial robustness without using any data.
  arXiv  Detail & Related papers  (2024-07-26T10:49:14Z)
• Reliable and Efficient Concept Erasure of Text-to-Image Diffusion Models [76.39651111467832]
  We introduce Reliable and Efficient Concept Erasure (RECE), a novel approach that modifies the model in 3 seconds without necessitating additional fine-tuning. To mitigate inappropriate content potentially represented by derived embeddings, RECE aligns them with harmless concepts in cross-attention layers. The derivation and erasure of new representation embeddings are conducted iteratively to achieve a thorough erasure of inappropriate concepts.
  arXiv  Detail & Related papers  (2024-07-17T08:04:28Z)
• Safeguard Text-to-Image Diffusion Models with Human Feedback Inversion [51.931083971448885]
  We propose a framework named Human Feedback Inversion (HFI), where human feedback on model-generated images is condensed into textual tokens guiding the mitigation or removal of problematic images. Our experimental results demonstrate our framework significantly reduces objectionable content generation while preserving image quality, contributing to the ethical deployment of AI in the public sphere.
  arXiv  Detail & Related papers  (2024-07-17T05:21:41Z)
• All but One: Surgical Concept Erasing with Model Preservation in Text-to-Image Diffusion Models [22.60023885544265]
  Large-scale datasets may contain sexually explicit, copyrighted, or undesirable content, which allows the model to directly generate them. Fine-tuning algorithms have been developed to tackle concept erasing in diffusion models. We present a new approach that solves all of these challenges.
  arXiv  Detail & Related papers  (2023-12-20T07:04:33Z)
• Implicit Concept Removal of Diffusion Models [92.55152501707995]
  Text-to-image (T2I) diffusion models often inadvertently generate unwanted concepts such as watermarks and unsafe images. We present the Geom-Erasing, a novel concept removal method based on the geometric-driven control.
  arXiv  Detail & Related papers  (2023-10-09T17:13:10Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.