Alignment, interpretability, red-teaming, bias and privacy: the research on what these systems do when they misbehave.
arXiv:2607. 24551v1 Announce Type: new Abstract: Maintenance regulations are complex legal texts that are difficult to exploit when addressing a specific case and challenging to integrate into operational systems.
arXiv:2607. 22706v1 Announce Type: new Abstract: This paper presents the MPR-CiteG framework, which achieved second place in the ScienceON AI Challenge by addressing two fundamental challenges in generative AI: inefficient retrieval and the absence of source verification.
arXiv:2607. 22676v1 Announce Type: new Abstract: Post-training is a key mechanism for adapting large language models to downstream tasks.
arXiv:2607. 22926v1 Announce Type: new Abstract: High-impact generative AI makes catastrophic misuse a lifecycle-control problem, not merely a prompt-filtering problem.
arXiv:2607. 22727v1 Announce Type: cross Abstract: Medical image segmentation models often report high benchmark accuracy under ideal imaging conditions, yet their failures under clinical degradation can be quiet: sensor noise, patient motion, low- resolution acquisition, and contrast variability may all alter model behavior without producing an obvious warning.
arXiv:2607. 23147v1 Announce Type: cross Abstract: Large language models now power autonomous agents capable of complex, multi-step tasks in different environments.
arXiv:2607. 23934v1 Announce Type: new Abstract: Machine unlearning, which aims to remove the influence of specific training data from a trained model, is a key requirement for privacy, accountability, and adaptive deployment.
arXiv:2607. 23029v1 Announce Type: cross Abstract: Federated learning enables collaborative model training across distributed clients without centralising their data, yet privacy remains a persistent concern because the shared model updates can leak information about local datasets.
arXiv:2501. 19060v4 Announce Type: replace-cross Abstract: Vision-language models (VLMs), such as CLIP, adapt effectively to downstream tasks through prompt tuning, but fine-tuning can misalign predictive confidence and accuracy, particularly on unseen classes.
arXiv:2607. 22652v1 Announce Type: new Abstract: Recent research has explored the integration of knowledge graphs (KGs) with large language models (LLMs) to enhance their performance on downstream knowledge-intensive tasks, particularly knowledge graph question answering (KGQA).
arXiv:2607. 22694v1 Announce Type: new Abstract: Attention collapse in autoregressive language models -- manifested as repetitive token loops where the model becomes trapped in self-reinforcing attractors -- is a persistent pathology that existing decoding-time heuristics fail to address at its root cause.
arXiv:2604. 11151v2 Announce Type: replace Abstract: We develop parameter-free algorithms for unconstrained online learning with regret guarantees that scale with the gradient variation $V_T(u) = \sum_{t=2}^T \|\nabla f_t(u)-\nabla f_{t-1}(u)\|^2$.
arXiv:2607. 22712v1 Announce Type: cross Abstract: Single-cell light microscopy images have become an important data source for characterizing cell phenotypes, but their complexity and heterogeneity pose challenges to high-throughput automated analysis.
arXiv:2607. 22671v1 Announce Type: new Abstract: Foundation-model safety benchmarks capture the AI risks of their time of publication: as models improve and governments pass new AI-safety legislation, their risk taxonomies become incomprehensive and their attack prompts become ineffective.
arXiv:2607. 24243v1 Announce Type: new Abstract: Mainstream AI research emphasises capability growth and tolerates low failure rates when average-case performance is high.
arXiv:2607. 22716v1 Announce Type: cross Abstract: In this paper, we show for the first time that visual token pruning enhances the robustness of Multimodal Large Language Models (MLLMs), mitigating vulnerabilities such as jailbreak attacks and hallucinations.
arXiv:2607. 22925v1 Announce Type: cross Abstract: A key question for AI safety is whether a language model expresses all of its reasoning in its output tokens.
arXiv:2605. 14981v2 Announce Type: replace Abstract: Gromov--Wasserstein (GW) distances compare graphs, shapes, and point clouds through internal distances, without requiring a common coordinate system.
arXiv:2607. 22927v1 Announce Type: new Abstract: Weights and biases are normally optimized as separate parameter tensors, yet they do not represent separate functions when the input to an affine layer has nonzero mean.
arXiv:2607. 24522v1 Announce Type: new Abstract: While on-policy distillation (OPD) effectively addresses sparse rewards and exposure bias in large language model post-training, its extension to flow models remains underexplored.