Pith. sign in

REVIEW 4 major objections 5 minor 35 references

TACTIC: Task-Agnostic Contrastive pre-Training for Inter-Agent Communication

T0 review · 4 major / 5 minor · reviewed 2026-08-10 · deepseek-v4-flash

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

Pith's one-line read TACTIC claims that a reward-free contrastive pretraining stage can make multi-agent communication generalize across sight ranges unseen during training, outperforming four baselines on SMACv2.

desk verdict Solid empirical paper with a real contribution, but the unspecified sight-range sampling distribution undercuts the 'generalization' headline until it's reported and ablated. read the letter →

arxiv 2501.02174 v1 pith:55ESJJFA submitted 2025-01-04 cs.MA

Peihong Yu , Manav Mishra , Syed Zaidi , Pratap Tokekar This is my paper · ORCID
classification cs.MA
keywords multi-agentreinforcementlearninginter-agentcommunicationcontrastivesightrangegeneralizationpartialobservabilitytask-agnosticpretrainingSMACv2QMIX
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

TACTIC is a two-stage method for cooperative multi-agent reinforcement learning that aims to make a team's communication robust to changes in how far each agent can see. The paper's central claim is that an offline, reward-free contrastive pretraining stage can teach agents to generate and interpret messages that capture global information, so that the final policy generalizes to sight ranges very different from the one it was trained on. This matters because existing communication methods are reported to degrade sharply when execution-time sight range differs from training-time sight range, forcing retraining for each new visibility condition. On the SMACv2 benchmark, TACTIC is reported to outperform four baselines in cross-sight-range generalization, with the largest margins at extremely small and extremely large sight ranges.

What carries the argument

The central object is the egocentric state embedding $\hat{z}_i$, computed from each agent's global state represented from that agent's perspective, which serves as the alignment target for communication. The mechanism that carries the argument is a pair of supervised contrastive losses: Global Information Alignment (GIA) pulls egocentric-state embeddings of all agents in the same episode and within a timestep window together, and Feature Integration Alignment (FIA) pulls the embedding $z_i$ produced by the message-observation integrator toward the corresponding $\hat{z}_i$. FIA is the part responsible for sight-range generalization, because the input observation $o_i^r$ is obtained by masking the egocentric state with a randomly sampled sight-range ratio $r$, forcing the message generator to learn to compensate for missing information at many observability levels. Reconstruction and dynamics losses, using a decoder and forward/inverse prediction networks, are auxiliary objectives that keep these embeddings from discarding the content needed to rebuild the egocentric state and to track temporal structure.

What would settle it

Train TACTIC with the offline masking procedure restricted to sight-range ratios near the training value, for example only $r=0.2$, and then evaluate at $r=5$; if the battle-win rate in this far-off-diagonal condition drops to the level of the QMIX-Att baseline instead of staying high, the claimed generalization beyond the pretraining distribution is refuted.

Watch

Extended reading notes

Core claim

On the paper's own terms, the discovery is that a task-agnostic contrastive objective can replace the usual need to retrain a communication policy for every observability setting. The offline stage trains a message generator and a message-observation integrator by aligning their output embeddings with embeddings of each agent's egocentric global state, using a supervised contrastive loss; a second contrastive loss aligns egocentric embeddings across agents. The generator is exposed to varied sight ranges by randomly masking each egocentric state with a sampled sight-range ratio before producing messages. Two auxiliary losses, reconstruction and forward/inverse dynamics, keep the representations informative and temporally coherent. After this stage, the communication modules are frozen and plugged into a QMIX-style value decomposition, and the resulting policies are evaluated on sight-range ratios from 0.2 to 5. The paper reports that TACTIC maintains usable battle-win rates across the whole tested range even when the training sight-range ratio differs from the test ratio by a factor of 25, while the baselines drop to near-zero win rates in those off-diagonal cases.

Load-bearing premise

The load-bearing premise is that the sight ranges sampled during offline pretraining cover the sight ranges used at evaluation, because the paper never specifies the sampling distribution over $r$, and without that coverage the reported generalization could be interpolation inside an already-seen observability range.

Editorial extensions

If this is right

  • A single trained model can be deployed across environments with different visibility conditions without retraining, provided the execution sight range falls in the range covered by pretraining.
  • Agents with very narrow sight ranges can maintain coordinated behavior through communication learned offline, reducing the penalty usually associated with limited observability.
  • Offline communication pretraining accelerates subsequent online policy learning, most clearly when the sight range is small and the number of agents is large.
  • Because the pretraining stage does not use reward signals, the same communication modules can in principle be reused for different downstream tasks within the same environment.
  • The task-agnostic pretraining stage means one communication module can replace per-scenario communication policies, making deployment cheaper for systems that must adapt to changing visibility.

Reading between the lines

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

  • A testable extension is to train the offline stage with a deliberately narrow distribution of sight-range ratios and then evaluate far outside it; if performance holds, the generalization is true extrapolation, and if not, the result is interpolation across a pretraining-covered range.
  • The offline stage's reliance on egocentric states derived from the global state means the method needs privileged information during data collection; an extension would be to align instead to a shared reconstructed state available from local observations alone.
  • The contrastive protocol may be interpreted as learning a compression of the joint state into messages; quantifying the mutual information between messages and the hidden global state could show how much of the communication channel is actually used.
  • The benchmark evidence is limited to a single environment family, so the task-agnostic claim would be tested more directly by applying the same offline pretraining to a different cooperative domain, such as navigation or robotic swarm tasks.
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

4 major / 5 minor

Summary. The paper proposes TACTIC, a two-stage method for cooperative MARL with communication. In an offline stage, a message generator and a message-observation integrator are trained using supervised contrastive losses (GIA and FIA) that align agent-specific integrated representations with egocentric-state embeddings, together with reconstruction and dynamics auxiliary losses. In an online stage, these modules are frozen and inserted into QMIX for policy learning. The method is evaluated on SMACv2 across three maps, team sizes of 5/10/20, training sight-range ratios of 0.2/1/5, and test sight-range ratios from 0.2 to 5, against QMIX, QMIX-Att, NDQ, and TarMAC. The central claim is that TACTIC generalizes better to sight ranges different from those seen during training, particularly under very small or very large sight ranges.

Significance. If the claims hold, TACTIC is a practical contribution: it uses task-agnostic offline data, including random trajectories, to pretrain communication modules that yield better sight-range generalization and faster online learning. The empirical scope is broad (three maps, three team sizes, three training SRRs, nine test SRRs, five seeds, four baselines), and the ablation shows that the two auxiliary losses are jointly important. The main caveats are that the random-sight-range augmentation is not precisely specified and is not ablated, and that no uncertainty quantification is provided, so the strength of the headline generalization claim is not yet established.

major comments (4)
  1. [Section 4.1 (FIA) and Section 5.1] The distribution over the randomly sampled sight range r used in P(s_hat_i, r) is never specified. Because the evaluation grid in Section 5.1 is exactly SRR in {0.2, 0.4, ..., 5.0}, if the pretraining r distribution covers this interval, the communication modules have already been trained on all test sight-range ratios, and the claimed 'generalization to sight ranges different from those seen in training' is only true of the online policy stage, not of the communication mechanism as a whole. Please state the sampling distribution, add an ablation that fixes r during FIA (e.g., r equal to the online training SRR) against the random-r version, and ideally evaluate at SRRs outside the pretraining range (e.g., 0.1 and 10) to test extrapolation.
  2. [Section 5.2, Figures 5-7] The heatmaps report only mean battle won rates over five seeds, with no standard deviations, standard errors, or significance tests. Since the paper's central claim is comparative (TACTIC outperforms four baselines), the absence of uncertainty quantification makes it impossible to know which numeric differences are meaningful; this is especially relevant for small differences such as those between Figure 9(a) and 9(b). Please report means with error bars or confidence intervals for the main heatmaps and learning curves.
  3. [Section 5.4, Figure 9] The ablations vary only dataset quality and the auxiliary losses; there is no ablation isolating the contrastive loss terms (GIA/FIA) or the random-r augmentation. Consequently, the results do not establish that the contrastive objective, rather than simple domain randomization over sight ranges during pretraining, is responsible for the advantage over baselines. Please add ablations with (i) FIA/GIA removed and (ii) FIA with a fixed r, and compare these to the full model.
  4. [Section 4.1 and Section 5.1] The offline training procedure omits several key hyperparameters: alpha and beta in Eq. (3), W_pos, W_neg, the SupCon temperature tau, and the message embedding dimension. Moreover, the size and composition of the offline dataset D are not fully specified; Section 5.4 mentions 6000 episodes 'per task', but the number of tasks, the collection policy for the main results, and the number of timesteps per episode are not given. Please include a complete hyperparameter and data-collection table.
minor comments (5)
  1. [Figure 3 caption] The caption says 'deconstruction loss' where the text and Eq. (3) say 'reconstruction loss'; please unify the terminology.
  2. [Section 5.3, Figure 8] The text says Figure 8 presents learning curves for 'every combination' of map, agent number, and sight-range ratio, but the figure as printed appears to omit several configurations (notably the 5v5 panels); please correct the text or complete the figure.
  3. [References] Reference [30] is cited for QMIX-Att, but the reference is titled 'Qatten: A General Framework for Cooperative Multiagent Reinforcement Learning'; please clarify the relationship or cite the correct source for QMIX-Att.
  4. [Figures 5-7] The color scales in the heatmaps are not described; please state whether all panels share a common color scale, since comparisons across panels currently require reading the numeric cell values.
  5. [Section 5.2] The phrase 'satisfying performances' would benefit from an explicit operational definition, such as a threshold battle win rate or a comparison rule relative to the training-condition performance.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the central claim is supported by external SMACv2 benchmark evaluation, and no load-bearing step reduces to its own input by construction.

full rationale

TACTIC's derivation chain is not circular. The communication modules are pretrained offline with a contrastive loss that aligns message-plus-observation representations with the egocentric state derived from the global state; this is a legitimate auxiliary training signal under the CTDE paradigm, not a restatement of the paper's empirical claims. The online policy is trained with frozen communication modules inside a QMIX architecture and evaluated on the external SMACv2 benchmark against QMIX, QMIX-Att, NDQ, and TarMAC. No parameter is fitted to the evaluation outcomes, and no result is defined in terms of the target claim. There is no load-bearing self-citation chain: the paper does not cite the authors' own prior work to justify its central premise. The only noteworthy limitation is that Section 4.1 says the offline stage obtains o_i^r by 'randomly sampling a sight range r' but never specifies the distribution; if that distribution covers the evaluation grid of sight-range ratios 0.2 to 5, then the communication module has seen those conditions during pretraining, and the 'different from those seen in training' language overstates the extrapolation at the component level. This is an experimental-design and interpretation concern, not a circularity: the benchmark results are externally measured and do not reduce to the training objective by construction.

Assumptions & free parameters 7 free parameters · 5 assumptions · 0 invented entities

The central claim rests on standard deep RL machinery (SupCon, QMIX, egocentric states) plus a design hypothesis: that contrastive alignment to egocentric states yields transferable communication. The main unstated assumptions are the distribution of sight-range augmentation and the availability of egocentric states in deployment. No new physical or conceptual entities are introduced; the message generator and integrator are learned neural modules.

free parameters (7)
  • alpha (reconstruction loss weight) = not reported
    Weighting factor for L_recon in Equation 3; only appears in ablation as 0 or 1, default value never stated.
  • beta (dynamics loss weight) = not reported
    Weighting factor for L_dyn in Equation 3; only appears in ablation as 0 or 1, default value never stated.
  • W_pos (positive pair timestep window) = not reported
    Defines positive pairs in GIA and FIA; value not given in text.
  • W_neg (negative pair timestep window) = not reported
    Defines negative pairs in GIA and FIA; value not given in text.
  • temperature tau in SupCon loss = not reported
    Temperature parameter in Equation 2; value not reported.
  • sight range sampling distribution for r = not reported
    Section 4.1 states r is randomly sampled but the distribution is not specified; this affects the generalization claim.
  • message embedding dimension = not reported
    Dimension of communication messages; not stated.
assumptions (5)
  • standard math Supervised contrastive loss (SupCon) is a valid representation learning objective.
    Equation 2 defines the loss; standard result from Khosla et al. [7].
  • standard math QMIX monotonic value function factorization supports centralized training and decentralized execution.
    Used as the online policy backbone in Section 4.2; standard result from Rashid et al. [15].
  • domain assumption Egocentric states are available for all agents during offline pretraining.
    Section 4.1 assumes the offline dataset D contains egocentric states derived from the global state; this is available in simulation but may not be in physical deployment.
  • domain assumption The random sight-range augmentation P(ŝ_i, r) in offline pretraining is representative of the sight ranges used in evaluation.
    Section 4.1 states r is randomly sampled but does not specify the distribution; the generalization result depends on this distribution covering or being informative for test SRRs.
  • ad hoc to paper Aligning integrated message-local representations with egocentric state embeddings improves communication and sight-range generalization.
    This is the core design hypothesis of TACTIC; it is plausible but not proven by theory, only by the empirical results.

how reviews work

0 comments
Cite this review

Pith. "Pith review of TACTIC: Task-Agnostic Contrastive pre-Training for Inter-Agent Communication." pith.science (2026). https://pith.science/paper/55ESJJFA

@misc{pith2026250102174,
  author       = {Pith},
  title        = {Pith review of: TACTIC: Task-Agnostic Contrastive pre-Training for Inter-Agent Communication},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/55ESJJFA}},
  note         = {Machine review of arXiv:2501.02174}
}
read the original abstract

The "sight range dilemma" in cooperative Multi-Agent Reinforcement Learning (MARL) presents a significant challenge: limited observability hinders team coordination, while extensive sight ranges lead to distracted attention and reduced performance. While communication can potentially address this issue, existing methods often struggle to generalize across different sight ranges, limiting their effectiveness. We propose TACTIC, Task-Agnostic Contrastive pre-Training strategy Inter-Agent Communication. TACTIC is an adaptive communication mechanism that enhances agent coordination even when the sight range during execution is vastly different from that during training. The communication mechanism encodes messages and integrates them with local observations, generating representations grounded in the global state using contrastive learning. By learning to generate and interpret messages that capture important information about the whole environment, TACTIC enables agents to effectively "see" more through communication, regardless of their sight ranges. We comprehensively evaluate TACTIC on the SMACv2 benchmark across various scenarios with broad sight ranges. The results demonstrate that TACTIC consistently outperforms traditional state-of-the-art MARL techniques with and without communication, in terms of generalizing to sight ranges different from those seen in training, particularly in cases of extremely limited or extensive observability.

Figures

Figures reproduced from arXiv: 2501.02174 by the authors.

Figure 1
Figure 1. TACTIC utilizes contrastive learning to align the integration of local observations 𝑜𝑖 and messages {𝑚𝑗𝑖 } with the full egocentric state 𝑠ˆ𝑖 for each agent 𝑖, enabling agents to "see" beyond their limited sight ranges through communi￾cation. such as robotics [1, 29, 32], autonomous vehicles [14, 18], and net￾work optimization [8, 34]. In MARL, agents often operate under partial observability, where each agent’s per… view at source ↗
Figure 2
Figure 2. QMIX-ATT and TACTIC’s performances on Protoss 10v10 from SMACv2 [4] with varying sight ranges (SRs). Dif￾ferent SRs are achieved by applying different sight-range ratios (SRRs) to the agents’ original SRs in the implementa￾tion. Policies trained at SRR=0.2, 1, and 5 are tested across a broader set of SRRs. QMIX-ATT Struggles to generalize to unseen SRs, while TACTIC generalizes much better. 2 RELATED WORK Communicat… view at source ↗
Figure 3
Figure 3. The offline training pipeline of the adaptive communication mechanism(Section 4.1). It includes three key components: [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗
Figures from the paper (6 more)
Figure 4
Figure 4. Figure 4: Online policy training pipeline of TACTIC, illustrat￾ing the integration of QMIX architecture with pre-trained communication components. The pre-trained message gen￾erator (Mess Gen) and message-observation integrator (Mess￾obs Integrator) remain fixed during the polic…
Figure 5
Figure 5. Figure 5: Performance of TACTIC and baseline models on policy generalizability across various sight ranges in the Protoss map. 0.2 0.4 0.6 0.8 1.0 2.0 3.0 4.0 5.0 Test SRR 0.2 1.0 5.0 Train SRR 0.50 0.50 0.50 0.48 0.48 0.44 0.43 0.42 0.41 0.36 0.43 0.55 0.70 0.70 0.60 0.57 0.55 …
Figure 6
Figure 6. Figure 6: Performance of TACTIC and baseline models on policy generalizability across various sight ranges in the Terran map [PITH_FULL_IMAGE:figures/full_fig_p006_6.png]
Figure 7
Figure 7. Figure 7: Performance of TACTIC and baseline models on policy generalizability across various sight ranges in the Zerg map. 5.2 Policy Generalization Across Sight Ranges To answer Q1, in this section, we report the generalization capabil￾ity of our trained policies across variou…
Figure 8
Figure 8. Figure 8: Training curves of the online policy learning stage in [PITH_FULL_IMAGE:figures/full_fig_p008_8.png]
Figure 9
Figure 9. Figure 9: Performance of TACTIC on policy generalizability across various sight ranges under 5 different offline training schemas. The environment setup is the Terran map with a 5v5 agent configuration. (a) Offline training with a Exploratory dataset; (b) Offline training with a…

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

35 extracted references · 19 canonical work pages

  1. [1]

    Lucian Busoniu, Robert Babuska, and Bart De Schutter. 2008. A comprehensive survey of multiagent reinforcement learning. IEEE Transactions on Systems, Man, and Cybernetics, Part C (Applications and Reviews) 38, 2 (2008), 156–172

    work page 2008

  2. [2]

    Abhishek Das, Théophile Gervet, Joshua Romoff, Dhruv Batra, Devi Parikh, Mike Rabbat, and Joelle Pineau. 2019. Tarmac: Targeted multi-agent communication. In International Conference on Machine Learning . PMLR, 1538–1546

    work page 2019

  3. [3]

    Yali Du, Bo Liu, Vincent Moens, Ziqi Liu, Zhicheng Ren, Jun Wang, Xu Chen, and Haifeng Zhang. 2021. Learning Correlated Communication Topology in Multi- Agent Reinforcement Learning. InProceedings of the 20th International Conference on Autonomous Agents and MultiAgent Systems . International Foundation for Autonomous Agents and Multiagent Systems, 456–464

    work page 2021

  4. [4]

    Benjamin Ellis, Jonathan Cook, Skander Moalla, Mikayel Samvelyan, Mingfei Sun, Anuj Mahajan, Jakob Foerster, and Shimon Whiteson. 2024. Smacv2: An improved benchmark for cooperative multi-agent reinforcement learning. Advances in Neural Information Processing Systems 36 (2024)

    2024

  5. [5]

    Cong Guan, Feng Chen, Lei Yuan, Chenghe Wang, Hao Yin, Zongzhang Zhang, and Yang Yu. 2022. Efficient Multi-agent Communication via Self-supervised Information Aggregation. Advances in Neural Information Processing Systems 35 (2022), 1020–1033

    2022

  6. [6]

    Shengchao Hu, Li Shen, Ya Zhang, and Dacheng Tao. 2024. Learning multi-agent communication from graph modeling perspective.arXiv preprint arXiv:2405.08550 (2024)

    arXiv 2024

  7. [7]

    Prannay Khosla, Piotr Teterwak, Chen Wang, Aaron Sarna, Yonglong Tian, Phillip Isola, Aaron Maschinot, Ce Liu, and Dilip Krishnan. 2020. Supervised Contrastive Learning. In Advances in Neural Information Processing Systems , Vol. 33. 18661– 18673

    work page 2020

  8. [8]

    Tianxu Li, Kun Zhu, Nguyen Cong Luong, Dusit Niyato, Qihui Wu, Yang Zhang, and Bing Chen. 2022. Applications of multi-agent reinforcement learning in future internet: A comprehensive survey. IEEE Communications Surveys & Tutorials 24, 2 (2022), 1240–1279

    work page 2022

Show all 35 references
  1. [9]

    Toru Lin, Jacob Huh, Christopher Stauffer, Ser Nam Lim, and Phillip Isola. 2021. Learning to ground multi-agent communication with autoencoders. Advances in Neural Information Processing Systems 34 (2021), 15230–15242

    2021

  2. [10]

    Yen-Cheng Liu, Junjiao Tian, Chih-Yao Ma, Nathan Glaser, Chia-Wen Kuo, and Zsolt Kira. 2020. Who2com: Collaborative perception via learnable handshake communication. In 2020 IEEE International Conference on Robotics and Automation (ICRA). IEEE, 6876–6883

    2020

  3. [11]

    Y. L. Lo, B. Sengupta, J. Foerster, and M. Noukhovitch. 2024. Learning Multi- Agent Communication with Contrastive Learning.arXiv preprint arXiv:2307.01403 (2024)

    2024 arXiv

  4. [12]

    Yaru Niu, Rohan R Paleja, and Matthew C Gombolay. 2021. Multi-Agent Graph- Attention Communication and Teaming. InProceedings of the AAMAS Conference, Vol. 21. International Foundation for Autonomous Agents and Multiagent Sys- tems

    2021

  5. [13]

    Frans A Oliehoek, Christopher Amato, et al . 2016. A concise introduction to decentralized POMDPs. Vol. 1. Springer

    2016

  6. [14]

    Kaige Qu, Weihua Zhuang, Qiang Ye, Wen Wu, and Xuemin Shen. 2024. Model- Assisted Learning for Adaptive Cooperative Perception of Connected Au- tonomous Vehicles. IEEE Transactions on Wireless Communications 23, 8 (2024), 8820–8835. https://doi.org/10.1109/TWC.2024.3354507

    2024

  7. [15]

    Tabish Rashid, Mikayel Samvelyan, Christian Schroeder de Witt, Gregory Far- quhar, Jakob Foerster, and Shimon Whiteson. 2018. QMIX: Monotonic Value Function Factorisation for Deep Multi-Agent Reinforcement Learning. arXiv e-prints, page. arXiv preprint arXiv:1803.11485 (2018)

    2018 arXiv

  8. [16]

    Mikayel Samvelyan, Tabish Rashid, Christian Schroeder De Witt, Gregory Far- quhar, Nantas Nardelli, Tim GJ Rudner, Chia-Man Hung, Philip HS Torr, Jakob Foerster, and Shimon Whiteson. 2019. The starcraft multi-agent challenge. arXiv preprint arXiv:1902.04043 (2019)

    2019 arXiv

  9. [17]

    Esmaeil Seraj, Zheyuan Wang, Rohan Paleja, Matthew Sklar, Anirudh Patel, and Matthew Gombolay. 2021. Heterogeneous graph attention networks for learning diverse communication. arXiv preprint arXiv:2108.09568 (2021)

    2021 arXiv

  10. [18]

    Shai Shalev-Shwartz, Shaked Shammah, and Amnon Shashua. 2016. Safe, multi-agent, reinforcement learning for autonomous driving. arXiv preprint arXiv:1610.03295 (2016)

    2016 arXiv

  11. [19]

    Jianzhun Shao, Hongchang Zhang, Yun Qu, Chang Liu, Shuncheng He, Yuhang Jiang, and Xiangyang Ji. 2023. Complementary Attention for Multi-Agent Rein- forcement Learning. (2023)

    2023

  12. [20]

    Jennifer She, Jayesh K Gupta, and Mykel J Kochenderfer. 2022. Agent-time attention for sparse rewards multi-agent reinforcement learning. arXiv preprint arXiv:2210.17540 (2022)

    2022 arXiv

  13. [21]

    Siqi Shen, Yongquan Fu, Huayou Su, Hengyue Pan, Peng Qiao, Yong Dou, and Cheng Wang. 2021. Graphcomm: A graph neural network based method for multi- agent reinforcement learning. In ICASSP 2021-2021 IEEE International Conference on Acoustics, Speech and Signal Processing (ICAS...

    2021

  14. [22]

    Singh, T

    A. Singh, T. Jain, and S. Sukhbaatar. 2019. Learning When to Communicate at Scale in Multiagent Cooperative and Competitive Tasks. In International Conference on Learning Representations

    2019

  15. [23]

    Sainbayar Sukhbaatar, Arthur Szlam, and Rob Fergus. 2016. Learning multiagent communication with backpropagation. In Proceedings of the 30th International Conference on Neural Information Processing Systems (Barcelona, Spain) (NIPS’16). Curran Associates Inc., Red Hook, NY, US...

    2016

  16. [24]

    Sainbayar Sukhbaatar, Arthur Szlam, and Rob Fergus. 2016. Learning multia- gent communication with backpropagation. In Advances in Neural Information Processing Systems. 2244–2252

    2016

  17. [25]

    Chuangchuang Sun, Macheng Shen, and Jonathan P How. 2020. Scaling up multiagent reinforcement learning for robotic systems: Learn an adaptive sparse communication graph. In 2020 IEEE/RSJ international conference on intelligent robots and systems (IROS) . IEEE, 11755–11762

    2020

  18. [26]

    van den Oord, Y

    A. van den Oord, Y. Li, and O. Vinyals. 2018. Representation Learning with Contrastive Predictive Coding. arXiv preprint arXiv:1807.03748 (2018)

    2018 arXiv

  19. [27]

    Tonghan Wang, Jianhao Wang, Chongyi Zheng, and Chongjie Zhang. 2019. Learning nearly decomposable value functions via communication minimization. arXiv preprint arXiv:1910.05366 (2019)

    2019 arXiv

  20. [28]

    Zhiwei Xu, Bin Zhang, Dapeng Li, Zeren Zhang, Guangchong Zhou, Hao Chen, and Guoliang Fan. 2023. Consensus Learning for Cooperative Multi-Agent Rein- forcement Learning. Proceedings of the AAAI Conference on Artificial Intelligence 37, 10 (Jun. 2023), 11726–11734. https://doi....

    2023 doi

  21. [29]

    Erfu Yang and Dongbing Gu. 2004. Multiagent reinforcement learning for multi- robot systems: A survey . Technical Report. tech. rep

    2004

  22. [30]

    Yaodong Yang, Jianye Hao, Ben Liao, Kun Shao, Guangyong Chen, Wulong Liu, and Hongyao Tang. 2020. Qatten: A General Framework for Cooperative Multiagent Reinforcement Learning. arXiv:2002.03939 [cs.MA] https://arxiv. org/abs/2002.03939

    2020 arXiv

  23. [31]

    Kejun Zhang, Jayesh K Gupta, Alfred O Hero III, and Mykel J Kochenderfer. 2021. Succinct and robust multi-agent communication with temporal message control. arXiv preprint arXiv:2107.06609 (2021)

    2021 arXiv

  24. [32]

    Kaiqing Zhang, Zhuoran Yang, and Tamer Başar. 2021. Multi-agent reinforce- ment learning: A selective overview of theories and algorithms. Handbook of Reinforcement Learning and Control (2021), 321–384

    2021

  25. [33]

    S. Q. Zhang, Q. Zhang, and J. Lin. 2019. Efficient Communication in Multi- Agent Reinforcement Learning via Variance Based Control. In Proceedings of the NeurIPS 32nd Conference on Neural Information Processing Systems , Vol. 32. Curran Associates, Inc

    2019

  26. [34]

    Zhi Zhang, Jiachen Yang, and Hongyuan Zha. 2019. Integrating independent and centralized multi-agent reinforcement learning for traffic signal network optimization. arXiv preprint arXiv:1909.10651 (2019)

    2019 arXiv

  27. [35]

    Ruijie Zheng, Xiyao Wang, Yanchao Sun, Shuang Ma, Jieyu Zhao, Huazhe Xu, Hal Daumé, and Furong Huang. 2024. TACO: temporal latent action-driven contrastive loss for visual reinforcement learning. In Proceedings of the 37th International Conference on Neural Information Process...

    2024

Pith tools

Reviewed August 10, 2026 · model on record in the stance chip above.