← Search

Xiang Yang

7 accepted papers

2026

Federated Multi-view Clustering for Remote Sensing Data

ICML 2026poster

The rapid expansion of remote sensing technology has generated massive amounts of unlabeled multi-view data distributed across different institutions. Analyzing this data presents significant challenges, as centralized processing incurs prohibitive communication costs and raises data privacy concern…

Cited by 0SourceScholar
2026

HURC-MM: Holistic Uncertainty-Responsive Control for Mobile Manipulator Grasping

RA-L 2026

We present a holistic uncertainty-responsive control system that enables mobile manipulators to perform human- like adaptive grasping by dynamically coordinating base-manipulator motion and uncertainty-aware decision-making. Our key innovations include: (1) a continuous visual monitoring strategy: D

Cited by 0SourceScholar
2026

M-LoRA: Efficient Serving for Concurrent LoRA Adapters with Memory-Aware Speculative Scheduler on Single GPU

IJCAI 2026

Low-Rank Adaptation (LoRA) is a popular approach that enables large language models (LLMs) to quickly adapt to domain-specific tasks by adding lightweight trainable adapters. Existing multi-LoRA serving systems typically exploit parameter sharing to serve hundreds of LoRA models with a single base m

Cited by 0Scholar
2026

SafeRoPE: Risk-specific Head-wise Embedding Rotation for Safe Generation in Rectified Flow Transformers

CVPR 2026

Recent Text-to-Image (T2I) models based on rectified-flow transformers (e.g., SD3, FLUX) achieve high generative fidelity but remain vulnerable to unsafe semantics, especially when triggered by multi-token interactions. Existing mitigation methods largely rely on fine-tuning or attention modulation

Cited by 0SourcecodeScholar
2026

Unified Safe In-context Image Generation in Multimodal Diffusion Transformers

ICML 2026poster

Diffusion transformers (DiTs) equipped with multimodal attention (MM-Attn) have become a dominant paradigm for image generation. However, preventing the generation of harmful content remains a critical challenge, particularly in imageto-image (I2I) editing tasks. Existing safety mechanisms are prima…

Cited by 0SourceScholar
2024

HPipe: Large Language Model Pipeline Parallelism for Long Context on Heterogeneous Cost-effective Devices

NAACL 2024industry

Micro-enterprises and individual developers emerge analysis demands for long sequence with powerful Large Language Models (LLMs). They try to deploy the LLMs at local, but only possess various commodity devices and the unreliable interconnection between devices. Existing parallel techniques do not l…

Cited by 6SourcePDFScholar