Towards NSFW-Free Text-to-Image Generation via Safety-Constraint Direct Preference Optimization

• URL: http://arxiv.org/abs/2504.14290v1
• Date: Sat, 19 Apr 2025 13:26:46 GMT
• Title: Towards NSFW-Free Text-to-Image Generation via Safety-Constraint Direct Preference Optimization
• Authors: Shouwei Ruan, Zhenyu Wu, Yao Huang, Ruochen Zhang, Yitong Sun, Caixin Kang, Xingxing Wei,
• Abstract summary: Existing studies fail to guarantee complete safety under potentially harmful concepts or struggle to balance safety with generation quality.<n>We propose Safety-Constrained Direct Preference Optimization (SC-DPO), a novel framework for safety alignment in T2I models.<n>SC-DPO integrates safety constraints into the general human preference calibration, aiming to maximize the likelihood of generating human-preferred samples.
• Score: 30.31991120463517
• License: http://creativecommons.org/licenses/by-nc-sa/4.0/
• Abstract: Ensuring the safety of generated content remains a fundamental challenge for Text-to-Image (T2I) generation. Existing studies either fail to guarantee complete safety under potentially harmful concepts or struggle to balance safety with generation quality. To address these issues, we propose Safety-Constrained Direct Preference Optimization (SC-DPO), a novel framework for safety alignment in T2I models. SC-DPO integrates safety constraints into the general human preference calibration, aiming to maximize the likelihood of generating human-preferred samples while minimizing the safety cost of the generated outputs. In SC-DPO, we introduce a safety cost model to accurately quantify harmful levels for images, and train it effectively using the proposed contrastive learning and cost anchoring objectives. To apply SC-DPO for effective T2I safety alignment, we constructed SCP-10K, a safety-constrained preference dataset containing rich harmful concepts, which blends safety-constrained preference pairs under both harmful and clean instructions, further mitigating the trade-off between safety and sample quality. Additionally, we propose a Dynamic Focusing Mechanism (DFM) for SC-DPO, promoting the model's learning of difficult preference pair samples. Extensive experiments demonstrate that SC-DPO outperforms existing methods, effectively defending against various NSFW content while maintaining optimal sample quality and human preference alignment. Additionally, SC-DPO exhibits resilience against adversarial prompts designed to generate harmful content.

Related papers

• SafetyDPO: Scalable Safety Alignment for Text-to-Image Generation [68.07258248467309]
  Text-to-image (T2I) models have become widespread, but their limited safety guardrails expose end users to harmful content and potentially allow for model misuse.<n>Current safety measures are typically limited to text-based filtering or concept removal strategies, able to remove just a few concepts from the model's generative capabilities.<n>We introduce SafetyDPO, a method for safety alignment of T2I models through Direct Preference Optimization (DPO)<n>We train safety experts, in the form of low-rank adaptation (LoRA) matrices, able to guide the generation process away from specific safety-related
  arXiv  Detail & Related papers  (2024-12-13T18:59:52Z)
• Enhancing Safety in Reinforcement Learning with Human Feedback via Rectified Policy Optimization [16.35399722653875]
  We propose Rectified Policy Optimization (RePO) to balance helpfulness and safety (harmlessness) in large language models (LLMs)<n>At the core of RePO is a policy update mechanism driven by rectified policy gradients, which penalizes the strict safety violation of every prompt, thereby enhancing safety across nearly all prompts.
  arXiv  Detail & Related papers  (2024-10-25T19:08:23Z)
• SAFREE: Training-Free and Adaptive Guard for Safe Text-to-Image And Video Generation [65.30207993362595]
  Unlearning/editing-based methods for safe generation remove harmful concepts from models but face several challenges. We propose SAFREE, a training-free approach for safe T2I and T2V. We detect a subspace corresponding to a set of toxic concepts in the text embedding space and steer prompt embeddings away from this subspace.
  arXiv  Detail & Related papers  (2024-10-16T17:32:23Z)
• Bi-Factorial Preference Optimization: Balancing Safety-Helpfulness in Language Models [94.39278422567955]
  Fine-tuning large language models (LLMs) on human preferences has proven successful in enhancing their capabilities.<n>However, ensuring the safety of LLMs during fine-tuning remains a critical concern.<n>We propose a supervised learning framework called Bi-Factorial Preference Optimization (BFPO) to address this issue.
  arXiv  Detail & Related papers  (2024-08-27T17:31:21Z)
• What Makes and Breaks Safety Fine-tuning? A Mechanistic Study [64.9691741899956]
  Safety fine-tuning helps align Large Language Models (LLMs) with human preferences for their safe deployment. We design a synthetic data generation framework that captures salient aspects of an unsafe input. Using this, we investigate three well-known safety fine-tuning methods.
  arXiv  Detail & Related papers  (2024-07-14T16:12:57Z)
• Towards Comprehensive Post Safety Alignment of Large Language Models via Safety Patching [74.62818936088065]
  textscSafePatching is a novel framework for comprehensive PSA.<n>textscSafePatching achieves a more comprehensive PSA than baseline methods.<n>textscSafePatching demonstrates its superiority in continual PSA scenarios.
  arXiv  Detail & Related papers  (2024-05-22T16:51:07Z)
• ConstrainedZero: Chance-Constrained POMDP Planning using Learned Probabilistic Failure Surrogates and Adaptive Safety Constraints [34.9739641898452]
  This work introduces the ConstrainedZero policy algorithm that solves CC-POMDPs in belief space by learning neural network approximations of the optimal value and policy. Results show that by separating safety constraints from the objective we can achieve a target level of safety without optimizing the balance between rewards and costs.
  arXiv  Detail & Related papers  (2024-05-01T17:17:22Z)
• Enhancing LLM Safety via Constrained Direct Preference Optimization [8.22888921018027]
  We introduce Constrained DPO (C-DPO), a novel extension of the recently proposed Direct Preference Optimization (DPO) approach for fine-tuning AI systems. By integrating dual gradient descent and DPO, our method identifies a nearly optimal trade-off between helpfulness and harmlessness without using reinforcement learning. Empirically, our approach provides a safety guarantee to LLMs that is missing in DPO while achieving significantly higher rewards under the same safety constraint.
  arXiv  Detail & Related papers  (2024-03-04T20:39:24Z)
• Concurrent Learning of Policy and Unknown Safety Constraints in Reinforcement Learning [4.14360329494344]
  Reinforcement learning (RL) has revolutionized decision-making across a wide range of domains over the past few decades. Yet, deploying RL policies in real-world scenarios presents the crucial challenge of ensuring safety. Traditional safe RL approaches have predominantly focused on incorporating predefined safety constraints into the policy learning process. We propose a novel approach that concurrently learns a safe RL control policy and identifies the unknown safety constraint parameters of a given environment.
  arXiv  Detail & Related papers  (2024-02-24T20:01:15Z)
• Bayesian Optimization with Formal Safety Guarantees via Online Conformal Prediction [36.14499894307206]
  Black-box zero-th order optimization is a central primitive for applications in fields as diverse as finance, physics, and engineering. In this paper, we study scenarios in which feedback is also provided on the safety of the attempted solution. A novel BO-based approach is introduced that satisfies safety requirements irrespective of properties of the constraint function.
  arXiv  Detail & Related papers  (2023-06-30T17:26:49Z)
• Meta-Learning Priors for Safe Bayesian Optimization [72.8349503901712]
  We build on a meta-learning algorithm, F-PACOH, capable of providing reliable uncertainty quantification in settings of data scarcity. As core contribution, we develop a novel framework for choosing safety-compliant priors in a data-riven manner. On benchmark functions and a high-precision motion system, we demonstrate that our meta-learned priors accelerate the convergence of safe BO approaches.
  arXiv  Detail & Related papers  (2022-10-03T08:38:38Z)
• Enforcing Hard Constraints with Soft Barriers: Safe Reinforcement Learning in Unknown Stochastic Environments [84.3830478851369]
  We propose a safe reinforcement learning approach that can jointly learn the environment and optimize the control policy. Our approach can effectively enforce hard safety constraints and significantly outperform CMDP-based baseline methods in system safe rate measured via simulations.
  arXiv  Detail & Related papers  (2022-09-29T20:49:25Z)
• Recursively Feasible Probabilistic Safe Online Learning with Control Barrier Functions [60.26921219698514]
  We introduce a model-uncertainty-aware reformulation of CBF-based safety-critical controllers. We then present the pointwise feasibility conditions of the resulting safety controller. We use these conditions to devise an event-triggered online data collection strategy.
  arXiv  Detail & Related papers  (2022-08-23T05:02:09Z)
• Constrained Policy Optimization via Bayesian World Models [79.0077602277004]
  LAMBDA is a model-based approach for policy optimization in safety critical tasks modeled via constrained Markov decision processes. We demonstrate LAMBDA's state of the art performance on the Safety-Gym benchmark suite in terms of sample efficiency and constraint violation.
  arXiv  Detail & Related papers  (2022-01-24T17:02:22Z)

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.