11 papers
LTLDiff: Finite Linear Temporal Logic-Guided Data Generation and Diffusion Policies for Multi-agent Robotic Manipulation
Chuhan Meng, Haiyan Yin
Multi-agent robotic manipulation tasks require coordination among agents to satisfy task-level temporal, logical, and safety constraints. Recently, diffusion policies have been use…
Causal Scaffolding for Physical Reasoning: A Benchmark for Causally-Informed Physical World Understanding in VLMs
Tianyi Tang, Zhuoyi Lin, Zeyu Feng +4
Understanding and reasoning about the physical world is the foundation of intelligent behavior, yet state-of-the-art vision-language models (VLMs) still fail at causal physical rea…
Lifting Traces to Logic: Programmatic Skill Induction with Neuro-Symbolic Learning for Long-Horizon Agentic Tasks
Jie-Jing Shao, Haiyan Yin, Yueming Lyu +5
Foundation model-driven agents often struggle with long-horizon planning due to the transient nature of purely prompting-based reasoning. While existing skill induction methods mit…
Aligning LLMs with Graph Neural Solvers for Combinatorial Optimization
Shaodi Feng, Zhuoyi Lin, Yaoxin Wu +4
Recent research has demonstrated the effectiveness of large language models (LLMs) in solving combinatorial optimization problems (COPs) by representing tasks and instances in natu…
Hybrid Orchestration of Edge AI and Microservices via Graph-based Self-Imitation Learning
Chen Yang, Jin Zheng, Yang Zhuolin +4
Modern edge AI applications increasingly rely on microservice architectures that integrate both AI services and conventional microservices into complex request chains with stringen…
Bridging Synthetic and Real Routing Problems via LLM-Guided Instance Generation and Progressive Adaptation
Jianghan Zhu, Yaoxin Wu, Zhuoyi Lin +5
Recent advances in Neural Combinatorial Optimization (NCO) methods have significantly improved the capability of neural solvers to handle synthetic routing instances. Nonetheless,…
Evaluating LLMs Without Oracle Feedback: Agentic Annotation Evaluation Through Unsupervised Consistency Signals
Cheng Chen, Haiyan Yin, Ivor Tsang
Large Language Models (LLMs), when paired with prompt-based tasks, have significantly reduced data annotation costs and reliance on human annotators. However, evaluating the qualit…
MermaidFlow: Redefining Agentic Workflow Generation via Safety-Constrained Evolutionary Programming
Chengqi Zheng, Jianda Chen, Yueming Lyu +5
Despite the promise of autonomous agentic reasoning, existing workflow generation methods frequently produce fragile, unexecutable plans due to unconstrained LLM-driven constructio…
Mastering Continual Reinforcement Learning through Fine-Grained Sparse Network Allocation and Dormant Neuron Exploration
Chengqi Zheng, Haiyan Yin, Jianda Chen +3
Continual Reinforcement Learning (CRL) is essential for developing agents that can learn, adapt, and accumulate knowledge over time. However, a fundamental challenge persists as ag…
ExpertFlow: Efficient Mixture-of-Experts Inference via Predictive Expert Caching and Token Scheduling
Xin He, Shunkang Zhang, Kaijie Tang +8
Sparse Mixture-of-Experts (MoE) models can outperform dense large language models at similar computation by activating only a small set of experts per token. However, stacking many…
CASA: Bridging the Gap between Policy Improvement and Policy Evaluation with Conflict Averse Policy Iteration
Changnan Xiao, Haosen Shi, Jiajun Fan +2
We study the problem of model-free reinforcement learning, which is often solved following the principle of Generalized Policy Iteration (GPI). While GPI is typically an interplay…