相关论文: Safe Transformer: An Explicit Safety Bit For Inter…

Automatic speech recognition (ASR) based on transducers is widely used. In training, a transducer maximizes the summed posteriors of all paths. The path with the highest posterior is commonly defined as the predicted alignment between the…

Synthesising safe controllers from visual data typically requires extensive supervised labelling of safety-critical data, which is often impractical in real-world settings. Recent advances in world models enable reliable prediction in…

AI systems can take harmful actions and are highly vulnerable to adversarial attacks. We present an approach, inspired by recent advances in representation engineering, that interrupts the models as they respond with harmful outputs with…

Transformer models are permutation equivariant. To supply the order and type information of the input tokens, position and segment embeddings are usually added to the input. Recent works proposed variations of positional encodings with…

SafePredict is a novel meta-algorithm that works with any base prediction algorithm for online data to guarantee an arbitrarily chosen correctness rate, $1-\epsilon$, by allowing refusals. Allowing refusals means that the meta-algorithm may…

Refusal on harmful prompts is a key safety behaviour in instruction-tuned large language models (LLMs), yet the internal causes of this behaviour remain poorly understood. We study two public instruction-tuned models, Gemma-2-2B-IT and…

The rise of hardware-level security threats, such as side-channel attacks, hardware Trojans, and firmware vulnerabilities, demands advanced detection mechanisms that are more intelligent and adaptive. Traditional methods often fall short in…

Upper-limb exoskeletons are primarily designed to provide assistive support by accurately interpreting and responding to human intentions. In home-care scenarios, exoskeletons are expected to adapt their assistive configurations based on…

This paper presents a novel approach for the safe control design of systems with parametric uncertainties in both drift terms and control-input matrices. The method combines control barrier functions and adaptive laws to generate a safe…

Selective Prediction is the task of rejecting inputs a model would predict incorrectly on. This involves a trade-off between input space coverage (how many data points are accepted) and model utility (how good is the performance on accepted…

Robot task planning is an important problem for autonomous robots in long-horizon challenging tasks. As large pre-trained models have demonstrated superior planning ability, recent research investigates utilizing large models to achieve…

Adept traffic models are critical to both planning and closed-loop simulation for autonomous vehicles (AV), and key design objectives include accuracy, diverse multimodal behaviors, interpretability, and downstream compatibility. Recently,…

Safety alignment aims to ensure that large language models (LLMs) refuse harmful requests by post-training on harmful queries paired with refusal answers. Although safety alignment is widely adopted in industry, the overrefusal problem…

In practical scenarios where training data is limited, many predictive signals in the data can be rather from some biases in data acquisition (i.e., less generalizable), so that one cannot prevent a model from co-adapting on such…

Improving the safety of vision-language models like CLIP via fine-tuning often comes at a steep price, causing significant drops in their generalization performance. We find this trade-off stems from rigid alignment strategies that force…

Secure signal processing is becoming a de facto model for preserving privacy. We propose a model based on the Fully Homomorphic Encryption (FHE) technique to mitigate security breaches. Our framework provides a method to perform a Fast…

Service and assistive robots are increasingly being deployed in dynamic social environments; however, ensuring transparent and explainable interactions remains a significant challenge. This paper presents a multimodal explainability module…

This study reveals a previously unexplored vulnerability in the safety alignment of Large Language Models (LLMs). Existing aligned LLMs predominantly respond to unsafe queries with refusals, which often begin with a fixed set of prefixes…

Modern safety-critical energy infrastructures are increasingly operated in a hierarchical and modular control framework which allows for limited data exchange between the modules. In this context, it is important for each module to…

Reinforcement learning policies are typically represented by black-box neural networks, which are non-interpretable and not well-suited for safety-critical domains. To address both of these issues, we propose constrained normalizing flow…