Jiang Bian (Microsoft Research)

Jiang Bian, Ph.D.

Assistant Managing Director of a stealth mode Neo Lab

Previously Assistant Managing Director, Microsoft Research Asia

Personal Email: jiang.bian.prc@gmail.com

About

I am a Partner Research Manager at Microsoft Research. And, I am currently playing the Director of MSR Asia Industry Innovation Center (MIIC). I am leading this center with the mission of building up innovative and disruptive technologies as well as forward-looking insights to drive the digital transformation for important industries. We have been focusing on designing cutting-edge machine learning algorithms into real-world application scenarios, including finance, supply-chain, healthcare and sustainability.

Prior to that, I was a Senior Scientist in Yidian Inc., a startup company in China, where I have been working on recommendation and search problems. I also used to work at Yahoo! Labs and did a lot of studies on content optimization and personalization for Yahoo!'s key content modules as well as local content search and recommendation for Yahoo!'s local services.

I received my B.S. from Peking University (2006) and Ph.D. from Georgia Institute of Technology (2010), both in Computer Science. My dissertation was on information retrieval, data mining, and machine learning techniques on general Web search and social media search, advised by Prof. Hongyuan Zha.

I authored tens of academic research papers receiving hundreds of citations, filed a couple of U.S. patents, and served as PC member/Peer reviewer for a few international conferences and journals.

Selected Projects

News

[May 2025] Our paper "Policy Filtration for RLHF to Mitigate Noise in Reward Models" has been accepted by ICML 2025.
[May 2025] Our paper "BRIDGE: Bootstrapping Text to Control Time-Series Generation via Multi-Agent Iterative Optimization and Diffusion Modelling" has been accepted by ICML 2025.
[May 2025] Our paper "From Complex to Atomic: Enhancing Augmented Generation via Knowledge-Aware Dual Rewriting and Reasoning" has been accepted by ICML 2025.
[May 2025] Our paper "DPO Meets PPO: Reinforced Token Optimization for RLHF" has been accepted by ICML 2025.
[May 2025] Our paper "Text-to-CAD Generation Through Infusing Visual Feedback in Large Language Models" has been accepted by ICML 2025.
[May 2025] Our paper "CAD-Editor: A Locate-then-Infill Framework with Automated Training Data Synthesis for Text-Based CAD Editing" has been accepted by ICML 2025.
[May 2025] Our paper "Unveiling Markov heads in Pretrained Language Models for Offline Reinforcement Learning" has been accepted by ICML 2025.
[May 2025] Our paper "AdaptiveStep: Automatically Dividing Reasoning Step through Model Confidence" has been accepted by ICML 2025.