Han Wang

The paper introduces FP-Agent, a classifier that demonstrates that while browser fingerprints are poor discriminators for AI browsing agents, behavioral fingerprints (like typing and scrolling patterns) are highly effective for distinguishing these agents from humans and from each other.

The paper introduces CoT-Guard, a small, cost-effective 4B-parameter model that significantly outperforms large, expensive monitors like GPT-5 in detecting hidden objectives in code generation tasks.

The paper introduces 'infilling extraction' to accurately model training data memorization in Diffusion Language Models (DLMs), finding that bidirectional masking significantly increases the extractability of verbatim training data compared to traditional prefix-only methods.

This paper analyzes the limitations of Counterfactual Knowledge Training (CFT) for LLM unlearning, identifying knowledge conflict and hallucination spillover as major pitfalls that hinder its effectiveness.

ZipRL introduces an adaptive context compression framework that significantly improves the performance and efficiency of LLMs in complex, multi-turn agent tasks by combining multi-granularity compression with Hindsight Response Replay.

The paper introduces a novel paradigm where a fine-tuned LLM acts as an ancillary predictor to forecast likely advertisers, significantly improving ad recommendation systems by augmenting candidate generation and providing priors for downstream ranking.

This paper proposes four guidelines and two novel data ordering methods (STR and SAW) to systematically optimize data organization, significantly enhancing the stability and performance of LLM training.

The paper addresses the challenge of multi-turn view planning for VLMs by proposing an iterative framework that uses self-exploration and view graph distillation, significantly improving planning performance over state-of-the-art models.

This paper introduces the concept of Budget-Aware Agents (BAGEN), showing that current LLM agents often fail to manage resources proactively, and proposes that incorporating early stop and interval estimation significantly improves efficiency.

The paper proposes EAGLE, a novel evidence-aligned multi-agent framework, demonstrating that requiring shared visual evidence among agents is crucial for achieving reliable and trustworthy consensus in multimodal Visual Question Answering (VQA).

The paper models healthcare mechanism design as program synthesis, demonstrating that an optimized, mixed-objective program can eliminate up-coding and reduce patient rejection while maintaining financial viability.

This paper proposes two horizon-control strategies, Progressive OPD (POPD) and Truncated OPD (TOPD), demonstrating that full rollouts are often unnecessary for On-Policy Distillation, leading to significant improvements in training efficiency.

The paper introduces RouteGuard, a router-expert framework, to improve the robustness and generalization of safety guardrails by specializing threat detection across multiple distinct unsafe categories.

The paper introduces RouteGuard, a router-expert framework, to improve the robustness and generalization of safety guardrails by specializing threat detection across multiple unsafe categories.

The paper introduces Dr. DocBench, a difficulty-aware, comprehensive benchmark designed to rigorously test expert-level and challenging document parsing capabilities for VLMs, demonstrating that current state-of-the-art models fail on complex, domain-specific structures.

The paper introduces RefMem-Bench, a new benchmark for measuring reflective memory in long-horizon dialogue, and proposes REMIND, a framework that significantly improves models' ability to synthesize fragmented cues into high-level interpretations.

The paper proposes GloResNet, a lightweight 3D CNN that effectively predicts brain injury in preterm infants using T2-weighted MRI, achieving an average accuracy of 75.18%.

The paper introduces AutoMedBench, a novel workflow-aware benchmark that evaluates autonomous medical-AI agents across a five-stage research process, revealing that agents struggle most with validation and submission.

The paper introduces RadioMaster, a novel multi-agent system that successfully translates high-level user intents into physically viable, real-world radio signals, significantly outperforming existing methods.

The paper introduces the threat model of sequential data poisoning, demonstrating that multiple, collaborating attackers can exploit compound vulnerabilities in LLM post-training pipelines that are invisible when analyzing individual stages.