A novel method is presented that enables a semi-autonomous, camera-equipped robot to recognise the task goal and plan actions to achieve it, and a solver is developed that uses a vision-based Large Language Model to convert photos of a Shoebox Task into a Planning Domain Definition Language (PDDL) representation and ge...
Inference-time harnesses, i.e., external procedures that call a model many times and keep state across calls, can substantially improve model capabilities. In this paper, we investigate the benefits of using such harnesses and the extent to which these benefits can be internalized through training. More precisely, we c...
Jingchu Gai, Adam Block, Andrej Risteski· 0 citations
Neural networks (NNs) are increasingly deployed on edge devices such as mobile and IoT platforms, offering low latency and improved privacy compared to cloud-based deployment. However, on-device models embedded in mobile applications can be unpacked and exposed to adversaries, enabling cross-node attacks in centrally d...
Yu-Yang Zhang, Kangjie Chen, Yi-Hui Jin et al.· 0 citations
Data selection plays a central role in modern large language model training, as the quality of training data strongly influences downstream performance. A common approach is to curate a development set that encodes the desired skills (e.g., reasoning or multilinguality) and select data that are similar to them. Existin...
Yu-Qiao Wen, Li-Li Mou· 0 citations
Reach audiences
Advertise in front of researchers, engineers, and readers.
A self-supervised representation learning framework built on a simple hypothesis: behavioral characteristics associated with identity and learning remain relatively stable over nearby time intervals, which establishes short-term movement as an informative behavioral"fingerprint" for studying learning and individual dif...
Ping Zhang, Quang-Huy Nguyen, Ying-Tie Lei et al.· 0 citations
We encounter a system through its responses, yet its state concerns how the system is. Predicting further responses promises knowledge beyond what is immediately observed. What that knowledge tells us about the system's state depends on what we mean by a state in the first place. We give this notion a model-relative me...
Ping Zhang, Quang-Huy Nguyen, Yingtie Lei et al.· 0 citations
A novel method is proposed, BMO, that obfuscates model parameters to diversify the decision boundary further resisting cross-node attack while preserving its benign accuracy, and achieves comparable performance in benign accuracy compared with baselines.
Yu-Yang Zhang, Yi-Hui Jin, Kangjie Chen et al.· 0 citations
OptEP is developed, a scalable continuous relaxation that directly optimizes differentiable expert-selection variables, followed by a lightweight reconstruction-based update of the retained expert weights, demonstrating that reconstruction-driven expert selection and weight adaptation can preserve substantially more of...
Jelena Markovic-Voronov, P. Prastakos, Kayhan Behdin et al.· 0 citations
Retrieval-augmented QA can fail even when the correct answer is in the retrieved evidence. This can happen when a competing answer satisfies some, but not all, of the question's requirements and is mentioned frequently in the retrieved documents, which can make it appear to be the best-supported answer. To test whether...
Jennifer Hsia, Zachary Chase Lipton, Aarti Singh· Proceedings of the 2026 Conf...· 0 citations
MEDIATOR is introduced, a discrete ontology-agnostic tokenizer that maps any clinical concept expressed as text into a shared vector-quantized codebook, turning EHR harmonization into open-vocabulary tokenization without manual mappings or a fixed vocabulary.
Tong Ding, Cai-Wei Tian, Dongmin Bang et al.· 0 citations
Together, these results support qualified reuse of English-built pipelines, but per-domain, per-language validation remains necessary--neither semantic similarity nor self-reported confidence reliably predicts cross-lingual accuracy.
J. Choi, Sherin V. George, Jia-Yang Zhao et al.· 0 citations
This investigation reveals important limitations of the current methods that try to leverage internal VLM signals to detect hallucinations and suggests that strong detection performance is often driven by detecting language-based hallucinations, and when methods rely on token position, they tend to be less visually gro...
Allison Chen, William Yang, Salma Abdel Magid et al.· 0 citations
