About

Yang Cai is a Professor of Computer Science at Yale University, with additional appointments in Economics. He holds a Ph.D. from the Massachusetts Institute of Technology and a B.S. from Peking University. His research focuses on the theory of computation, economics and computation, optimization, and learning. Cai has received numerous awards including the COLT Best Paper Award in 2025, the FOCS Test of Time Award in 2022, an NSF CAREER Award in 2020, a Sloan Research Fellowship in 2019, and several fellowships and scholarships. His work involves characterizations and estimators for treatment effects, equilibria in non-concave games, convergence in multi-player learning, Bayesian mechanism design, and approximation of gains from trade in markets, among other topics. Cai is recognized for his contributions to the fields of algorithms, data, and market design, and is actively involved in advancing research at Yale's Center for Algorithms, Data, and Market Design.

Research topics

• Computer Science
• Artificial Intelligence
• Nanotechnology
• Algorithm
• Mathematical optimization
• Metallurgy
• Electrical engineering
• Chemical engineering
• Materials science
• Physics
• Combinatorics
• Photochemistry
• Microeconomics
• Mathematics
• Chemistry
• Optoelectronics
• Engineering
• Economics

Selected publications

• From Average-Iterate to Last-Iterate Convergence in Games: A Reduction and Its Applications

  ArXiv.org · 2025-06-04

  preprintOpen access1st authorCorresponding

  The convergence of online learning algorithms in games under self-play is a fundamental question in game theory and machine learning. Among various notions of convergence, last-iterate convergence is particularly desirable, as it reflects the actual decisions made by the learners and captures the day-to-day behavior of the learning dynamics. While many algorithms are known to converge in the average-iterate, achieving last-iterate convergence typically requires considerably more effort in both the design and the analysis of the algorithm. Somewhat surprisingly, we show in this paper that for a large family of games, there exists a simple black-box reduction that transforms the average iterates of an uncoupled learning dynamics into the last iterates of a new uncoupled learning dynamics, thus also providing a reduction from last-iterate convergence to average-iterate convergence. Our reduction applies to games where each player's utility is linear in both their own strategy and the joint strategy of all opponents. This family includes two-player bimatrix games and generalizations such as multi-player polymatrix games. By applying our reduction to the Optimistic Multiplicative Weights Update algorithm, we obtain new state-of-the-art last-iterate convergence rates for uncoupled learning dynamics in multi-player zero-sum polymatrix games: (1) an $O(\frac{\log d}{T})$ last-iterate convergence rate under gradient feedback, representing an exponential improvement in the dependence on the dimension $d$ (i.e., the maximum number of actions available to either player); and (2) an $\widetilde{O}(d^{\frac{1}{5}} T^{-\frac{1}{5}})$ last-iterate convergence rate under bandit feedback, improving upon the previous best rates of $\widetilde{O}(\sqrt{d} T^{-\frac{1}{8}})$ and $\widetilde{O}(\sqrt{d} T^{-\frac{1}{6}})$.

  PublisherOA PDFDOI

• Multi-Dimensional Screening with Endogenous Information Disclosure

  ArXiv.org · 2025-02-25

  preprintOpen access1st authorCorresponding

  We study multi-product monopoly pricing where the seller jointly designs the selling mechanism and the information structure for the buyer to learn his values. Unlike the case with exogenous information, we show that when the seller controls information, even uniform pricing guarantees at least half of the optimal revenue. Moreover, for negatively affiliated or exchangeable value distributions, deterministic pricing is revenue-optimal. Our results highlight the power of information design in making pricing mechanisms approximately optimal in multi-dimensional settings.

  PublisherOA PDFDOI

• Clinical Efficacy of 5-Aminolevulinic Acid Photodynamic Therapy in the Treatment of Oral Leukoplakia: A Comparative Study with Long-Term Follow-Up

  SSRN Electronic Journal · 2025-01-01

  preprintOpen access
  PublisherDOI

• Single-Panel Digital Twin Motion Based on Unity3D Large Reflective Surface Antenna Actuator

  2025-05-23

  article

  Based on the Unity3D engine, the digital twin motion of a single panel in a large reflecting surface antenna is investigated. By building a high-precision 3D model and combining the actual motion characteristics of the antenna panel, it is hoped to simulate the motion behaviour of the panel using the physics engine and animation system of Unity3D. In order to synchronise the digital twin with the actual physical system, a real-time program-driven approach is used to dynamically reflect the motion state and positional changes of the panel. The system not only provides visual support for antenna design and optimisation, but can also be used for fault diagnosis and performance prediction, providing an efficient and intuitive digital solution for the operation and maintenance management of large reflector antennas.

  PublisherDOI

• Unprecedented large-scale aquifer recovery through human intervention

  Nature Communications · 2025-08-07 · 44 citations

  articleOpen access

  Groundwater depletion is a critical global challenge, particularly in intensively cultivated drylands, with few documented cases of successful recovery. Here, we report a striking reversal of long-term groundwater decline in the North China Plain, one of the world’s most severely depleted aquifers. Based on a comprehensive analysis of groundwater levels from over 2000 monitoring wells spanning the past two decades, we show that groundwater levels have risen at an average rate of ~0.7 m year−1 since 2020, surpassing 2005 levels by 2024. This recovery is driven by a combination of large-scale surface water diversion from the humid south and stringent groundwater pumping regulations, further amplified by wet years (e.g., 2021). From 2005 to 2023, these policies reduced annual groundwater abstraction by ~12 km3 and increased environmental water allocations to over 7 km3 since 2021, promoting aquifer recharge and restoring environmental flows. Our findings demonstrate that rapid, large-scale groundwater recovery is achievable through integrated water management and targeted policy interventions across extensive regions (~130,000 km2). Unprecedented groundwater recovery ( ~ 0.7 m/yr) driven by water diversions, strict pumping regulations, and a wet climate occurred in the North China Plain after decades of depletion, showing large-scale recovery is possible under human intervention.

  PublisherOA PDFDOI

• Diffusion Tensor Magnetic Resonance Image Registration Based on Parallel Dual-Channel VoxelMorph

  IEEE Journal of Biomedical and Health Informatics · 2025-01-01

  article

  Diffusion Tensor Magnetic Resonance Imaging (DTI) is a non-invasive technique for studying brain structure in vivo by measuring the diffusion properties of water molecules. Unlike conventional medical imaging that captures scalar intensity data, DTI data is typically stored as a 4D volume, where each voxel in 3D space is a 3×3 Cartesian tensor. DTI characterizes tensor-based diffusion profiles and captures information about the orientation of fiber bundles. During the alignment process, voxels need to be spatially transformed while maintaining the correspondence of tensor orientations, which leads to complex computations. Traditional DTI registration methods often suffer from slow iteration speed and low accuracy, posing challenges for clinical applications. In this paper, a novel DTI Registration method Based on Parallel Dual-channel Voxel Morph (DTI-RBPDV) is proposed. The core of the method is a two-branch convolutional neural network architecture. With a view to enhancing the alignment performance, it processes two input patterns simultaneously: (1) fractional anisotropy (FA) images and (2) principal eigenvectors from to-be-aligned and fixed DTI volumes to enhance the accuracy of deformation field prediction. In the network decoder layer, integration of attention mechanisms has also been implemented. These channel space attention modules dynamically highlight salient anatomical features and orientation consistency, improving the model's sensitivity to key structural alignments. Experimental results show that DTI-RBPDV effectively addresses the limitations of slow iterative computation and the challenges of applying deep learning to high-dimensional DTI data by significantly improving the registration accuracy and computational speed.

  PublisherDOI

• Proximal Regret and Proximal Correlated Equilibria: A New Tractable Solution Concept for Online Learning and Games

  ArXiv.org · 2025-11-03

  preprintOpen access1st authorCorresponding

  Learning and computation of equilibria are central problems in game theory, theory of computation, and artificial intelligence. In this work, we introduce proximal regret, a new notion of regret based on proximal operators that lies strictly between external and swap regret. When every player employs a no-proximal-regret algorithm in a general convex game, the empirical distribution of play converges to proximal correlated equilibria (PCE), a refinement of coarse correlated equilibria. Our framework unifies several emerging notions in online learning and game theory-such as gradient equilibrium and semicoarse correlated equilibrium-and introduces new ones. Our main result shows that the classic Online Gradient Descent (GD) algorithm achieves an optimal $O(\sqrt{T})$ bound on proximal regret, revealing that GD, without modification, minimizes a stronger regret notion than external regret. This provides a new explanation for the empirically superior performance of gradient descent in online learning and games. We further extend our analysis to Mirror Descent in the Bregman setting and to Optimistic Gradient Descent, which yields faster convergence in smooth convex games.

  PublisherOA PDFDOI

• The Power of Two-Sided Recruitment in Two-Sided Markets

  2024-06-10 · 1 citations

  articleOpen access1st authorCorresponding

  We consider the problem of maximizing the gains from trade (GFT) in two-sided markets. The seminal impossibility result by Myerson and Satterthwaite (1983) shows that even for bilateral trade, there is no individually rational (IR), Bayesian incentive compatible (BIC) and budget balanced (BB) mechanism that can achieve the full GFT. Moreover, the optimal BIC, IR and BB mechanism that maximizes the GFT is known to be complex and heavily depends on the prior. In this paper, we pursue a Bulow-Klemperer-style question, i.e., does augmentation allow for prior-independent mechanisms to compete against the optimal mechanism? Our first main result shows that in the double auction setting with m i.i.d. buyers and n i.i.d. sellers, by augmenting O(1) buyers and sellers to the market, the GFT of a simple, dominant strategy incentive compatible (DSIC), and prior-independent mechanism in the augmented market is at least the optimal in the original market, when the buyers’ distribution first-order stochastically dominates the sellers’ distribution. The mechanism we consider is a slight variant of the standard Trade Reduction mechanism due to McAfee (1992). For comparison, Babaioff, Goldner, and Gonczarowski (2020) showed that if one is restricted to augmenting only one side of the market, then n(m + 4√m) additional agents are sufficient for their mechanism to beat the original optimal and ⌊ log2 m ⌋ additional agents are necessary for any prior-independent mechanism. Next, we go beyond the i.i.d. setting and study the power of two-sided recruitment in more general markets. Our second main result is that for any ε > 0 and any set of O(1/ε) buyers and sellers where the buyers’ value exceeds the sellers’ value with constant probability, if we add these additional agents into any market with arbitrary correlations, the Trade Reduction mechanism obtains a (1−ε)-approximation of the GFT of the augmented market. Importantly, the newly recruited agents are agnostic to the original market.

  PublisherOA PDFDOI

• Field-of-View Studies for Stereo Interactions

  AHFE international · 2024-01-01 · 2 citations

  article1st authorCorresponding

  Rapidly growing extended reality technologies brought us more stereo interfaces and the costs became more affordable. We anticipated a rapid adaptation of the stereo display in everyday life and professional practice. Unfortunately, similar to 3D TVs, stereo displays have not been widely used in real-time. In this study, we want to know how the users interact with the stereo display systems, say, “stereo interaction.” We focus on the factors of field-of-view and distance to the screen. We found the field of view of many well-known stereo displays is very limited. For certain devices, only one location is for the user, not for the team to view the 3D effects. We also studied the stability of the 3D image in different lighting, poses, and motion. Finally, we zoom into the laparoscopic surgery training and summarize our findings.

  PublisherDOI

• Multi-Scale Semantic Segmentation with Modified MBConv Blocks

  arXiv (Cornell University) · 2024-02-07

  preprintOpen access

  Recently, MBConv blocks, initially designed for efficiency in resource-limited settings and later adapted for cutting-edge image classification performances, have demonstrated significant potential in image classification tasks. Despite their success, their application in semantic segmentation has remained relatively unexplored. This paper introduces a novel adaptation of MBConv blocks specifically tailored for semantic segmentation. Our modification stems from the insight that semantic segmentation requires the extraction of more detailed spatial information than image classification. We argue that to effectively perform multi-scale semantic segmentation, each branch of a U-Net architecture, regardless of its resolution, should possess equivalent segmentation capabilities. By implementing these changes, our approach achieves impressive mean Intersection over Union (IoU) scores of 84.5% and 84.0% on the Cityscapes test and validation datasets, respectively, demonstrating the efficacy of our proposed modifications in enhancing semantic segmentation performance.

  PublisherOA PDFDOI

Recent grants

• CAREER: Towards a Robust Theory of Mechanism Design

  NSF · $600k · 2020–2024

Frequent coauthors

• Constantinos Daskalakis

  30 shared

• S. Matthew Weinberg

  Princeton University

  26 shared

• Mingfei Zhao

  Google (United States)

  16 shared

• Nikhil R. Devanur

  13 shared

• Argyris Oikonomou

  11 shared

• Grigoris Velegkas

  8 shared

• Wei Yuan

  Tohoku University

  8 shared

• Yi Yang

  8 shared

Labs

• Yang Cai's LabPI

  Not provided

Education

• Ph.D., EECS

  Massachusetts Institute of Technology

  2013

• Bachelor of Science, EECS

  Peking University

  2008

Awards & honors

• COLT Best Paper Award (2025)
• FOCS Test of Time Award (2022)
• NSF CAREER Award (2020)
• Sloan Research Fellowship (2019)
• William Dawson Scholar (2015)