Pith. sign in

REVIEW 6 major objections 6 minor 2 cited by

Multi-Agent Reinforcement Learning for Dynamic Mobility Resource Allocation with Hierarchical Adaptive Grouping

T0 review · 6 major / 6 minor · reviewed 2026-08-06 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

Pith's one-line read Hierarchical adaptive grouping lifts bike rebalancing to a 77.21% service ratio.

desk verdict Plausible two-level adaptive grouping for MARL parameter sharing; the design is coherent and the ablations are clean, but the single-run experimental evidence and missing reproducibility details keep it from being more than a preliminary result. read the letter →

arxiv 2507.20377 v2 pith:4CGRPYVG submitted 2025-07-27 cs.AI

Farshid Nooshi , Suining He This is my paper · ORCID
classification cs.AI
keywords multi-agentreinforcementlearningdynamicparametersharingmobilityresourcerebalancinghierarchicalgroupingbikeadaptiveagentidentityembeddingsfulfilledserviceratio
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

The paper sets out to show that multi-agent reinforcement learning for urban mobility rebalancing can be made both memory-efficient and adaptive by sharing parameters hierarchically instead of keeping every agent independent, forcing every agent into one policy, or fixing clusters once. Using 1,232,838 New York City bike trips, it reports that its HAG-PS method reaches a 77.21% fulfilled service ratio, about eight percentage points above the strongest comparison method and more than 26 points above fully independent learning, while also rebalancing more bikes. The working hypothesis is that three design pieces - hierarchical global and local grouping, adaptive split-and-merge of agent groups from trajectory embeddings, and small learnable identity vectors - together let agents specialize without multiplying memory. If the result holds, the paper gives city-scale operators a concrete way to navigate the usual multi-agent trade-off between policy scale and behavioral diversity.

What carries the argument

The central object is a two-level parameter-sharing architecture. A global feature trunk, shared by all agents within a global group, maps the state into an embedding; beneath it, each local group maintains a compact actor-critic network head. After every episode, a variational LSTM encodes each agent's recent trajectory of states, actions, and rewards into an embedding, and the algorithm splits a local group when its symmetrized KL divergence around the group mean exceeds a threshold while the group is large enough, merging two groups when their mean embeddings are close enough. Split groups inherit their parent's network heads and then re-cluster into a fixed maximum number of subgroups, while learnable identity embeddings let agents keep individual specialization even under weight sharing. An adaptive regrouping interval, set by an exponentially smoothed divergence measure, makes cluster updates frequent when agents drift and rare once they stabilize. This mechanism is what carries the claim: it is the reason the policy can specialize at neighborhood scale without requiring a per-agent network.

What would settle it

Run HAG-PS and DyPS on realized trip-by-trip demand for a second city, or on the same New York data with many random seeds, and compare fulfilled service ratio; if DyPS reaches or exceeds HAG-PS's 77.21%, or if the reported margins do not reproduce, the central claim is falsified.

Watch

Extended reading notes

Core claim

The paper's central claim is that HAG-PS outperforms all compared baselines on its emulated benchmark: a fulfilled service ratio of 77.21% with 472,212 bikes rebalanced, versus 69.09% and 462,696 bikes for DyPS, 64.77% and 453,180 for SePS, 58.40% and 407,316 for CDS, 51.18% and 357,864 for No-Share, and 43.84% and 333,372 for Share-All. The ablations show each component pulling its weight: removing identity embeddings costs about 0.3 percentage points, disabling split-and-merge costs about 2.1 points, flattening the hierarchy costs about 4.0 points, and fixing the regrouping period costs about 1.1 points. The authors interpret this as evidence that the combination of hierarchy, adaptive grouping, and identity specialization drives the improvement, not any single trick.

Load-bearing premise

The whole result rests on the simulator described by the paper's update and reward equations being a trustworthy stand-in for real bike sharing: demand is represented by historical pickup means and standard deviations, relocation is decided before demand is realized, unserved requests are dropped, and inventory is floored at zero, so the 77.21% figure would not transfer if that emulation misrepresents real operations.

Editorial extensions

If this is right

  • The complete HAG-PS method reaches a 77.21% fulfilled service ratio and 472,212 bikes rebalanced, surpassing every baseline on both metrics within the same simulator.
  • The ablations imply that the hierarchy contributes about 4.0 percentage points, split-and-merge operations about 2.1 points, the adaptive regrouping period about 1.1 points, and identity embeddings about 0.3 points to the fulfilled service ratio.
  • Because the numbers of global and local groups are bounded, model memory stays bounded as the number of agents scales, making the approach a candidate for city-wide deployment.
  • The one-step inventory update with a floor at zero translates relocation actions into served versus dropped demand, so the reported gain is measured on the outcome operators care about: fulfilled pickup requests.
  • The adaptive grouping mechanism responds to behavioral drift, which means the policy can track seasonal or event-driven shifts in mobility demand without manual re-clustering.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • Editorial inference: the same adaptive grouping loop should be portable to other spatial coordination tasks, such as e-scooter repositioning or ride-hailing fleet matching, because it only needs region-level state, action, and reward histories.
  • Editorial inference: because the simulator summarizes demand by historical pickup means and standard deviations rather than realized trip sequences, a stronger test would run HAG-PS against DyPS on stochastic, trip-by-trip demand to see whether the reported 8-point gap persists.
  • Editorial inference: a natural next experiment is repeated training runs with multiple random seeds; reporting seed variance would confirm whether the 77.21% figure and the ranking over baselines are stable or a single-lottery outcome.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

6 major / 6 minor

Summary. The paper proposes HAG-PS, a multi-agent reinforcement learning method for dynamic bike rebalancing. Agents are regional coordinators; policies are shared through a two-level hierarchy (global feature trunk per group, compact local actor-critic heads), with adaptive splitting/merging of agent groups based on KL divergence of VLSTM trajectory embeddings, learnable identity embeddings for agent specialization, and an adaptive regroup period. The method is evaluated on an emulator built from 1.23 million NYC bike-sharing trips in January 2024 over 106 regions. Table 1 reports a fulfilled service ratio of 77.21% for HAG-PS versus 69.09% for the strongest baseline (DyPS), along with ablation results for each removed component. The paper also discusses related work and future extensions.

Significance. If the empirical claims are confirmed, HAG-PS would be a useful contribution to MARL for mobility resource allocation: it directly targets the scalability and adaptivity limitations of existing parameter-sharing methods, and the paper provides multiple ablations isolating the effect of each proposed component. The use of real trip data, a careful PPO implementation following the 37-detail checklist, and a comparison against several published baselines are strengths. However, the central quantitative claim is currently preliminary: it rests on a single run per method with no error bars, and the headline metric is essentially the same as the training reward's service-ratio term. Confirmation with statistical rigor would make the contribution solid.

major comments (6)
  1. [Section 4.2, Table 1] The central empirical claim is supported by a single training run per configuration; PPO is stochastic in initialization, trajectory sampling, and simulator demand draws, yet Table 1 reports no standard deviation, seed count, or significance test. The margins between HAG-PS and its ablations are as small as 0.30 percentage points (HAG-PS w/o ID) and 1.14 points (HAG-PS w/o ARP), which are plausibly within run-to-run noise. Please run at least 5 seeds per method, report mean and standard deviation, and provide a paired significance test for the HAG-PS versus DyPS comparison and for the ablation comparisons.
  2. [Section 3.1, Eqs. (3.1) and (3.2)] The environment update and the pre-service inventory equation use inconsistent indexing. In Eq. (3.1), the sum is over j with loc(j) ∈ N(k), but the term a^{(*,j)}_t is never defined, and the action a^i_t was defined as a four-direction vector for agent i. In Eq. (3.2), the sum is over loc(j) ∈ N(i), where i is an agent index, while the equation is meant to compute inventory for region k, and the text says a^{(*,j)}_t is 'contributed by agent i' inside a sum over j. Please define the mapping from agents to regions and write both equations with a consistent region index k and a signed net inflow term a^{(k,j)}_t.
  3. [Section 3.1, Eq. (3.5) and Section 4.1] The reward coefficients are inconsistent between the problem formulation and the experimental settings. Eq. (3.5) defines coefficients λ, α, and β, while Section 4.1 reports 'α = 5.0, β = 15.0, and γ = 3.0' and never gives a value for λ. Moreover, γ does not appear in Eq. (3.5), and the description 'discourage excessive relocations' corresponds to β in the equation. Since the reward is the training objective and the reported metric is derived from its first term, the value of λ must be specified, and the coefficient names must be aligned.
  4. [Section 4.1] The hyperparameters governing the adaptive grouping mechanism—D_split, τ_merge, S_min, and S_max—are used in the splitting and merging rules in Section 3.2, but no values are reported. These are central to the paper's novelty, and the claim of 'five stable hyper-parameters' is not substantiated because only the regroup-period parameters (Δ0, η, δ, ζ) are given. Please report all hyperparameter values and, ideally, a sensitivity analysis.
  5. [Section 3.2 and Section 4] The paper's second stated research challenge is memory-efficient parameter sharing, but no experiment measures parameter counts, memory footprint, or training time. Table 1 reports only fulfilled service ratio and total bikes rebalanced. The claim that HAG-PS bounds memory via capped group counts is qualitative; please include a comparison of model size and memory usage against Share-All, SePS, and DyPS to substantiate the memory-efficiency contribution.
  6. [Section 4.2, Eq. (4.1) vs Eq. (3.5)] The headline metric, fulfilled service ratio in Eq. (4.1), is the episode-level aggregate of the first reward term in Eq. (3.5), (1 − U/d). Thus Table 1 largely measures how well each method optimizes the authors' own reward, rather than an independent outcome such as realized demand served under a held-out demand process. This does not invalidate the baseline comparison, since all methods are trained with the same reward, but the abstract's claim of 'improved bike availability' would be better supported by reporting an independent metric or by framing the result as better optimization of the designed objective.
minor comments (6)
  1. [Section 2] Typo: 'invidual learnability' should be 'individual learnability'.
  2. [Section 3.1] Grammar and index consistency: 'each agent i serve a region' should be 'serves'; in the sentence introducing Eq. (3.5), 'for each agent i at time interval k' mixes agent and region indices, and the equation uses t rather than k.
  3. [Section 3.2, Eq. (3.11)] The definition of D̄_t is ambiguous: it is described as a running average 'within the local group' but is written as an average over groups of D^{(l)}_t. Please clarify the indexing of the running average.
  4. [Section 4.1] The statement 'Each training epoch consists of 64 episodes (corresponding to 64 simulated months)' is confusing because the dataset covers only January 2024; please specify how the 64 months are generated (e.g., resampling, bootstrapping, or repeated demand draws) and how demand is sampled within an episode.
  5. [Section 4.1] The input-state description says 'one-step demand history represented by a single dimension,' but earlier Section 3.1 states that state features include means and standard deviations of pick-ups over H intervals; please reconcile the dimension count and specify H and the total state vector size.
  6. [Table 1] Table 1 shows that Share-All (43.84%) underperforms No-Share (51.18%), which is an interesting result; the paper does not discuss why full sharing hurts in this setting, so please add a brief interpretation, as it affects the motivation for the proposed partial-sharing scheme.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the reported service ratio is an emulation result under a shared reward, not a derived prediction.

full rationale

The derivation chain is self-contained. HAG-PS is trained to maximize J(θ) = E[1/N Σ_i Σ_t γ^t r_i^t] (Eq. 3.6), with r_i^t defined in Eq. 3.5 as λ(1 − U_i^t/d_i^t) − αU_i^t/d_i^t − β‖a_i^t‖_1/m. The evaluation metric Avail_T in Eq. 4.1 is the global fulfilled-service ratio 1 − ΣU/Σd. Although this metric is aligned with the first reward term, that alignment is standard RL objective design, not a definitional reduction: the reported 77.21% is an empirical outcome of optimizing the reward in the simulator, and the baselines optimize the same reward under the same dynamics. No fitted parameter is renamed as a prediction; no load-bearing claim is justified solely by the authors' prior work; the cited prior methods (SePS, DyPS, CDS) are external baselines. The main validity concerns—single training run per method, no variance or seed reporting, and the λ/α/β/γ coefficient inconsistency in Section 4.1—are correctness and reproducibility issues, not circularity.

Assumptions & free parameters 4 free parameters · 4 assumptions · 3 invented entities

The central claim rests on a hand-tuned environment, a hand-designed reward, and grouping hyperparameters that are not all disclosed. No external benchmark outside the authors' simulator validates the method; all support is internal with public trip data used as input. The axiom count reflects assumptions that would need independent verification before deployment.

free parameters (4)
  • Reward weight lambda (service ratio term) = not reported; text lists alpha=5.0, beta=15.0, gamma=3.0 but no lambda value
    Eq 3.5 defines lambda, alpha, beta as hand-set weights; the value of lambda is missing from the paper.
  • Reward weights alpha and beta = alpha=5.0, beta=15.0
    Set by hand in Section 4.1 to balance service, under-service penalty, and relocation cost; affects all policies.
  • Grouping thresholds D_split, tau_merge, S_min, S_max = not specified
    Section 3.2 defines split and merge conditions but never reports the numeric thresholds or minimum and maximum group sizes.
  • Regrouping period constants = Delta0=8, delta=0.02, zeta=3, eta=0.90
    Section 4.1 tunes the adaptive regroup period; these constants control how often split-merge happens.
assumptions (4)
  • domain assumption Demand d_k^t is representable by historical pickup means and standard deviations plus one-step demand history
    Section 3.1 builds states from mu_k^t, sigma_k^t and one-step demand history; this discards trip-level temporal correlation beyond those aggregates.
  • domain assumption Relocations only move bikes between adjacent regions, and inventory is floored at zero (Eq 3.1)
    Eq 3.1 enforces a von Neumann neighborhood and a max(0, .) floor; real operators can move bikes across longer distances.
  • standard math PPO with the 37-detail checklist trains reliably without convergence guarantees
    Section 4.1 cites the PPO implementation checklist and applies PPO; no proof of convergence or stationarity is provided.
  • ad hoc to paper The reported fulfilled service ratio is the dominant objective being optimized
    Eq 4.1 equals the global version of the first reward term in Eq 3.5, so the evaluation metric overlaps the hand-designed training objective.
invented entities (3)
  • VLSTM trajectory encoder
    purpose: Compresses H-step (state, action, reward) trajectories into embeddings z_i^t used for split-merge decisions
    Architectural component with no external falsifiable prediction; validated only by ablation.
  • Learnable identity embeddings e_i
    purpose: Allows agents within a shared group to specialize beyond parameter copy
    Ablation shows a small 0.3% service-ratio drop when removed; no external prediction.
  • Two-level hierarchical group structure (global trunk, local actor-critic heads)
    purpose: Enables dynamic parameter sharing at district and neighborhood scales
    Algorithmic construct; ablation shows a 4% service-ratio drop without it, but it has no external handle.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Multi-Agent Reinforcement Learning for Dynamic Mobility Resource Allocation with Hierarchical Adaptive Grouping." pith.science (2026). https://pith.science/paper/4CGRPYVG

@misc{pith2026250720377,
  author       = {Pith},
  title        = {Pith review of: Multi-Agent Reinforcement Learning for Dynamic Mobility Resource Allocation with Hierarchical Adaptive Grouping},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/4CGRPYVG}},
  note         = {Machine review of arXiv:2507.20377}
}
read the original abstract

Allocating mobility resources (e.g., shared bikes/e-scooters, ride-sharing vehicles) is crucial for rebalancing the mobility demand and supply in the urban environments. We propose in this work a novel multi-agent reinforcement learning named Hierarchical Adaptive Grouping-based Parameter Sharing (HAG-PS) for dynamic mobility resource allocation. HAG-PS aims to address two important research challenges regarding multi-agent reinforcement learning for mobility resource allocation: (1) how to dynamically and adaptively share the mobility resource allocation policy (i.e., how to distribute mobility resources) across agents (i.e., representing the regional coordinators of mobility resources); and (2) how to achieve memory-efficient parameter sharing in an urban-scale setting. To address the above challenges, we have provided following novel designs within HAG-PS. To enable dynamic and adaptive parameter sharing, we have designed a hierarchical approach that consists of global and local information of the mobility resource states (e.g., distribution of mobility resources). We have developed an adaptive agent grouping approach in order to split or merge the groups of agents based on their relative closeness of encoded trajectories (i.e., states, actions, and rewards). We have designed a learnable identity (ID) embeddings to enable agent specialization beyond simple parameter copy. We have performed extensive experimental studies based on real-world NYC bike sharing data (a total of more than 1.2 million trips), and demonstrated the superior performance (e.g., improved bike availability) of HAG-PS compared with other baseline approaches.

Figures

Figures reproduced from arXiv: 2507.20377 by the authors.

Figure 1
Figure 1. Illustration of overall architecture of HAG-PS. • Problem Formulation. We formulate our mobility resource allocation problem as a finite–horizon multi-agent Markov game. For a horizon of 𝑇 time intervals, and a discount factor 𝛾 ∈ (0, 1], HAG-PS aims to find the MARL parameters 𝜽 that maximize the objective function of 𝐽 (𝜽) = E h 1 𝑁 ∑︁ 𝑁 𝑖=1 𝑇 ∑︁−1 𝑡=0 𝛾 𝑡 𝑟 𝑖 𝑡 i . (3.6) 3.2 Hierarchical Adaptive Grouping We have… view at source ↗

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 2 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. Mobility-Aware Cache Framework for Scalable LLM-Based Human Mobility Simulation

    cs.AI 2026-02 conditional novelty 6.0 of 10

    A latent-space reasoning cache with a lightweight decoder cuts the cost of LLM-based human mobility simulation by roughly 40-90% while keeping trajectory quality comparable.

  2. Bridging Individual and Collective Realism in LLM-Based Human Mobility Simulation via Mobility Scaling-Law Guidance

    cs.MA 2026-02 conditional novelty 6.0 of 10

    M2LSimu uses population-level mobility statistics as a reward signal to iteratively adjust LLM prompts, improving simulated trajectories' match to real mobility patterns.

Reference graph

Works this paper leans on

24 extracted references · 21 canonical work pages · cited by 2 Pith papers

  1. [1]

    Xiaohui Bei and Shengyu Zhang. 2018. Algorithms for trip-vehicle assignment in ride-sharing. In Proc. AAAI, Vol. 32

    work page 2018

  2. [2]

    Rahman, and Ste- fano V

    Filippos Christianos, Georgios Papoudakis, Muhammad A. Rahman, and Ste- fano V. Albrecht. 2021. Scaling Multi-Agent Reinforcement Learning with Se- lective Parameter Sharing. In Proceedings of the 38th International Conference on 4 Multi-Agent Reinforcement Learning for Dynamic Mobility Resource Allocation with Hierarchical Adaptive Grouping Machine Learn...

    work page 2021

  3. [3]

    James C Chu, Hsin-Chia Lin, Fan-Yu Liao, and Yu-Hsuan Yu. 2023. Dynamic repo- sitioning problem of dockless electric scooter sharing systems. Transportation Letters 15, 9 (2023), 1066–1082

    work page 2023

  4. [4]

    Tianshu Chu, Jie Wang, Lara Codecà, and Zhaojian Li. 2020. Multi-Agent Deep Reinforcement Learning for Large-Scale Traffic Signal Control. IEEE Transactions on Intelligent Transportation Systems 21, 3 (2020), 1086–1095

    work page 2020

  5. [5]

    Xiangxiang Chu and Hangjun Ye. 2017. Parameter Sharing Deep Deterministic Policy Gradient for Cooperative MARL. arXiv:1710.00336

    arXiv 2017

  6. [6]

    Foerster, Yannis M

    Jakob N. Foerster, Yannis M. Assael, Nando de Freitas, and Shimon Whiteson

    work page

  7. [7]

    Gupta, Maxim Egorov, and Mykel J

    Jayesh K. Gupta, Maxim Egorov, and Mykel J. Kochenderfer. 2017. Cooperative Multi-Agent Control Using Deep Reinforcement Learning. In Adaptive Learning Agents Workshop @ AAMAS

    work page 2017

  8. [8]

    Suining He and Kang G. Shin. 2019. Spatio-Temporal Adaptive Pricing for Balancing Mobility-on-Demand Networks. ACM Transactions on Intelligent Systems and Technology (TIST) 10, 4 (July 2019), 39:1–39:28

    work page 2019

Show all 24 references
  1. [9]

    Suining He and Kang G. Shin. 2022. Spatio-Temporal Capsule-Based Reinforce- ment Learning for Mobility-on-Demand Coordination. IEEE Transactions on Knowledge and Data Engineering 34, 3 (Mar 2022), 1446–1461

    2022

  2. [10]

    Runqiu Hu, Zhizheng Zhang, Xinwei Ma, and Yuchuan Jin. 2021. Dynamic rebalancing optimization for bike-sharing system using priority-based MOEA/D algorithm. IEEE Access 9 (2021), 27067–27084

    2021

  3. [11]

    Shengyi Huang, Rousslan Fernand Julien Dossa, Antonin Raffin, Anssi Kanervisto, and Weixun Wang. 2022. The 37 Implementation Details of Proximal Policy Optimization. In ICLR Blog Track

    2022

  4. [12]

    Chenghao Li, Tonghan Wang, Chengjie Wu, Qianchuan Zhao, Jun Yang, and Chongjie Zhang. 2021. Celebrating Diversity in Shared Multi-Agent Reinforce- ment Learning. In Proc. NeurIPS. 3991–4002

    2021

  5. [13]

    Hui Li, Xin Sun, Xueqian Wang, and Shaojie Shen. 2020. i-Rebalance: Person- alized Vehicle Repositioning for Supply-Demand Balance. In Proceedings of the Thirty-Fourth AAAI Conference on Artificial Intelligence (AAAI) . 480–487

    2020

  6. [14]

    Xinghua Li, Xinyuan Zhang, Cheng Cheng, Wei Wang, and Chao Yang. 2022. Dynamic Repositioning in Dock-Less Bike-Sharing System: A Multi-Agent Rein- forcement Learning Approach. In IEEE Intelligent Transportation Systems Confer- ence (ITSC). 3352–3357

    2022

  7. [15]

    Yilin Liu, Guiyang Luo, Quan Yuan, Jinglin Li, Lei Jin, Bo Chen, and Rui Pan

  8. [16]

    Darvariu, Maria A

    Alessandro Staffolani, Victor-A. Darvariu, Maria A. Cheema, and Mirco Mu- solesi. 2025. A Cost -Aware Adaptive Bike Repositioning Agent Using Deep Reinforcement Learning. IEEE Transactions on Intelligent Transportation Systems (2025)

    2025

  9. [17]

    Terry, Nathaniel Grammel, Sanghyun Son, Benjamin Black, and Aakriti Agrawal

    Justin K. Terry, Nathaniel Grammel, Sanghyun Son, Benjamin Black, and Aakriti Agrawal. 2020. Revisiting Parameter Sharing in Multi-Agent Deep Reinforcement Learning. arXiv:2005.13625

    2020 arXiv

  10. [18]

    Jingwei Wang, Qianyue Hao, Wenzhen Huang, Xiaochen Fan, Zhentao Tang, Bin Wang, Jianye Hao, and Yong Li. 2024. DyPS: Dynamic Parameter Sharing in Multi-Agent Reinforcement Learning for Spatio-Temporal Resource Allocation. In Proc. ACM KDD. 3128–3139

    2024

  11. [19]

    Tonghan Wang, Jingyang Zhou, Qianchuan Zhao, and et al. 2021. RODE: Learn- ing Roles to Decompose Multi-Agent Tasks. In 9th International Conference on Learning Representations (ICLR)

    2021

  12. [20]

    Xi Yang, Suining He, and Mahan Tabatabaie. 2023. Equity-Aware Cross-Graph Reinforcement Learning for Bike Station Network Expansion. In Proceedings of the 2023 ACM SIGSPATIAL International Conference on Advances in Geographic Information Systems (SIGSPATIAL). 45:1–45:12

    2023

  13. [21]

    Yaodong Yang, Rui Luo, Minne Li, Ming Zhou, Weinan Zhang, and Jun Wang

  14. [2016]

    In Advances in Neural Information Processing Systems 29 (NeurIPS)

    Learning to Communicate with Deep Multi-Agent Reinforcement Learning. In Advances in Neural Information Processing Systems 29 (NeurIPS) . 2137–2145

  15. [2018]

    In Proceedings of the 35th International Conference on Machine Learning (ICML)

    Mean Field Multi -Agent Reinforcement Learning. In Proceedings of the 35th International Conference on Machine Learning (ICML) . 5571–5580. 5

  16. [2023]

    In Proceedings of the 32nd International Joint Conference on Artificial Intelligence (IJCAI)

    GPLight: Grouped Multi-Agent Reinforcement Learning for Large-Scale Traffic Signal Control. In Proceedings of the 32nd International Joint Conference on Artificial Intelligence (IJCAI). 172–180

Pith tools

Reviewed August 6, 2026 · model on record in the stance chip above.