9 papers · 1 filter
When Low Prediction Error Misleads Planning: Diagnosing Representation, Dynamics, and Decision Failures in Latent World Models
Rui Min, Xianyao Li, Fang Xu +1
The component that dominates a latent world model's prediction error need not be the one whose repair most improves action selection. We show this by comparing action sequences fro…
Minimal Recurrent Behavioral Memory for Imitation under Partial Observability
Xianyao Li, Fang Xu, Rui Min +2
What is the least recurrent memory needed to reproduce a specified expert under partial observability? The instantaneous requirement is the conditional entropy of the expert's beha…
Same World, Different Knowledge: When Isolated Audits Misjudge World-Model Repairs
Rui Min, Xianyao Li, Fang Xu +2
A repair favored under an isolated input fault can be inferior when deployed modules share the faulty information. We introduce an information-interface audit for world models, dis…
DreamLedger: Where to Refuse World-Model Imagination Using Execution-Settled Credit
Xianyao Li, Ruitong Tian, Rui Min +2
World-model predictions inform robot actions, yet instantaneous reliability signals do not retain the outcomes of comparable past predictions. DreamLedger registers consumed predic…
Some Large Language Models Exhibit Consistent Risk Attitudes
Bowen Sun, Rui Min, Yuxi Wang +3
As artificial intelligence systems are deployed in open-ended, high-stakes settings, a critical dimension remains unmeasured: how perceived risk is translated into action. We test…
Scalable Token-Level Hallucination Detection in Large Language Models
Rui Min, Tianyu Pang, Chao Du +2
Large language models (LLMs) have demonstrated remarkable capabilities, but they still frequently produce hallucinations. These hallucinations are difficult to detect in reasoning-…
RemoteZero: Geospatial Reasoning with Zero Labels
Liang Yao, Fan Liu, Shengxiang Xu +4
Geospatial reasoning requires models to identify image regions that satisfy complex and often implicit user intents. Recent reinforcement learning approaches improve reasoning with…
RemoteAgent: Bridging Vague Human Intents and Earth Observation with RL-based Agentic MLLMs
Liang Yao, Shengxiang Xu, Fan Liu +7
Earth Observation (EO) systems are essentially designed to support domain experts who often express their requirements through vague natural language rather than precise, machine-f…
GeoBrowse: A Geolocation Benchmark for Agentic Tool Use with Expert-Annotated Reasoning Traces
Xinyu Geng, Yanjing Xiao, Yuyang Zhang +5
Deep research agents integrate fragmented evidence through multi-step tool use. BrowseComp offers a text-only testbed for such agents, but existing multimodal benchmarks rarely req…