Feng Liu
6 papers · Latest:
UniCorrn: Unified Correspondence Transformer Across 2D and 3D
UniCorrn unifies 2D-2D, 2D-3D, and 3D-3D geometric matching using a single Transformer model with shared weights, outperforming SOTA.
Bridge: Basis-Driven Causal Inference Marries VFMs for Domain Generalization
Bridge enhances object detection domain generalization by using basis-driven causal inference to block confounders and refine representations.
Green Shielding: A User-Centric Approach Towards Trustworthy AI
Green Shielding proposes a user-centric approach to build trustworthy AI by characterizing how benign input variations affect LLM behavior.
Discrete Preference Learning for Personalized Multimodal Generation
DPPMG learns discrete modal-specific preferences to generate personalized and consistent multimodal content from user interactions.
Combating Data Laundering in LLM Training
This paper introduces Synthesis Data Reversion (SDR) to effectively detect laundered proprietary data used in LLM training, even when original detection signals are erased.
Baichuan 2: Open Large-scale Language Models
Baichuan 2 is a series of large-scale, open-source multilingual language models that achieve state-of-the-art performance across general and specialized benchmarks.
📬 Weekly AI Paper Digest
Get the top 10 AI/ML arXiv papers from the week — summarized, scored, and delivered to your inbox every Monday.