arXiv:2605. 29076v2 Announce Type: replace-cross Abstract: LLMs have advanced text classification, yet existing paradigms face a trade-off: supervised (label only) fine-tuning is scalable but offers limited reasoning on complex text and lacks broader model transparency, while discrete prompt optimization offers human-readable instructions but struggles with performance and scalability.
By Tianyang Zhou, Wenbo Chen, Pierre Jinghong Liang, Leman Akoglu
arXiv:2606. 04326v1 Announce Type: cross Abstract: Concept bottleneck models predict outcomes from high-level concepts detected in inputs.
By Julian Skirzynski, Harry Cheon, Shreyas Kadekodi, Meredith Stewart, Berk Ustun
arXiv:2601. 06196v3 Announce Type: replace-cross Abstract: Large language models (LLMs) frequently generate factually incorrect or unsupported content, commonly referred to as hallucinations.
By Bodla Krishna Vamshi, Rohan Bhatnagar, Haizhao Yang
arXiv:2606. 04767v1 Announce Type: new Abstract: The robustness of deep neural networks is crucial for safety-critical deployments, yet existing evaluation methods are often attack-dependent and lack interpretability.
By Chong Zhang, Xiang Li, Jia Wang, Qiufeng Wang, Xiaobo Jin
arXiv:2606. 04777v1 Announce Type: new Abstract: Clustering is increasingly used to support high-impact decisions, yet standard objectives such as $k$-means can produce clusterings that treat demographic groups unequally.
By Antonia Karra, Vasiliki Papanikou, Georgios Vardakas, Evaggelia Pitoura, Aristidis Likas
arXiv:2606. 04092v1 Announce Type: cross Abstract: Flow matching models learn to transport samples from a simple prior distribution to a complex data distribution.
By Shimon Malnick, Matan Rusanovsky, Ohad Fried, Shai Avidan
Large language models (LLMs) are rigorously aligned to refuse harmful requests, a process that inherently cultivates a latent capacity to evaluate and recognize unsafe content. In this work, we reveal that this advanced safety awareness inadvertently introduces a fatal vulnerability.
To leverage the full potential of multimodal data, we need representations that go beyond the state-of-the-art alignment and fusion approaches and exploit all cross-modal interactions without sacrificing modality-specific information. Learning disentangled representations is a principled way to identify these underlying shared and unique factors that are hidden in observational data.
When an AI agent calls an API and hits a validation error, it needs more than what went wrong -- it needs what to do next. A self-reflective API returns, on validation failure, a machine-readable recovery\_feedback.
Reinforcement learning with verifiable rewards (e. g.
Mitigating social bias in Large Language Models (LLMs) presents a distinct alignment challenge: unlike verifiable tasks, bias lacks a single ground truth, creating a high-variance, subjective reward landscape. Previous preference-based fine-tuning methods have major trade-offs: Direct Preference Optimization (DPO) is limited by the lack of exploration inherent in offline training, while Proximal Policy Optimization (PPO) can lead to training instability due to potentially unreliable critic estimates.
Generative models -- diffusion and flow matching -- are increasingly used to solve partial differential equation (PDE) inverse problems, enforcing the governing physics as a \emph{hard constraint} (via projection or guidance) and reporting the resulting samples as a Bayesian posterior with calibrated uncertainty. We show that this widely adopted recipe samples the wrong distribution.
Temporal credit assignment is central to both biological and artificial intelligence, yet its interaction with non-linear function approximation is poorly understood. We identify a systematic failure mode in deep reinforcement learning (RL) termed Trace-Mediated Peak Bias (TMPB).
OpenAI outlines a blueprint for U. S.
OpenAI outlines its public policy agenda for AI, including safety, youth protection, workforce transition, and global standards to ensure AI benefits society.
Vision-Language Models (VLMs) struggle when applied to medical image-text data, yet the tools available to diagnose this failure remain limited. Existing representation alignment metrics are symmetric, collapsing both modalities into a single score and hiding which modality drives cross-modal degradation.
Multimodal Large Language Models (MLLMs) are increasingly used for video understanding, yet their reliability under multi-video inputs remains poorly understood. We study positional bias in multi-video summarization, where the quality of a per-video summary can change with the video's input slot even when the underlying content is unchanged.
arXiv:2606. 02860v1 Announce Type: cross Abstract: Catastrophic forgetting is often framed as a representational problem: after sequential training, a model appears to lose the features that supported performance on earlier tasks.
By Archie Chaudhury
arXiv:2606. 03262v1 Announce Type: new Abstract: Neural operators learn mappings between infinite-dimensional function spaces and provide a data-driven surrogate modeling paradigm for parametric partial differential equations (PDEs).
By Keke Wu, Yixuan Zhang, Jingrun Chen
arXiv:2605. 18629v2 Announce Type: replace Abstract: Sparse autoencoders (SAEs) are one of the main methods to interpret the inner workings of deep neural networks (DNNs), decomposing activations into higher-dimensional features.
By Micha{\l} Brzozowski, Neo Christopher Chung