Skip to main content

Human-Centered 

 

Research Quick Takes

Posing with posters at NeurIPS

Dec 18, 2025

LISP Lab at NeurIPS

Three members of Professor Peter Chin's LISP Lab—PhD students Mai Pham and Junyan Cheng, and post-doc Xavier Cadet—presented at the Thirty-Ninth Annual Conference on Neural Information Processing Systems (NeurIPS 2025) which drew a record-breaking 26,000 attendees. Their presentations addressed optimal auction design, multi-agent cooperation, and language models for autonomous scientific discovery.

Ruixu (Rachel) Huang

Dec 11, 2025

Guide for Generating Spatial Data

PhD student Ruixu (Rachel) Huang is a co-lead author of "Systematic benchmarking of imaging spatial transcriptomics platforms in FFPE tissues" published in Nature Communications. A collaboration between the Goods Lab and the Broad Institute of MIT and Harvard, the study is the first to compare commercial platforms for generating spatial data.

A printed solar absorber

Dec 04, 2025

Better Printed Solar Cells

Postdoc Yanan Li, PhD students Julia Huddy and Masha Klymenko, and Professor Will Scheideler coauthored "Spatial-Uniformity–Driven Bayesian Optimization for Rapid Development of Printed Perovskite Solar Cells" published in Small. (This came out of work recently funded by DOE in Scheideler's SENSE Lab.) "Metal halide perovskites are a promising emerging solar technology, but challenges in reliability and large‑area scalability still hinder widescale adoption. This work uses a machine‑learning–driven Bayesian optimization approach to improve the uniformity of printed perovskite films—addressing a key bottleneck for scaling low‑cost, roll‑to‑roll manufacturing and enabling higher‑efficiency, more reliable solar cells," said Scheideler.

The three study coauthors

Nov 20, 2025

Toward Optimal Auctions

PhD student Mai Pham, will present her paper, coauthored with professors Vikrant Vaze and Peter Chin, titled "Advancing Differentiable Mechanism Design: Neural architectures for combinatorial auctions" for a workshop at the Conference on Neural Information Processing Systems. Although auctions are considered an effective way of allocating limited resources when demand is high, designing auctions that are simultaneously optimal for the participants, system operator, and greater society is challenging. The paper presents a new approach that leverages modern deep learning architectures and algorithms to meet this challenge.

Instagram logo