Haozhu Wang 

I'm an AI researcher working on foundation models at AWS. Besides my research, I enjoy collaborating with domain experts to solve challenging problems in physical sciences and healthcare. My long-term research goal is to create ML systems that can assist human experts in scientific discoveries and innovations.

Email  /  CV  /  Google Scholar  /  GitHub  /  LinkedIn / 

[April-2023] We announced OptoGPT, the first foundation model for optical multilayer thin film design. Check out the arXiv preprint!

[April-2023] I will serve as a reviewer for NeurIPS 2023 and Synergy of Scientific and Machine Learning Modelling Workshop @ ICML 2023!

[Feb-2023] I will serve as a reviewer for ICML 2023!

[Feb-2023] Our survey article Structural color generation: from layered thin films to optical metasurfaces has been published on Nanophotonics!

[Aug-2022] Our paper Dynamic prediction of work status for workers with occupational injuries: assessing the value of longitudinal observations has been published on Journal of the American Medical Informatics Association!

[Aug-2022] Our AWS Machine Learning blog on real-time fraud detection is online.

[July-2022] I served as a session chair in ICML 2022.

[Mar-2022] I joined Amazon ML Solutions Lab as a Research Scientist.

[Jan-2022] I defended my thesis Learning to Optimize: Applications in Physical Designs and Manufacturing on Jan 14th, 2022.

We developed OptoGPT, the first foundation model for optical thin film structure inverse design. After being trained on a large dataset of 10 million optical thin film designs, OptoGPT demonstrates remarkable capabilities including: 1) autonomous global design exploration, 2) efficient designs for various tasks, 3) the ability to output diverse designs, and 4) seamless integration of user-defined constraints. We believe OptoGPT is a major leap towards accelerating optical science with foundation models.

Comprehensive survey of the structural color research field. I provided a discussion of applying machine learning to structural color device designs.

We developed a forecasting model to predict return-to-work after occupational injuries based on longitudinal claim data. The model may allow case managers to better allocate medical resources and help speed up patients' recover process.

We propose a hybrid machine learning and optimization method that combines mixture density networks and particle swarm optimization for accurate and efficient structural color inverse design.

We provide extensive benchmarking results on accuracy, diversity, robustness for commonly used deep learning models in nanophotonic inverse designs. The findings can help researchers select models that best suit their design problems.

We provide an analytical approach for evaluating the robustness of photonic integrated circuits. The method is verified by genetic algorithms.

Training a novel sequence generation network with Proximal Policy Optimization for automatically discovering near-optimal optical designs.

Apply LSTM and LR models for spatio-temporal COVID-19 risk prediction.

Apply Q-learning to insurance claim data to learn near-optimal dynamic treatment regimes.

Expert-yielded-estimates regularizer for incorporating expert knowledge into linear models.

Group-ordered-weighted lasso (GrOWL) for deep model compression.

Surface plasmon polariton laser with a novel device structure.

Superconduting nanowire single photon detector as a highly-sensitive imager.

The source code of this website is from Jon Barron.

(last update: Sep 2022)