arXiv subjects

Xinwei Song

Publications and source records attributed to Xinwei Song.

4 recordsLinked to original sources

arXiv2026

NormAct: Benchmarking Embodied Agents' Proactive Compliance with Unspoken Social Norms

Embodied agents driven by multimodal large language models (MLLMs) can often complete everyday tasks from visual observations, but goal achievement does not establish whether they proactively respect unstated social norms. Existing benchmarks assess explicit norm judgments or constrained behavior, but rarely test whether agents infer and apply scene-relevant norms during ordinary tasks. We introduce NormAct, a benchmark of 550 TongSim scenarios in which the same goal permits norm-compliant or norm-violating action sequences. Norm-relevant evidence is embedded in each scenario while the applicable rule is omitted from the goal instruction. By progressively increasing normative guidance while holding the goal and scene fixed, NormAct tests whether compliant behavior emerges autonomously or only after prompting. Across three MLLM planners, goal achievement substantially exceeds norm compliance without guidance (67.4% versus 24.7%), while both broad and rule-specific guidance improve compliance, indicating that planners can often comply when prompted but not reliably on their own. With a fixed planner, general-norm retrieval is less effective than norm-relevant scene descriptions or generated norm cues, suggesting that identifying relevant visual evidence is a greater challenge than accessing general norm knowledge. NormAct therefore supports the development of embodied agents that pursue everyday goals while proactively respecting unstated social norms.

cs.AI↗

arXiv2026

Learning to cooperate with emergent reputation via multi-agent reinforcement learning

Reputation, the aggregation of peer assessments diffused through social networks, is a pivotal mechanism for promoting cooperation in social dilemmas ubiquitous to distributed multi-agent systems comprising agents with limited perception and cognitive capabilities. Exploring efficient reputation systems, comprising reputation assessment rules and reputation-based policies, is a long-standing challenge. Previous work assumes predefined reputation assessment rules or models reputation as an intrinsic reward to learn policies, compromising the methods' ability for generalization and adaptation. To address this, we propose a distributed multi-agent reinforcement learning method $\textbf{COOPER}$ ($\textbf{COOP}$eration with $\textbf{E}$mergent $\textbf{R}$eputation), which jointly learns reputation assessment rules and reputation-based policies entirely from environment rewards. Notably, leveraging the underlying mechanisms of reputation, we deliberately design the constituent modules of $\textbf{COOPER}$ and the data flows among them, overcoming the latency and noise in the feedback signal, caused by the deep entanglement between reputation and policy. Experiments on the donation game and the coin game in grid world environments demonstrate that $\textbf{COOPER}$ effectively adapts to various existing reputation systems and co-players. Furthermore, we observe the co-emergence of reputation norms and cooperation in self-play settings. These results hold robustly across diverse social network topologies, underscoring the generalizability and efficacy of our approach.

cs.GT↗

arXiv2023

Connected Trading Cycles

Incentivizing the existing participants to invite new participants to join an auction, matching or cooperative game have been extensively studied recently. One common challenge to design such incentive in these games is that the invitees and inviters are competitors. To have such an incentive, we normally have to sacrifice some of the traditional properties. Especially, in a housing market (one kind of one-sided matching), we cannot maintain the traditional stability and optimality. The previous studies proposed some new matching mechanisms to have the invitation incentive (part of the incentive compatibility), but did not have any guarantee on stability and optimality. In this paper, we propose new notions of stability and optimality which are achievable with incentive compatibility. We weaken stability and optimality on a special structure (complete components) on networks. We first prove that the weakened notions are the best we can achieve with incentive compatibility. Then, we propose three mechanisms (Swap With Neighbors, Leave and Share, and Connected Trading Cycles) to satisfy the desirable properties. Connected Trading Cycles is the first mechanism to satisfy the best stability and optimality compatible with incentive compatibility.

cs.GT↗

arXiv2023

Truthful and Stable One-sided Matching on Networks

Mechanism design on social networks is a hot research direction recently, and we have seen many interesting results in auctions and matching. Compared to the traditional settings, the new goal of the network settings is that we need to design incentives to incentivize the participants of the game to invite their neighbors on the network to join the game. This is challenging because they are competing for something (e.g., resources or matches) in the game. In one-sided matching, especially house exchange, the well-known unique truthful, stable and optimal solution called Top Trading Cycle (TTC) cannot achieve the new goal. Existing works have tried to add constraints on TTC to obtain the incentive, but it only works in trees and it does not guarantee any stability. In this paper, we move this forward and propose the first mechanism called Leave and Share (LS) which not only achieves the goal in all networks but also gives the most stable solution in the new settings. In terms of optimality, as it is impossible to achieve it in any network, we conduct simulations to compare it with the extensions of TTC.

cs.GT↗