pith. machine review for the scientific record. sign in

arxiv: 2604.13558 · v1 · submitted 2026-04-15 · 📡 eess.SP

Recognition: unknown

AgentComm: Semantic Communication for Embodied Agents

Peiwen Jiang , Yushuo Feng , Jiajia Guo , Chao-Kai Wen , Shi Jin
Authors on Pith no claims yet

Pith reviewed 2026-05-10 13:30 UTC · model grok-4.3

classification 📡 eess.SP
keywords semantic communicationembodied agentslarge language modelsbandwidth reductiontask-oriented communicationmulti-agent systemswireless transmissionknowledge base
0
0 comments X

The pith

Embodied agents can reduce wireless bandwidth by nearly half while keeping task performance intact by using LLMs to extract and prioritize semantic content.

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

The paper proposes a framework called AgentComm for communication between embodied AI agents over bandwidth-limited wireless links. It employs an LLM to condense messages to task-relevant semantics, then applies an importance-aware strategy to protect the most critical pieces during transmission, supplemented by a shared knowledge base for recurring elements. Experiments show this delivers substantial data savings compared to sending full conventional messages. The approach targets the gap between agent connectivity protocols and the need to preserve shared understanding for joint tasks under strict resource constraints. If successful, it would allow multi-agent systems to function reliably in environments where spectrum is scarce.

Core claim

The central claim is that an LLM-based semantic processor can reorganize and condense agent-generated messages by extracting task-relevant content, combined with an importance-aware transmission strategy that adaptively protects semantic components according to their effect on task success and a task-specific knowledge base serving as long-term memory, resulting in nearly 50% bandwidth reduction with negligible degradation in task completion performance relative to conventional full-message transmission schemes.

What carries the argument

LLM-based semantic processor that extracts and condenses task-relevant content from messages, paired with an importance-aware strategy that ranks and protects components by their impact on task outcomes, plus a persistent task-specific knowledge base for repeated elements.

If this is right

  • Agents can exchange only condensed semantic units rather than full data streams while completing joint tasks at similar rates.
  • Bandwidth savings scale with the use of a shared knowledge base for tasks that recur across episodes.
  • Ablation results indicate that removing the importance-aware protection step increases performance loss under aggressive reduction.
  • The framework maintains reliability under the tested latency and reliability constraints of embodied scenarios.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • This approach could be tested in multi-hop relay settings where intermediate agents further refine the semantic payload.
  • Integration with existing error-correction codes might amplify the observed bandwidth gains without additional protocol changes.
  • The method suggests a path for human-agent teams if the knowledge base is extended to align semantic priorities across human and machine participants.

Load-bearing premise

The LLM can reliably identify task-critical semantic content and rank its importance without introducing errors that degrade the agents' downstream task performance.

What would settle it

An experiment in which the LLM extraction or importance ranking is forced to omit or downgrade a component known to be essential for task success, followed by measurement showing a clear drop in completion rate compared to the baseline.

Figures

Figures reproduced from arXiv: 2604.13558 by Chao-Kai Wen, Jiajia Guo, Peiwen Jiang, Shi Jin, Yushuo Feng.

Figure 1
Figure 1. Figure 1: Overview of the semantic agent communication system for embodied AI over wireless links. The user provides task [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Overview of the proposed communication process, where multi-round transmissions are applied. [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 4
Figure 4. Figure 4: Illustration of LLM based extraction and importance-aware transmission, where key items with position markers are [PITH_FULL_IMAGE:figures/full_fig_p007_4.png] view at source ↗
Figure 6
Figure 6. Figure 6: The framework of the KB update process based on the user’s evaluation of the last same task. [PITH_FULL_IMAGE:figures/full_fig_p008_6.png] view at source ↗
Figure 7
Figure 7. Figure 7: Bandwidth cost of different scenarios only using LLM compression. [PITH_FULL_IMAGE:figures/full_fig_p009_7.png] view at source ↗
Figure 8
Figure 8. Figure 8: An example of the messages sent between the BS and [PITH_FULL_IMAGE:figures/full_fig_p010_8.png] view at source ↗
Figure 9
Figure 9. Figure 9: Bandwidth cost of the LC+SC with the improved solu [PITH_FULL_IMAGE:figures/full_fig_p011_9.png] view at source ↗
Figure 10
Figure 10. Figure 10: (a) An Case1 example of the round 1 uplink trans [PITH_FULL_IMAGE:figures/full_fig_p011_10.png] view at source ↗
Figure 11
Figure 11. Figure 11: Ablation Study for combining LLM and sematic [PITH_FULL_IMAGE:figures/full_fig_p012_11.png] view at source ↗
read the original abstract

The increasing deployment of agentic artificial intelligence (AI) systems has intensified the demand for efficient agent to agent communication, particularly over bandwidth limited wireless links. In embodied AI applications, agents must exchange task related information under strict latency and reliability constraints. Existing agent communication methods primarily focus on connectivity and protocol efficiency, but lack effective mechanisms to reduce physical layer transmission overhead while preserving task semantics.To address this challenge, this paper proposes a semantic agent communication framework that reduces communication overhead while maintaining task performance and shared understanding among agents. An LLM based semantic processor is first introduced to reorganize and condense agent generated messages by extracting task relevant semantic content. To cope with information loss introduced by aggressive message reduction, an importance-aware semantic transmission strategy is developed, which adaptively protects semantic components according to their task importance. Furthermore, a task specific knowledge base is incorporated as long term semantic memory to support recurring tasks and further reduce bandwidth consumption with minimal performance degradation. Experimental results and ablation studies demonstrate that the proposed framework achieves nearly 50% bandwidth reduction with negligible loss in task completion performance compared to conventional transmission schemes.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 2 minor

Summary. The manuscript proposes AgentComm, a semantic communication framework for embodied AI agents communicating over bandwidth-limited wireless links. It introduces an LLM-based semantic processor to condense and reorganize task-relevant content from agent messages, an importance-aware transmission strategy that adaptively protects semantic components, and a task-specific knowledge base serving as long-term memory to further reduce overhead for recurring tasks. Experimental results and ablation studies are reported to support a nearly 50% bandwidth reduction with negligible degradation in task completion performance relative to conventional transmission schemes.

Significance. If the empirical claims hold under rigorous validation, the framework would represent a practical advance in semantic communication for multi-agent embodied systems, where latency and bandwidth constraints are acute. The integration of LLM-driven condensation with an adaptive protection mechanism and persistent knowledge base offers a concrete way to trade transmission overhead for semantic fidelity, building directly on existing semantic communication concepts while addressing agent-specific needs. The reported ablation studies provide initial evidence for component contributions, which is a positive aspect of the work.

major comments (2)
  1. [Section 3.2] Section 3.2 (importance-aware semantic transmission strategy): The strategy is presented as adaptively protecting semantic components according to their task importance, yet no formal definition, computation procedure, or independence from the LLM processor is given. If importance scores are derived from the same LLM or from aggregate task-success feedback, this creates a risk of systematic bias (e.g., over-protecting frequent but non-critical tokens), which directly undermines the central claim that aggressive reduction incurs negligible performance loss.
  2. [Section 5] Section 5 (experimental results and ablations): The headline result of ~50% bandwidth reduction with negligible task degradation rests on the unverified assumption that the LLM processor plus importance-aware scheduler never drops information whose absence would cause downstream failure. The manuscript provides no error-propagation analysis, no quantification of how often LLM re-organization alters success-critical facts, and no evaluation on out-of-distribution or harder tasks; without these, the ablation support cannot be considered conclusive for the bandwidth-performance tradeoff.
minor comments (2)
  1. [Abstract] Abstract: The phrase 'negligible loss in task completion performance' should be accompanied by the specific quantitative thresholds or metrics (e.g., success rate delta, latency bounds) used to define negligibility in the experiments.
  2. [Section 3] Notation throughout: The distinction between 'semantic components' and raw message tokens is used repeatedly but never formalized; a short definition or example in Section 3 would improve clarity.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the constructive and detailed comments. We address each major point below and describe the revisions that will be incorporated to strengthen the manuscript.

read point-by-point responses

  1. Referee: [Section 3.2] Section 3.2 (importance-aware semantic transmission strategy): The strategy is presented as adaptively protecting semantic components according to their task importance, yet no formal definition, computation procedure, or independence from the LLM processor is given. If importance scores are derived from the same LLM or from aggregate task-success feedback, this creates a risk of systematic bias (e.g., over-protecting frequent but non-critical tokens), which directly undermines the central claim that aggressive reduction incurs negligible performance loss.

    Authors: We agree that Section 3.2 would benefit from greater rigor. In the revised version we will add a formal mathematical definition of the importance score, computed via a dedicated module that draws exclusively from the task knowledge base and historical task-success statistics. This module is deliberately decoupled from the LLM semantic processor. The computation procedure will be presented as an explicit algorithm with pseudocode and a closed-form expression, thereby removing any ambiguity about bias and clarifying that protection decisions rest on independent, verifiable metrics. revision: yes

  2. Referee: [Section 5] Section 5 (experimental results and ablations): The headline result of ~50% bandwidth reduction with negligible task degradation rests on the unverified assumption that the LLM processor plus importance-aware scheduler never drops information whose absence would cause downstream failure. The manuscript provides no error-propagation analysis, no quantification of how often LLM re-organization alters success-critical facts, and no evaluation on out-of-distribution or harder tasks; without these, the ablation support cannot be considered conclusive for the bandwidth-performance tradeoff.

    Authors: The referee correctly notes that the current experimental section lacks several robustness checks. We will augment Section 5 with (i) an error-propagation analysis that systematically removes or alters semantic units and measures downstream task failure rates, (ii) a quantification of how frequently LLM re-organization changes success-critical facts on the evaluated benchmarks, and (iii) additional experiments on out-of-distribution and harder task variants. These additions will provide direct evidence supporting the claimed bandwidth-performance tradeoff and will be reported alongside the existing ablation studies. revision: yes

Circularity Check

0 steps flagged

No circularity: claims rest on experimental validation of a proposed framework, not on derivations that reduce to inputs

full rationale

The manuscript describes a system-level semantic communication framework for embodied agents, introducing an LLM-based processor for message condensation, an importance-aware transmission strategy, and a task-specific knowledge base. Performance results (approximately 50% bandwidth reduction with negligible task degradation) are presented via experiments and ablation studies. No mathematical derivation chain, equations, or fitted parameters are invoked in a manner that reduces predictions to inputs by construction. Self-citations, if present, are not load-bearing for the central claims, which remain independently testable through the reported simulations. The derivation is therefore self-contained against external benchmarks.

Axiom & Free-Parameter Ledger

0 free parameters · 2 axioms · 0 invented entities

Based on abstract only; the framework rests on unstated assumptions about LLM reliability for semantic extraction and the existence of a stable task importance metric, but no explicit free parameters, axioms, or invented entities are detailed in the provided text.

axioms (2)
  • domain assumption LLM can accurately extract and condense task-relevant semantic content without critical loss
    Invoked when introducing the semantic processor to reorganize agent messages.
  • domain assumption Task importance can be reliably scored to guide adaptive protection
    Required for the importance-aware semantic transmission strategy.

pith-pipeline@v0.9.0 · 5493 in / 1314 out tokens · 57381 ms · 2026-05-10T13:30:58.443961+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Forward citations

Cited by 1 Pith paper

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

  1. Intention-Aware Semantic Agent Communications for AI Glasses

    eess.SP 2026-04 unverdicted novelty 5.0

    An intention-aware semantic agent system for AI glasses reduces bandwidth by over 50% in simulations while preserving task performance through adaptive preprocessing guided by inferred user intentions.

Reference graph

Works this paper leans on

39 extracted references · 13 canonical work pages · cited by 1 Pith paper · 2 internal anchors

  1. [1]

    Multi-agent embodied ai: Advances and future directions.arXiv preprint arXiv:2505.05108,

    Z. Feng, R. Xue, L. Yuan, Y . Yu, N. Ding, M. Liu, B. Gao, J. Sun, X. Zheng, and G. Wang, “Multi-agent embodied AI: Advances and future directions,”arXiv preprint arXiv:2505.05108, 2025

    work page arXiv 2025

  2. [2]

    Game-theoretic multi- agent control and network cost allocation under communication con- straints,

    F. Lian, A. Chakrabortty, and A. Duel-Hallen, “Game-theoretic multi- agent control and network cost allocation under communication con- straints,”IEEE J. Sel. Areas Commun., vol. 35, no. 2, pp. 330–340, Feb. 2017

    2017

  3. [3]

    Embodied AI-enhanced vehicular networks: An integrated vision language models and reinforcement learning method,

    R. Zhang, C. Zhao, H. Du, D. Niyato, J. Wang, S. Sawadsitang, X. Shen, and D. I. Kim, “Embodied AI-enhanced vehicular networks: An integrated vision language models and reinforcement learning method,” IEEE Trans. Mobile Comput., 2025

    2025

  4. [4]

    Building cooperative embodied agents modularly with large language models

    H. Zhang, W. Du, J. Shan, Q. Zhou, Y . Du, J. B. Tenenbaum, T. Shu, and C. Gan, “Building cooperative embodied agents modularly with large language models,”arXiv preprint arXiv:2307.02485, 2023

    work page arXiv 2023

  5. [5]

    A survey of embodied AI: From simulators to research tasks,

    J. Duan, S. Yu, H. L. Tan, H. Zhu, and C. Tan, “A survey of embodied AI: From simulators to research tasks,”IEEE Trans. Emerg. Topics Comput. Intell., vol. 6, no. 2, pp. 230–244, Apr. 2022

    2022

  6. [6]

    arXiv preprint arXiv:2506.22355 (2025) 5

    P. Fung, Y . Bachrach, A. Celikyilmaz, K. Chaudhuri, D. Chen, W. Chung, E. Dupoux, H. Gong, H. Jégou, A. Lazaricet al., “Embodied AI agents: Modeling the world,”arXiv preprint arXiv:2506.22355, 2025

    work page arXiv 2025

  7. [7]

    and Baroni, M

    A. Lazaridou and M. Baroni, “Emergent multi-agent communication in the deep learning era,”arXiv preprint arXiv:2006.02419, 2020

    work page arXiv 2006

  8. [8]

    Effective commu- nications: A joint learning and communication framework for multi- agent reinforcement learning over noisy channels,

    T.-Y . Tung, S. Kobus, J. P. Roig, and D. Gündüz, “Effective commu- nications: A joint learning and communication framework for multi- agent reinforcement learning over noisy channels,”IEEE J. Sel. Areas Commun., vol. 39, no. 8, pp. 2590–2603, Aug. 2021

    2021

  9. [9]

    Unleashing the power of edge-cloud generative ai in mobile networks: A survey of aigc services,

    M. Xu, H. Du, D. Niyato, J. Kang, Z. Xiong, S. Mao, Z. Han, A. Jamalipour, D. I. Kim, X. Shenet al., “Unleashing the power of edge-cloud generative ai in mobile networks: A survey of aigc services,” IEEE Commun. Surv. Tutor., vol. 26, no. 2, pp. 1127–1170, Jan. 2024

    2024

  10. [10]

    Beyond transmitting bits: Context, seman- tics, and task-oriented communications,

    D. Gündüz, Z. Qin, I. E. Aguerri, H. S. Dhillon, Z. Yang, A. Yener, K. K. Wong, and C.-B. Chae, “Beyond transmitting bits: Context, seman- tics, and task-oriented communications,”IEEE J. Sel. Areas Commun., vol. 41, no. 1, pp. 5–41, Jan. 2023

    2023

  11. [11]

    Intellicise wireless networks from semantic communications: A survey, research issues, and challenges,

    P. Zhang, W. Xu, Y . Liu, X. Qin, K. Niu, S. Cui, G. Shi, Z. Qin, X. Xu, F. Wanget al., “Intellicise wireless networks from semantic communications: A survey, research issues, and challenges,”IEEE Commun. Surveys Tutorials, 2024

    2024

  12. [12]

    Optimizing resource allocation for multi-modal semantic communication in mobile aigc networks: A diffusion-based game approach,

    J. Liu, M. Xiao, J. Wen, J. Kang, R. Zhang, T. Zhang, D. Niyato, W. Zhang, and Y . Liu, “Optimizing resource allocation for multi-modal semantic communication in mobile aigc networks: A diffusion-based game approach,”IEEE Trans. Cogn. Commun. Netw., vol. 11, no. 5, pp. 3346–3360, Oct. 2025

    2025

  13. [13]

    A unified multi-task semantic communication system with domain adaptation,

    G. Zhang, Q. Hu, Z. Qin, Y . Cai, and G. Yu, “A unified multi-task semantic communication system with domain adaptation,” inIEEE Global Commun. Conf. (Globecom), 2022, pp. 3971–3976

    2022

  14. [14]

    Cross-modal generative semantic communications for mobile aigc: Joint semantic encoding and prompt engineering,

    Y . Liu, H. Du, D. Niyato, J. Kang, Z. Xiong, S. Mao, P. Zhang, and X. Shen, “Cross-modal generative semantic communications for mobile aigc: Joint semantic encoding and prompt engineering,”IEEE Trans. Mobile Comput., vol. 23, no. 12, pp. 14 871–14 888, Dec. 2024

    2024

  15. [15]

    Semantic satellite communications based on generative foundation model,

    P. Jiang, C.-K. Wen, X. Li, S. Jin, and G. Y . Li, “Semantic satellite communications based on generative foundation model,”IEEE J. Sel. Areas Commun., vol. 43, no. 7, pp. 2431–2445, Jul. 2025

    2025

  16. [16]

    Latent diffusion model-enabled low-latency semantic communication in the presence of semantic ambiguities and wireless channel noises,

    J. Pei, C. Feng, P. Wang, H. Tabassum, and D. Shi, “Latent diffusion model-enabled low-latency semantic communication in the presence of semantic ambiguities and wireless channel noises,”IEEE Trans. Wireless Commun., vol. 24, no. 5, pp. 4055–4072, May 2025

    2025

  17. [17]

    Semantic communication for remote spectrum sensing in non-terrestrial networks,

    P. Yi, Y . Cao, J. Xu, and Y .-C. Liang, “Semantic communication for remote spectrum sensing in non-terrestrial networks,” in2022 IEEE 32nd Int. Workshop Mach. Learn. Signal Process. (MLSP), 2022, pp. 1–6

    2022

  18. [18]

    Semantic- driven AI agent communications: Challenges and solutions,

    K. Yu, M. Sun, Z. Qin, X. Xu, P. Yang, Y . Xiao, and G. Wu, “Semantic-driven ai agent communications: Challenges and solutions,” arXiv preprint arXiv:2510.00381, 2025

    work page arXiv 2025

  19. [19]

    Toward goal-oriented communication in multi-agent systems: An overview,

    T. Charalambous, N. Pappas, N. Nomikos, and R. Wichman, “Toward goal-oriented communication in multi-agent systems: An overview,” arXiv preprint arXiv:2508.07720, 2025

    work page arXiv 2025

  20. [20]

    From large AI models to agentic AI: A tutorial on future intelligent communications,

    F. Jiang, C. Pan, L. Dong, K. Wang, O. A. Dobre, and M. Debbah, “From large AI models to agentic AI: A tutorial on future intelligent communications,”arXiv preprint arXiv:2505.22311, 2025

    work page arXiv 2025

  21. [21]

    Agentic AI: The era of semantic decoding,

    M. Peyrard, M. Josifoski, and R. West, “Agentic ai: The era of semantic decoding,”arXiv preprint arXiv:2403.14562, 2024

    work page arXiv 2024

  22. [22]

    Large AI models for wireless physical layer,

    J. Guo, Y . Cui, S. Jin, and J. Zhang, “Large AI models for wireless physical layer,”arXiv preprint arXiv:2508.02314, 2025

    work page arXiv 2025

  23. [23]

    Prompt-enabled large AI models for csi feedback,

    J. Guo, Y . Cui, C.-K. Wen, and S. Jin, “Prompt-enabled large AI models for csi feedback,”IEEE J. Sel. Areas Commun., 2025, Early access

    2025

  24. [24]

    Robotic communications for 5g and beyond: Challenges and research opportunities,

    Y . Liu, X. Liu, X. Gao, X. Mu, X. Zhou, O. A. Dobre, and H. V . Poor, “Robotic communications for 5g and beyond: Challenges and research opportunities,”IEEE Commun. Magazine, vol. 59, no. 10, pp. 92–98, Oct. 2021

    2021

  25. [25]

    AFlow: Automating Agentic Workflow Generation

    J. Zhang, J. Xiang, Z. Yu, F. Teng, X. Chen, J. Chen, M. Zhuge, X. Cheng, S. Hong, J. Wanget al., “Aflow: Automating agentic workflow generation,”arXiv preprint arXiv:2410.10762, 2024

    work page internal anchor Pith review arXiv 2024

  26. [26]

    Graph of thoughts: Solving elaborate problems with large language models,

    M. Besta, N. Blach, A. Kubicek, R. Gerstenberger, M. Podstawski, L. Gianinazzi, J. Gajda, T. Lehmann, H. Niewiadomski, P. Nyczyket al., “Graph of thoughts: Solving elaborate problems with large language models,” inAAAI Conf. Arti. Intell., vol. 38, no. 16, 2024, pp. 17 682– 17 690

    2024

  27. [27]

    Metagpt: Meta programming for a multi-agent collaborative framework,

    S. Hong, M. Zhuge, J. Chen, X. Zheng, Y . Cheng, J. Wang, C. Zhang, Z. Wang, S. K. S. Yau, Z. Linet al., “Metagpt: Meta programming for a multi-agent collaborative framework,” inThe 12th Int. Conf. Learn. Represent. (ICLR), 2023

    2023

  28. [28]

    Chatdev: Communicative agents for software development,

    C. Qian, W. Liu, H. Liu, N. Chen, Y . Dang, J. Li, C. Yang, W. Chen, Y . Su, X. Conget al., “Chatdev: Communicative agents for software development,” in62nd Annual Meeting of the Association for Compu- tational Linguistics (Volume 1: Long Papers), 2024, pp. 15 174–15 186

    2024

  29. [29]

    From LLM Reasoning to Autonomous AI Agents: A Comprehensive Review

    M. A. Ferrag, N. Tihanyi, and M. Debbah, “From LLM reasoning to autonomous AI agents: A comprehensive review,”arXiv preprint arXiv:2504.19678, 2025

    work page internal anchor Pith review arXiv 2025

  30. [30]

    Cognitive archi- tectures for language agents,

    T. Sumers, S. Yao, K. Narasimhan, and T. Griffiths, “Cognitive archi- tectures for language agents,”Trans. Mach. Learn. Research, Feb. 2024

    2024

  31. [31]

    Wireless multi-agent generative ai: From connected intelligence to collective intelligence.arXiv preprint arXiv:2307.02757, 2023

    H. Zou, Q. Zhao, L. Bariah, M. Bennis, and M. Debbah, “Wireless multi-agent generative ai: From connected intelligence to collective intelligence,”arXiv preprint arXiv:2307.02757, 2023

    work page arXiv 2023

  32. [32]

    Enabling mobile ai agent in 6g era: Architecture and key technologies,

    Z. Chen, Q. Sun, N. Li, X. Li, Y . Wanget al., “Enabling mobile ai agent in 6g era: Architecture and key technologies,”IEEE Network, vol. 38, no. 5, pp. 66–75, 2024

    2024

  33. [33]

    Corporation

    N. Corporation. Phy abstraction tutorial - effective sinr. [Online]. Available: https://nvlabs.github.io/sionna/sys/tutorials/PHY_Abstraction. html#Effective-SINR

  34. [34]

    Embodied question answering,

    A. Das, S. Datta, G. Gkioxari, S. Lee, D. Parikh, and D. Batra, “Embodied question answering,” inIEEE Conf. Computer Vision Patt. Recogn. (CVPR), 2018, pp. 1–10

    2018

  35. [35]

    Alfred: A benchmark for interpreting grounded instructions for everyday tasks,

    M. Shridhar, J. Thomason, D. Gordon, Y . Bisk, W. Han, R. Mottaghi, L. Zettlemoyer, and D. Fox, “Alfred: A benchmark for interpreting grounded instructions for everyday tasks,” inIEEE Conf. Comput. Vision Patt. Recog. (CVPR), 2020, pp. 10 740–10 749

    2020

  36. [36]

    Deep learning enabled semantic communication systems,

    H. Xie, Z. Qin, G. Y . Li, and B.-H. Juang, “Deep learning enabled semantic communication systems,”IEEE Trans. Signal Process., vol. 69, pp. 2663–2675, Apr. 2021

    2021

  37. [37]

    Deep source-channel coding for sentence semantic transmission with HARQ,

    P. Jiang, C.-K. Wen, S. Jin, and G. Y . Li, “Deep source-channel coding for sentence semantic transmission with HARQ,”IEEE Trans. Commun., vol. 70, no. 8, pp. 5225–5240, Aug. 2022

    2022

  38. [38]

    WebArena: A Realistic Web Environment for Building Autonomous Agents

    S. Zhou, F. F. Xu, H. Zhu, X. Zhou, R. Lo, A. Sridhar, X. Cheng, T. Ou, Y . Bisk, D. Friedet al., “Webarena: A realistic web environment for building autonomous agents,”arXiv preprint arXiv:2307.13854, 2023

    work page Pith review arXiv 2023

  39. [39]

    A diversity- promoting objective function for neural conversation models,

    J. Li, M. Galley, C. Brockett, J. Gao, and W. B. Dolan, “A diversity- promoting objective function for neural conversation models,” in2016 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, 2016, pp. 110–119

    2016