Yexin Lai — accepted AI-conference papers

2026

PromptPilot: Game-Theoretic Multi-Agent Prompt Optimization for Segment Anything

Guangze Shi, Yingjie Mi, Jia Shen, Feixue Shao, Jiarui Cao, Yexin Lai +4

ICML 2026poster

Optimizing prompts for foundation models like SAM represents a challenging high-dimensional black-box optimization problem, fundamentally plagued by the credit assignment ambiguity. To address this, we introduce PromptPilot, a task-agnostic reinforcement learning framework that structurally decompos…

Cited by 0 Source ↗Scholar ↗

2025

Plug-and-Play PPO: An Adaptive Point Prompt Optimizer Making SAM Greater

Xueyu Liu, Rui Wang, Yexin Lai, Guangze Shi, Feixue Shao, Fang Hao +4

CVPR 2025poster

Powered by extensive curated training data, the Segment Anything Model (SAM) demonstrates impressive generalization capabilities in open-world scenarios, effectively guided by user-provided prompts. However, the class-agnostic characteristic of SAM renders its segmentation accuracy highly dependent…

Cited by 0 Source ↗PDF ↗code ↗Scholar ↗