Author name cluster

Zun Li

Possible papers associated with this exact author name in Arrow. This page groups case-insensitive exact name matches and is not a full identity disambiguation profile.

13 papers

1 author row

AAAI Conference 2026 Conference Paper

CCAHCL: Multi-Level Hypergraph Contrastive Learning for Connected Component Awareness

Zhuo Li
Gengyu Lyu
Yuena Lin
Ziang Chen
Zhiyuan Ma
Zhen Yang
Zun Li

Hypergraph contrastive learning has emerged as a powerful unsupervised paradigm for hypergraph representation learning. Traditional hypergraph contrastive learning methods typically leverage neighbor aggregation strategy to obtain entity (node and hyperedge) representations within each connected component, and then utilize contrastive losses (e.g., node- or hyperedge-level) to update the encoders. However, since entities are usually focused equally on their respective losses, large connected components with numerous entities tend to provide a dominant contribution to the whole learning process, which inevitably hinders the effective learning of entity representations within small connected components. To address this issue, we propose a novel Connected-Component-Aware Hypergraph Contrastive Learning method (CCAHCL). Different from previous methods that only construct node or hyperedge representations, our method additionally constructs the connected component representations, and accordingly designs a hierarchical contrastive loss to balance the model's focus on different scales of connected components. Specifically, we first use the traditional neighbor aggregation strategy to aggregate and update entity (node and hyperedge) representations. Then, these entity representations are further aggregated to generate the connected component representations, where entity features are incorporated into connected components and their structural information is propagated back to enrich their corresponding entities. Afterwards, we employ node-level and hyperedge-level losses to learn the enriched entity representations, and further propose a novel connected-component-level contrastive loss to balance the model's focus on all different connected components, naturally avoiding the learning bias on large connected components. Extensive experiments on various datasets demonstrate that our proposed model achieves superior performance against other state-of-the-art methods.

PDF Details DOI

AAMAS Conference 2026 Conference Paper

Code-Space Response Oracles: Generating Interpretable Multi-Agent Policies with Large Language Models

Daniel Hennes
Zun Li
John Schultz
Marc Lanctot

Policy-Space Response Oracles (PSRO) have enabled the computation of approximate Nash equilibria in complex games. However, standard implementations rely on Deep Reinforcement Learning oracles, producing "black-box" neural network policies that are opaque, difficult to verify, and sample-inefficient. We introduce Code-Space Response Oracles (CSRO), a framework that tasks a Large Language Model (LLM) to synthesize code policies. CSRO reframes best-response computation as a code generation task, producing policies as executable, human-readable Python code. We demonstrate that CSRO, particularly when augmented with evolutionary refinement (AlphaEvolve), achieves performance competitive with baselines while offering superior interpretability and leveraging the LLM’s pretraining knowledge.

PDF

IJCAI Conference 2025 Conference Paper

Combining Deep Reinforcement Learning and Search with Generative Models for Game-Theoretic Opponent Modeling

Zun Li
Marc Lanctot
Kevin R. McKee
Luke Marris
Ian Gemp
Daniel Hennes
Paul Muller
Kate Larson

Opponent modeling methods typically involve two crucial steps: building a belief distribution over opponents' strategies, and exploiting this opponent model by playing a best response. However, existing approaches typically require domain-specific heurstics to come up with such a model, and algorithms for approximating best responses are hard to scale in large, imperfect information domains. In this work, we introduce a scalable and generic multiagent training regime for opponent modeling using deep game-theoretic reinforcement learning. We first propose Generative Best Respoonse (GenBR), a best response algorithm based on Monte-Carlo Tree Search (MCTS) with a learned deep generative model that samples world states during planning. This new method scales to large imperfect information domains and can be plug and play in a variety of multiagent algorithms. We use this new method under the framework of Policy Space Response Oracles (PSRO), to automate the generation of an offline opponent model via iterative game-theoretic reasoning and population-based training. We propose using solution concepts based on bargaining theory to build up an opponent mixture, which we find identifying profiles that are near the Pareto frontier. Then GenBR keeps updating an online opponent model and reacts against it during gameplay. We conduct behavioral studies where human participants negotiate with our agents in Deal-or-No-Deal, a class of bilateral bargaining games. Search with generative modeling finds stronger policies during both training time and test time, enables online Bayesian co-player prediction, and can produce agents that achieve comparable social welfare and Nash bargaining score negotiating with humans as humans trading among themselves.

PDF Details DOI

EAAI Journal 2025 Journal Article

Multi-scale motion-based relational reasoning for group activity recognition

Yihao Zheng
Zhuming Wang
Ke Gu
Lifang Wu
Zun Li
Ye Xiang

Existing group activity recognition methods generally use optical flow image to represent motion within videos, which often struggle to capture the movements of individuals inaccurately. In this paper, we explore the effectiveness of more kinds of motion information for group activity recognition. We propose a novel multi-scale MOtion-based relational reasoning framework for Group Activity Recognition (MOGAR). It combines joint motion (intra-individual level) with trajectory (individual-level) and individual position (inter-individual level) to acquire richer activity representation. Specifically, it involves two branches: the trajectory branch utilizes individuals’ trajectories and positions to extract the motion feature at the individual and inter-individual levels. The joint branch extracts the motion features at the intra-individual level. Furthermore, the gated recurrent units (GRU) and Transformers are employed to enhance the corresponding features through gating mechanism and self-attention mechanism. The features from the two branches are concatenated for group activity recognition. The experiments on two public datasets demonstrate that our method achieves competitive performance and has potential benefits in terms of computational complexity.

Details DOI

EAAI Journal 2025 Journal Article

Statistical Information Assisted Interaction Reasoning for skeleton-only group activity recognition

Zhuming Wang
Zun Li
Yihao Zheng
Lifang Wu

Group activity recognition is an essential task for video understanding and has shown significant value in practical application scenarios. Most existing methods mainly operate on visual information directly. Some skeleton-based methods replace the visual information with human skeletons to reduce computational pressure. However, due to the concise representation of skeletons, these methods typically leverage less information and show performance gaps with visual-based ones. In this paper, we introduce the statistical information of human joints and semantic information of action labels to provide more useful classification clues, and further propose a Statistical Information Assisted Interaction Reasoning Framework (SIAIR) for skeleton-only group activity recognition. It summarizes the joints statistical information to constructs the distribution maps, and explores semantic information of action labels to integrate both the skeleton-based individual representation and the statistical information. Extensive experiments on two public datasets validate the effectiveness of introducing statistical information, and the superiority of our framework over the state-of-the-art methods.

Details DOI

IJCAI Conference 2024 Conference Paper

A Meta-Game Evaluation Framework for Deep Multiagent Reinforcement Learning

Zun Li
Michael P. Wellman

Evaluating deep multiagent reinforcement learning (MARL) algorithms is complicated by stochasticity in training and sensitivity of agent performance to the behavior of other agents. We propose a meta-game evaluation framework for deep MARL, by framing each MARL algorithm as a meta-strategy, and repeatedly sampling normal-form empirical games over combinations of meta-strategies resulting from different random seeds. Each empirical game captures both self-play and cross-play factors across seeds. These empirical games provide the basis for constructing a sampling distribution, using bootstrapping, over a variety of game analysis statistics. We use this approach to evaluate state-of-the-art deep MARL algorithms on a class of negotiation games. From statistics on individual payoffs, social welfare, and empirical best-response graphs, we uncover strategic relationships among self-play, population-based, model-free, and model-based MARL methods. We also investigate the effect of run-time search as a meta-strategy operator, and find via meta-game analysis that the search version of a meta-strategy generally leads to improved performance.

PDF Details DOI

AAMAS Conference 2023 Conference Paper

Artificial Intelligence Algorithms for Strategic Reasoning over Complex Multiagent Systems

Zun Li

My Ph. D. research focuses on developing practical algorithms in computer games by assembling a variety of artificial intelligence methods (game-tree search, machine learning, graphical models, etc.). In this extended abstract, I will briefly review three of my previous works that studied normal-form games, Bayesian games, and extensive-form games through modern AI lenses. Then I will cast three possible future directions that I am dedicating to.

PDF

AAMAS Conference 2023 Conference Paper

Search-Improved Game-Theoretic Multiagent Reinforcement Learning in General and Negotiation Games

Zun Li
Marc Lanctot
Kevin R. McKee
Luke Marris
Ian Gemp
Daniel Hennes
Kate Larson
Yoram Bachrach

Multiagent reinforcement learning (MARL) has benefited significantly from population-based and game-theoretic training regimes. One approach, Policy-Space Response Oracles (PSRO), employs standard reinforcement learning to compute response policies via approximate best responses and combines them via meta-strategy selection. We augment PSRO by adding a novel search procedure with generative sampling of world states, and introduce two new meta-strategy solvers based on the Nash bargaining solution. We evaluate PSRO’s ability to compute approximate Nash equilibrium, and its performance in negotiation games: Colored Trails and Dealor-no-Deal. We conduct behavioral studies where human participants negotiate with our agents (𝑁 = 346). Search with generative modeling finds stronger policies during both training time and test time, enables online Bayesian co-player prediction, and can produce agents that achieve comparable social welfare negotiating with humans as humans trading among themselves.

PDF

AAAI Conference 2021 Conference Paper

Evolution Strategies for Approximate Solution of Bayesian Games

Zun Li
Michael P. Wellman

We address the problem of solving complex Bayesian games, characterized by high-dimensional type and action spaces, many (> 2) players, and general-sum payoffs. Our approach applies to symmetric one-shot Bayesian games, with no given analytic structure. We represent agent strategies in parametric form as neural networks, and apply natural evolution strategies (NES) (Wierstra et al. 2014) for deep model optimization. For pure equilibrium computation, we formulate the problem as bi-level optimization, and employ NES in an iterative algorithm to implement both inner-loop best response optimization and outer-loop regret minimization. In simple games including first- and second-price auctions, it is capable of recovering known analytic solutions. For mixed equilibrium computation, we adopt an incremental strategy generation framework, with NES as strategy generator producing a finite sequence of approximate best-response strategies. We then calculate equilibria over this finite strategy set via a model-based optimization process. Both our pure and mixed equilibrium computation methods employ NES to efficiently search for strategies over the function space, given only black-box simulation access to noisy payoff samples. We experimentally demonstrate the efficacy of all methods on two simultaneous sealed-bid auction games with distinct type distributions, and observe that the solutions exhibit qualitatively different behavior in these two environments.

PDF Details

AAAI Conference 2020 Conference Paper

Bounding Regret in Empirical Games

Steven Jecmen
Arunesh Sinha
Zun Li
Long Tran-Thanh

Empirical game-theoretic analysis refers to a set of models and techniques for solving large-scale games. However, there is a lack of a quantitative guarantee about the quality of output approximate Nash equilibria (NE). A natural quantitative guarantee for such an approximate NE is the regret in the game (i. e. the best deviation gain). We formulate this deviation gain computation as a multi-armed bandit problem, with a new optimization goal unlike those studied in prior work. We propose an efﬁcient algorithm Super-Arm UCB (SAUCB) for the problem and a number of variants. We present sample complexity results as well as extensive experiments that show the better performance of SAUCB compared to several baselines.

PDF Details

AAAI Conference 2020 Conference Paper

Structure Learning for Approximate Solution of Many-Player Games

Zun Li
Michael Wellman

Games with many players are difﬁcult to solve or even specify without adopting structural assumptions that enable representation in compact form. Such structure is generally not given and will not hold exactly for particular games of interest. We introduce an iterative structure-learning approach to search for approximate solutions of many-player games, assuming only black-box simulation access to noisy payoff samples. Our ﬁrst algorithm, K-Roles, exploits symmetry by learning a role assignment for players of the game through unsupervised learning (clustering) methods. Our second algorithm, G3L, seeks sparsity by greedy search over local interactions to learn a graphical game model. Both algorithms use supervised learning (regression) to ﬁt payoff values to the learned structures, in compact representations that facilitate equilibrium calculation. We experimentally demonstrate the efﬁcacy of both methods in reaching quality solutions and uncovering hidden structure, on both perfectly and approximately structured game instances.

PDF Details

TIST Journal 2019 Journal Article

Co-saliency Detection with Graph Matching

Zun Li
Congyan Lang
Jiashi Feng
Yidong Li
Tao Wang
Songhe Feng

Recently, co-saliency detection, which aims to automatically discover common and salient objects appeared in several relevant images, has attracted increased interest in the computer vision community. In this article, we present a novel graph-matching based model for co-saliency detection in image pairs. A solution of graph matching is proposed to integrate the visual appearance, saliency coherence, and spatial structural continuity for detecting co-saliency collaboratively. Since the saliency and the visual similarity have been seamlessly integrated, such a joint inference schema is able to produce more accurate and reliable results. More concretely, the proposed model first computes the intra-saliency for each image by aggregating multiple saliency cues. The common and salient regions across multiple images are thus discovered via a graph matching procedure. Then, a graph reconstruction scheme is proposed to refine the intra-saliency iteratively. Compared to existing co-saliency detection methods that only utilize visual appearance cues, our proposed model can effectively exploit both visual appearance and structure information to better guide co-saliency detection. Extensive experiments on several challenging image pair databases demonstrate that our model outperforms state-of-the-art baselines significantly.

Details DOI

AAMAS Conference 2018 Conference Paper

On Designing Optimal Data Purchasing Strategies for Online Ad Auctions

Zun Li
Zhenzhe Zheng
Fan Wu
Guihai Chen

In online advertising, advertisers can purchase consumer relevant data from data marketplaces with a certain expenditure, and exploit the purchased data to guide the bidding process in ad auctions. One of the pressing problem faced by advertisers is to design the optimal data purchasing strategy (how much data to purchase to be competitive in bidding process) in online ad auctions. In this paper, we model the data purchasing strategy design as a convex optimization problem, jointly considering the expenditure paid during data purchasing and the benefits obtained from ad auctions. Using the techniques from Baysian game theory and convex analysis, we derive the optimal purchasing strategies for advertisers in different market scenarios. We also theoretically prove that the resulting strategy profile is the unique one that achieves Nash Equilibrium. Our analysis shows that the proposed data purchasing strategy can handle diverse ad auctions and valuation learning models. Our numerical results empirically reveal how the equilibrium state changes with variation of the strategic environment.

PDF