Withings first announced its U-Scan, a non-invasive device you install in your toilet to track various health metrics, at CES 2023 two years ago. Unlike Kohler’s Dekoda, which uses a camera to analyze your poop, the U-Scan detects and samples…
arXiv:2501.00919v2 Announce Type: replace Abstract: Representational similarity analysis (RSA) is widely used to analyze the alignment between humans and neural networks; however, conclusions based on this approach can be misleading without considering the underlying representational geometry. Our work introduces a…
arXiv:2502.21142v2 Announce Type: replace-cross Abstract: Humans leverage rich internal models of the world to reason about the future, imagine counterfactuals, and adapt flexibly to new situations. In Reinforcement Learning (RL), world models aim to capture how the environment evolves in…
arXiv:2510.23631v1 Announce Type: new Abstract: Alignment of large language models (LLMs) has predominantly relied on pairwise preference optimization, where annotators select the better of two responses to a prompt. While simple, this approach overlooks the opportunity to learn from richer…
arXiv:2510.23632v1 Announce Type: new Abstract: The rapid growth of high-resolution scientific simulations and observation systems is generating massive spatiotemporal datasets, making efficient, error-bounded compression increasingly important. Meanwhile, decoder-only large language models (LLMs) have demonstrated remarkable capabilities in modeling complex sequential…
arXiv:2510.01161v2 Announce Type: replace Abstract: Reinforcement learning has been central to recent advances in large language model reasoning, but most algorithms rely on on-policy training that demands fresh rollouts at every update, limiting efficiency and scalability. Asynchronous RL systems alleviate…
arXiv:2510.23633v1 Announce Type: new Abstract: Pretrained diffusion models have demonstrated strong capabilities in zero-shot inverse problem solving by incorporating observation information into the generation process of the diffusion models. However, this presents an inherent dilemma: excessive integration can disrupt the…
arXiv:2510.22931v2 Announce Type: replace Abstract: Continual Learning (CL) is essential for enabling self-evolving large language models (LLMs) to adapt and remain effective amid rapid knowledge growth. Yet, despite its importance, little attention has been given to establishing statistical reliability guarantees…
arXiv:2510.23634v1 Announce Type: new Abstract: Motivated by applications for set containment problems, we consider the following fundamental problem: can we design set-to-vector functions so that the natural partial order on sets is preserved, namely $Ssubseteq T text{ if and only…
arXiv:2508.06041v3 Announce Type: replace Abstract: How can we effectively handle queries for on-device large language models (LLMs) with varying runtime constraints, such as latency and accuracy? Multi-scale quantization addresses this challenge by enabling memory-efficient runtime model adaptation of LLMs through…
