4 papers
ECA: Efficient Continual Alignment for Open-Ended Image-to-Text Generation
Jiangtao Kong, Peijun Zhao, Chun-Fu Chen +4
Incremental Learning (IL) for Open-ended Image-to-Text Generation (OpenITG) enables models to continuously generate accurate, contextually relevant text for new images while preser…
WestWorld: A Knowledge-Encoded Scalable Trajectory World Model for Diverse Robotic Systems
Yuchen Wang, Jiangtao Kong, Sizhe Wei +6
Trajectory world models play a crucial role in robotic dynamics learning, planning, and control. While recent works have explored trajectory world models for diverse robotic system…
Explore the vulnerability of black-box models via diffusion models
Jiacheng Shi, Yanfu Zhang, Huajie Shao +1
Recent advancements in diffusion models have enabled high-fidelity and photorealistic image generation across diverse applications. However, these models also present security and…
Hybrid Memory Replay: Blending Real and Distilled Data for Class Incremental Learning
Jiangtao Kong, Jiacheng Shi, Ashley Gao +3
Incremental learning (IL) aims to acquire new knowledge from current tasks while retaining knowledge learned from previous tasks. Replay-based IL methods store a set of exemplars f…