Ying Liu

ViaMOBO reveals that understanding variable interactions can drastically enhance optimization efficiency in high-dimensional spaces, outperforming traditional methods.

Reward hacking can be mitigated with a simple one-line fix that improves out-of-distribution performance while keeping training robust.

PAC approximations can significantly enhance the efficiency of optimizing parametric Markov decision processes, achieving faster runtimes without sacrificing accuracy.

Achieving a 91.51% acceptable-answer rate in enterprise question answering reveals the potential of staged adaptation to balance proprietary knowledge acquisition with general capabilities.

Achieving state-of-the-art scene completion and generation without relying on partial scans, FPSGen redefines how point clouds can be generated in challenging environments.

SERPO achieves up to 20.63 points improvement on HealthBench by enabling language models to self-evolve their evaluation criteria in real-time without external feedback.

Achieving near-linear scalability in multi-robot task allocation while effectively managing environmental uncertainty could revolutionize how we deploy heterogeneous robotic systems in complex settings.

PolyUQuest achieves superior answer correctness and verifiability by integrating structural web data into its retrieval process, outperforming traditional RAG systems.

Semantic clustering transforms the Tsetlin Machine into a powerful yet interpretable model, rivaling BERT's performance without sacrificing transparency.

LLMs don't just reflect gender bias in public vs. private spaces; they encode nuanced, micro-level mappings that substantially exceed real-world distributions, shaping spatial gender narratives in unexpected ways.

By encoding physical priors of low-light degradation directly into the diffusion process, SADM achieves state-of-the-art low-light image enhancement and denoising in a single stage, eliminating the need for complex multi-stage pipelines.