Yian Ma

Assistant Professor@University of California, San Diego·美国·OpenReview

研究方向

Bayesian inference · MCMC · stochastic gradient MCMC · Scalable inference · stochastic gradient methods · variance reduction · Time series · hidden Markov model · state space models · Sequential decision making

Yian Ma

Zephyrus: An Agentic Framework for Weather Science

Beyond Length: Quantifying Long-Range Information for Long-Context LLM Pretraining Data

On the Complexity Theory of Masked Discrete Diffusion: From $\mathrm{poly}(1/\epsilon)$ to Nearly $\epsilon$-Free

Functional-level Uncertainty Quantification for Calibrated Fine-tuning on LLMs

SimulRAG: Simulator-based RAG for Grounding LLMs in Long-form Scientific QA

Theory of Autoregressive Diffusion Model: Inference Complexity and Conditional Dependency Learning

LaDiR: Latent Diffusion Enhances LLMs for Text Reasoning

Nearly Dimension-Independent Convergence of Mean-Field Black-Box Variational Inference

Purifying Approximate Differential Privacy with Randomized Post-processing

Efficiently Scaling LLM Reasoning Programs with Certaindex

Preference Optimization on Pareto Sets: On a Theory of Multi-Objective Optimization

ClimaQA: An Automated Evaluation Framework for Climate Question Answering Models

A Skewness-Based Criterion for Addressing Heteroscedastic Noise in Causal Discovery

Learning to Steer Learners in Games

Functional-level Uncertainty Quantification for Calibrated Fine-tuning on LLMs

Discovering Latent Causal Graphs from Spatiotemporal Data

Diff-BBO: Diffusion-Based Inverse Modeling for Black-Box Optimization

Discovering Latent Structural Causal Models from Spatio-Temporal Data