Zi-Hao Huang

Achieve "What You See Is What You Get" fidelity in single-image 3D texturing by decoupling material estimation and using observation latents as fixed anchors.

Parsing Korean requires a nuanced approach, as fine-grained morphological details significantly enhance accuracy over simpler representations.

Eliminating arity actions in constituency parsing leads to a simpler, more efficient parser that competes with traditional methods while maintaining accuracy.

OmniView-Space redefines spatial reasoning in MLLMs, achieving unprecedented accuracy by leveraging dynamic, egocentric evidence mapping.

iLLaDA's fully bidirectional diffusion training outperforms traditional autoregressive models, achieving remarkable gains across key language benchmarks.

Despite the advancements in multimodal agents, even the best models struggle with interactive spatial reasoning, achieving only a 17.4% success rate in complex real-world tasks.

Arc-standard dependency parsing isn't just about graphs; it's secretly building ordered trees, unlocking a new perspective on projectivity and parsing.

Zero-shot synthesis of articulated human-object interactions is now possible by treating diffusion-generated videos as supervision for 4D scene reconstruction, unlocking physically grounded interactions beyond rigid manipulation.