arXiv subjects

Tsung-Yi Ho

Publications and source records attributed to Tsung-Yi Ho.

At least 55 records · Page 3Linked to original sources

arXiv2025

Defensive Prompt Patch: A Robust and Interpretable Defense of LLMs against Jailbreak Attacks

Safety, security, and compliance are essential requirements when aligning large language models (LLMs). However, many seemingly aligned LLMs are soon shown to be susceptible to jailbreak attacks. These attacks aim to circumvent the models' safety guardrails and security mechanisms by introducing jailbreak prompts into malicious queries. In response to these challenges, this paper introduces Defensive Prompt Patch (DPP), a novel prompt-based defense mechanism specifically designed to protect LLMs against such sophisticated jailbreak strategies. Unlike previous approaches, which have often compromised the utility of the model for the sake of safety, DPP is designed to achieve a minimal Attack Success Rate (ASR) while preserving the high utility of LLMs. Our method uses strategically designed interpretable suffix prompts that effectively thwart a wide range of standard and adaptive jailbreak techniques. Empirical results conducted on LLAMA-2-7B-Chat and Mistral-7B-Instruct-v0.2 models demonstrate the robustness and adaptability of DPP, showing significant reductions in ASR with negligible impact on utility. Our approach not only outperforms existing defense strategies in balancing safety and functionality, but also provides a scalable and interpretable solution applicable to various LLM platforms.

cs.CR↗

arXiv2025

Rethinking Backdoor Attacks on Dataset Distillation: A Kernel Method Perspective

Dataset distillation offers a potential means to enhance data efficiency in deep learning. Recent studies have shown its ability to counteract backdoor risks present in original training samples. In this study, we delve into the theoretical aspects of backdoor attacks and dataset distillation based on kernel methods. We introduce two new theory-driven trigger pattern generation methods specialized for dataset distillation. Following a comprehensive set of analyses and experiments, we show that our optimization-based trigger design framework informs effective backdoor attacks on dataset distillation. Notably, datasets poisoned by our designed trigger prove resilient against conventional backdoor attack detection and mitigation methods. Our empirical results validate that the triggers developed using our approaches are proficient at executing resilient backdoor attacks.

cs.LG↗

arXiv2025

Unsupervised Out-of-Distribution Detection in Medical Imaging Using Multi-Exit Class Activation Maps and Feature Masking

Out-of-distribution (OOD) detection is essential for ensuring the reliability of deep learning models in medical imaging applications. This work is motivated by the observation that class activation maps (CAMs) for in-distribution (ID) data typically emphasize regions that are highly relevant to the model's predictions, whereas OOD data often lacks such focused activations. By masking input images with inverted CAMs, the feature representations of ID data undergo more substantial changes compared to those of OOD data, offering a robust criterion for differentiation. In this paper, we introduce a novel unsupervised OOD detection framework, Multi-Exit Class Activation Map (MECAM), which leverages multi-exit CAMs and feature masking. By utilizing mult-exit networks that combine CAMs from varying resolutions and depths, our method captures both global and local feature representations, thereby enhancing the robustness of OOD detection. We evaluate MECAM on multiple ID datasets, including ISIC19 and PathMNIST, and test its performance against three medical OOD datasets, RSNA Pneumonia, COVID-19, and HeadCT, and one natural image OOD dataset, iSUN. Comprehensive comparisons with state-of-the-art OOD detection methods validate the effectiveness of our approach. Our findings emphasize the potential of multi-exit networks and feature masking for advancing unsupervised OOD detection in medical imaging, paving the way for more reliable and interpretable models in clinical practice.

cs.CV↗

arXiv2024

Token Highlighter: Inspecting and Mitigating Jailbreak Prompts for Large Language Models

Large Language Models (LLMs) are increasingly being integrated into services such as ChatGPT to provide responses to user queries. To mitigate potential harm and prevent misuse, there have been concerted efforts to align the LLMs with human values and legal compliance by incorporating various techniques, such as Reinforcement Learning from Human Feedback (RLHF), into the training of the LLMs. However, recent research has exposed that even aligned LLMs are susceptible to adversarial manipulations known as Jailbreak Attacks. To address this challenge, this paper proposes a method called Token Highlighter to inspect and mitigate the potential jailbreak threats in the user query. Token Highlighter introduced a concept called Affirmation Loss to measure the LLM's willingness to answer the user query. It then uses the gradient of Affirmation Loss for each token in the user query to locate the jailbreak-critical tokens. Further, Token Highlighter exploits our proposed Soft Removal technique to mitigate the jailbreak effects of critical tokens via shrinking their token embeddings. Experimental results on two aligned LLMs (LLaMA-2 and Vicuna-V1.5) demonstrate that the proposed method can effectively defend against a variety of Jailbreak Attacks while maintaining competent performance on benign questions of the AlpacaEval benchmark. In addition, Token Highlighter is a cost-effective and interpretable defense because it only needs to query the protected LLM once to compute the Affirmation Loss and can highlight the critical tokens upon refusal.

cs.CR↗

arXiv2024

Retention Score: Quantifying Jailbreak Risks for Vision Language Models

The emergence of Vision-Language Models (VLMs) is a significant advancement in integrating computer vision with Large Language Models (LLMs) to enhance multi-modal machine learning capabilities. However, this progress has also made VLMs vulnerable to sophisticated adversarial attacks, raising concerns about their reliability. The objective of this paper is to assess the resilience of VLMs against jailbreak attacks that can compromise model safety compliance and result in harmful outputs. To evaluate a VLM's ability to maintain its robustness against adversarial input perturbations, we propose a novel metric called the \textbf{Retention Score}. Retention Score is a multi-modal evaluation metric that includes Retention-I and Retention-T scores for quantifying jailbreak risks in visual and textual components of VLMs. Our process involves generating synthetic image-text pairs using a conditional diffusion model. These pairs are then predicted for toxicity score by a VLM alongside a toxicity judgment classifier. By calculating the margin in toxicity scores, we can quantify the robustness of the VLM in an attack-agnostic manner. Our work has four main contributions. First, we prove that Retention Score can serve as a certified robustness metric. Second, we demonstrate that most VLMs with visual components are less robust against jailbreak attacks than the corresponding plain VLMs. Additionally, we evaluate black-box VLM APIs and find that the security settings in Google Gemini significantly affect the score and robustness. Moreover, the robustness of GPT4V is similar to the medium settings of Gemini. Finally, our approach offers a time-efficient alternative to existing adversarial attack methods and provides consistent model robustness rankings when evaluated on VLMs including MiniGPT-4, InstructBLIP, and LLaVA.

cs.AI↗

arXiv2024

Echo: Simulating Distributed Training At Scale

Simulation offers unique values for both enumeration and extrapolation purposes, and is becoming increasingly important for managing the massive machine learning (ML) clusters and large-scale distributed training jobs. In this paper, we build Echo to tackle three key challenges in large-scale training simulation: (1) tracing the runtime training workloads at each device in an ex-situ fashion so we can use a single device to obtain the actual execution graphs of 1K-GPU training, (2) accurately estimating the collective communication without high overheads of discrete-event based network simulation, and (3) accounting for the interference-induced computation slowdown from overlapping communication and computation kernels on the same device. Echo delivers on average 8% error in training step -- roughly 3x lower than state-of-the-art simulators -- for GPT-175B on a 96-GPU H800 cluster with 3D parallelism on Megatron-LM under 2 minutes.

cs.LG↗

arXiv2024

NeuralFuse: Learning to Recover the Accuracy of Access-Limited Neural Network Inference in Low-Voltage Regimes

Deep neural networks (DNNs) have become ubiquitous in machine learning, but their energy consumption remains problematically high. An effective strategy for reducing such consumption is supply-voltage reduction, but if done too aggressively, it can lead to accuracy degradation. This is due to random bit-flips in static random access memory (SRAM), where model parameters are stored. To address this challenge, we have developed NeuralFuse, a novel add-on module that handles the energy-accuracy tradeoff in low-voltage regimes by learning input transformations and using them to generate error-resistant data representations, thereby protecting DNN accuracy in both nominal and low-voltage scenarios. As well as being easy to implement, NeuralFuse can be readily applied to DNNs with limited access, such cloud-based APIs that are accessed remotely or non-configurable hardware. Our experimental results demonstrate that, at a 1% bit-error rate, NeuralFuse can reduce SRAM access energy by up to 24% while recovering accuracy by up to 57%. To the best of our knowledge, this is the first approach to addressing low-voltage-induced bit errors that requires no model retraining.

cs.LG↗

arXiv2024

Gradient Cuff: Detecting Jailbreak Attacks on Large Language Models by Exploring Refusal Loss Landscapes

Large Language Models (LLMs) are becoming a prominent generative AI tool, where the user enters a query and the LLM generates an answer. To reduce harm and misuse, efforts have been made to align these LLMs to human values using advanced training techniques such as Reinforcement Learning from Human Feedback (RLHF). However, recent studies have highlighted the vulnerability of LLMs to adversarial jailbreak attempts aiming at subverting the embedded safety guardrails. To address this challenge, this paper defines and investigates the Refusal Loss of LLMs and then proposes a method called Gradient Cuff to detect jailbreak attempts. Gradient Cuff exploits the unique properties observed in the refusal loss landscape, including functional values and its smoothness, to design an effective two-step detection strategy. Experimental results on two aligned LLMs (LLaMA-2-7B-Chat and Vicuna-7B-V1.5) and six types of jailbreak attacks (GCG, AutoDAN, PAIR, TAP, Base64, and LRL) show that Gradient Cuff can significantly improve the LLM's rejection capability for malicious jailbreak queries, while maintaining the model's performance for benign user queries by adjusting the detection threshold.

cs.CR↗

arXiv2024

GREAT Score: Global Robustness Evaluation of Adversarial Perturbation using Generative Models

Current studies on adversarial robustness mainly focus on aggregating local robustness results from a set of data samples to evaluate and rank different models. However, the local statistics may not well represent the true global robustness of the underlying unknown data distribution. To address this challenge, this paper makes the first attempt to present a new framework, called GREAT Score , for global robustness evaluation of adversarial perturbation using generative models. Formally, GREAT Score carries the physical meaning of a global statistic capturing a mean certified attack-proof perturbation level over all samples drawn from a generative model. For finite-sample evaluation, we also derive a probabilistic guarantee on the sample complexity and the difference between the sample mean and the true mean. GREAT Score has several advantages: (1) Robustness evaluations using GREAT Score are efficient and scalable to large models, by sparing the need of running adversarial attacks. In particular, we show high correlation and significantly reduced computation cost of GREAT Score when compared to the attack-based model ranking on RobustBench (Croce,et. al. 2021). (2) The use of generative models facilitates the approximation of the unknown data distribution. In our ablation study with different generative adversarial networks (GANs), we observe consistency between global robustness evaluation and the quality of GANs. (3) GREAT Score can be used for remote auditing of privacy-sensitive black-box models, as demonstrated by our robustness evaluation on several online facial recognition services.

cs.LG↗

arXiv2024

Evaluating Text-to-Image Generative Models: An Empirical Study on Human Image Synthesis

In this paper, we present an empirical study introducing a nuanced evaluation framework for text-to-image (T2I) generative models, applied to human image synthesis. Our framework categorizes evaluations into two distinct groups: first, focusing on image qualities such as aesthetics and realism, and second, examining text conditions through concept coverage and fairness. We introduce an innovative aesthetic score prediction model that assesses the visual appeal of generated images and unveils the first dataset marked with low-quality regions in generated human images to facilitate automatic defect detection. Our exploration into concept coverage probes the model's effectiveness in interpreting and rendering text-based concepts accurately, while our analysis of fairness reveals biases in model outputs, with an emphasis on gender, race, and age. While our study is grounded in human imagery, this dual-faceted approach is designed with the flexibility to be applicable to other forms of image generation, enhancing our understanding of generative models and paving the way to the next generation of more sophisticated, contextually aware, and ethically attuned generative models. Code and data, including the dataset annotated with defective areas, are available at \href{https://github.com/cure-lab/EvaluateAIGC}{https://github.com/cure-lab/EvaluateAIGC}.

cs.CV↗

arXiv2024

Elephant in the Room: Unveiling the Impact of Reward Model Quality in Alignment

The demand for regulating potentially risky behaviors of large language models (LLMs) has ignited research on alignment methods. Since LLM alignment heavily relies on reward models for optimization or evaluation, neglecting the quality of reward models may cause unreliable results or even misalignment. Despite the vital role reward models play in alignment, previous works have consistently overlooked their performance and used off-the-shelf reward models arbitrarily without verification, rendering the reward model ``\emph{an elephant in the room}''. To this end, this work first investigates the quality of the widely-used preference dataset, HH-RLHF, and curates a clean version, CHH-RLHF. Based on CHH-RLHF, we benchmark the accuracy of a broad range of reward models used in previous alignment works, unveiling the unreliability of using them both for optimization and evaluation. Furthermore, we systematically study the impact of reward model quality on alignment performance in three reward utilization paradigms. Extensive experiments reveal that better reward models perform as better human preference proxies. This work aims to awaken people to notice this huge elephant in alignment research. We call attention to the following issues: (1) The reward model needs to be rigorously evaluated, whether for alignment optimization or evaluation. (2) Considering the role of reward models, research efforts should not only concentrate on alignment algorithm, but also on developing more reliable human proxy.

cs.CL↗

arXiv2024

When Does Visual Prompting Outperform Linear Probing for Vision-Language Models? A Likelihood Perspective

Adapting pre-trained models to new tasks can exhibit varying effectiveness across datasets. Visual prompting, a state-of-the-art parameter-efficient transfer learning method, can significantly improve the performance of out-of-distribution tasks. On the other hand, linear probing, a standard transfer learning method, can sometimes become the best approach. We propose a log-likelihood ratio (LLR) approach to analyze the comparative benefits of visual prompting and linear probing. By employing the LLR score alongside resource-efficient visual prompts approximations, our cost-effective measure attains up to a 100-fold reduction in run time compared to full training, while achieving prediction accuracies up to 91%. The source code is available at https://github.com/IBM/VP-LLR.

cs.CV↗

arXiv2024

Neural Clamping: Joint Input Perturbation and Temperature Scaling for Neural Network Calibration

Neural network calibration is an essential task in deep learning to ensure consistency between the confidence of model prediction and the true correctness likelihood. In this paper, we propose a new post-processing calibration method called Neural Clamping, which employs a simple joint input-output transformation on a pre-trained classifier via a learnable universal input perturbation and an output temperature scaling parameter. Moreover, we provide theoretical explanations on why Neural Clamping is provably better than temperature scaling. Evaluated on BloodMNIST, CIFAR-100, and ImageNet image recognition datasets and a variety of deep neural network models, our empirical results show that Neural Clamping significantly outperforms state-of-the-art post-processing calibration methods. The code is available at github.com/yungchentang/NCToolkit, and the demo is available at huggingface.co/spaces/TrustSafeAI/NCTV.

cs.LG↗

arXiv2024

The Devil is in the Neurons: Interpreting and Mitigating Social Biases in Pre-trained Language Models

Pre-trained Language models (PLMs) have been acknowledged to contain harmful information, such as social biases, which may cause negative social impacts or even bring catastrophic results in application. Previous works on this problem mainly focused on using black-box methods such as probing to detect and quantify social biases in PLMs by observing model outputs. As a result, previous debiasing methods mainly finetune or even pre-train language models on newly constructed anti-stereotypical datasets, which are high-cost. In this work, we try to unveil the mystery of social bias inside language models by introducing the concept of {\sc Social Bias Neurons}. Specifically, we propose {\sc Integrated Gap Gradients (IG$^2$)} to accurately pinpoint units (i.e., neurons) in a language model that can be attributed to undesirable behavior, such as social bias. By formalizing undesirable behavior as a distributional property of language, we employ sentiment-bearing prompts to elicit classes of sensitive words (demographics) correlated with such sentiments. Our IG$^2$ thus attributes the uneven distribution for different demographics to specific Social Bias Neurons, which track the trail of unwanted behavior inside PLM units to achieve interoperability. Moreover, derived from our interpretable technique, {\sc Bias Neuron Suppression (BNS)} is further proposed to mitigate social biases. By studying BERT, RoBERTa, and their attributable differences from debiased FairBERTa, IG$^2$ allows us to locate and suppress identified neurons, and further mitigate undesired behaviors. As measured by prior metrics from StereoSet, our model achieves a higher degree of fairness while maintaining language modeling ability with low cost.

cs.CL↗

arXiv2024

RIGID: A Training-free and Model-Agnostic Framework for Robust AI-Generated Image Detection

The rapid advances in generative AI models have empowered the creation of highly realistic images with arbitrary content, raising concerns about potential misuse and harm, such as Deepfakes. Current research focuses on training detectors using large datasets of generated images. However, these training-based solutions are often computationally expensive and show limited generalization to unseen generated images. In this paper, we propose a training-free method to distinguish between real and AI-generated images. We first observe that real images are more robust to tiny noise perturbations than AI-generated images in the representation space of vision foundation models. Based on this observation, we propose RIGID, a training-free and model-agnostic method for robust AI-generated image detection. RIGID is a simple yet effective approach that identifies whether an image is AI-generated by comparing the representation similarity between the original and the noise-perturbed counterpart. Our evaluation on a diverse set of AI-generated images and benchmarks shows that RIGID significantly outperforms existing trainingbased and training-free detectors. In particular, the average performance of RIGID exceeds the current best training-free method by more than 25%. Importantly, RIGID exhibits strong generalization across different image generation methods and robustness to image corruptions.

cs.CV↗

arXiv2024

Achieving Fairness Through Channel Pruning for Dermatological Disease Diagnosis

Numerous studies have revealed that deep learning-based medical image classification models may exhibit bias towards specific demographic attributes, such as race, gender, and age. Existing bias mitigation methods often achieve high level of fairness at the cost of significant accuracy degradation. In response to this challenge, we propose an innovative and adaptable Soft Nearest Neighbor Loss-based channel pruning framework, which achieves fairness through channel pruning. Traditionally, channel pruning is utilized to accelerate neural network inference. However, our work demonstrates that pruning can also be a potent tool for achieving fairness. Our key insight is that different channels in a layer contribute differently to the accuracy of different groups. By selectively pruning critical channels that lead to the accuracy difference between the privileged and unprivileged groups, we can effectively improve fairness without sacrificing accuracy significantly. Experiments conducted on two skin lesion diagnosis datasets across multiple sensitive attributes validate the effectiveness of our method in achieving state-of-the-art trade-off between accuracy and fairness. Our code is available at https://github.com/Kqp1227/Sensitive-Channel-Pruning.

cs.CV↗

arXiv2024

TroLLoc: Logic Locking and Layout Hardening for IC Security Closure against Hardware Trojans

Due to cost benefits, supply chains of integrated circuits (ICs) are largely outsourced nowadays. However, passing ICs through various third-party providers gives rise to many security threats, like piracy of IC intellectual property or insertion of hardware Trojans, i.e., malicious circuit modifications. In this work, we proactively and systematically protect the physical layouts of ICs against post-design insertion of Trojans. Toward that end, we propose TroLLoc, a novel scheme for IC security closure that employs, for the first time, logic locking and layout hardening in unison. TroLLoc is fully integrated into a commercial-grade design flow, and TroLLoc is shown to be effective, efficient, and robust. Our work provides in-depth layout and security analysis considering the challenging benchmarks of the ISPD'22/23 contests for security closure. We show that TroLLoc successfully renders layouts resilient, with reasonable overheads, against (i) general prospects for Trojan insertion as in the ISPD'22 contest, (ii) actual Trojan insertion as in the ISPD'23 contest, and (iii) potential second-order attacks where adversaries would first (i.e., before Trojan insertion) try to bypass the locking defense, e.g., using advanced machine learning attacks. Finally, we release all our artifacts for independent verification [2].

cs.CR↗

arXiv2024

The Dawn of AI-Native EDA: Opportunities and Challenges of Large Circuit Models

Within the Electronic Design Automation (EDA) domain, AI-driven solutions have emerged as formidable tools, yet they typically augment rather than redefine existing methodologies. These solutions often repurpose deep learning models from other domains, such as vision, text, and graph analytics, applying them to circuit design without tailoring to the unique complexities of electronic circuits. Such an AI4EDA approach falls short of achieving a holistic design synthesis and understanding, overlooking the intricate interplay of electrical, logical, and physical facets of circuit data. This paper argues for a paradigm shift from AI4EDA towards AI-native EDA, integrating AI at the core of the design process. Pivotal to this vision is the development of a multimodal circuit representation learning technique, poised to provide a comprehensive understanding by harmonizing and extracting insights from varied data sources, such as functional specifications, RTL designs, circuit netlists, and physical layouts. We champion the creation of large circuit models (LCMs) that are inherently multimodal, crafted to decode and express the rich semantics and structures of circuit data, thus fostering more resilient, efficient, and inventive design methodologies. Embracing this AI-native philosophy, we foresee a trajectory that transcends the current innovation plateau in EDA, igniting a profound shift-left in electronic design methodology. The envisioned advancements herald not just an evolution of existing EDA tools but a revolution, giving rise to novel instruments of design tools that promise to radically enhance design productivity and inaugurate a new epoch where the optimization of circuit performance, power, and area (PPA) is achieved not incrementally, but through leaps that redefine the benchmarks of electronic systems' capabilities.

cs.AR↗