pith. sign in

arxiv: 2605.25916 · v1 · pith:XMLQPK4Gnew · submitted 2026-05-25 · 💻 cs.LG · cs.DC· cs.NI

Joint Optimization of Training and Inference in Federated Edge Learning via Constrained Multi-Objective Deep Reinforcement Learning

Zhen Li , Jun Cai , Chao Yang , Haoran Gao This is my paper

Pith reviewed 2026-06-29 22:31 UTC · model grok-4.3

classification 💻 cs.LG cs.DCcs.NI
keywords federated edge learningmulti-objective optimizationdeep reinforcement learningproximal policy optimizationinference accuracylatencyenergy consumptionPareto front
0
0 comments X

The pith

Constrained multi-objective proximal policy optimization jointly optimizes training and inference in federated edge learning.

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

The paper introduces an online framework that manages both federated model training and inference requests on resource-constrained edge devices. It uses a tandem-queue conversion to link inference requests to training data and includes data and model freshness in the accuracy metric to reflect time dynamics. The joint decisions on mode selection, communication, and computation resources are formulated as a multi-objective Markov decision process. The C-MOPPO algorithm solves this by first training policies across different objective preferences and then applying constrained policy optimization to produce dense, high-quality Pareto fronts. Experiments indicate that the method yields well-balanced results across accuracy, latency, and energy that exceed those of baseline approaches in multiple system settings.

Core claim

The C-MOPPO algorithm first learns a set of policies with different preferences across inference accuracy, latency, and energy objectives, then leverages constrained policy optimization to enrich the Pareto front and obtain high-quality, dense solutions for the multi-objective Markov decision process that incorporates tandem-queue conversion and freshness terms.

What carries the argument

The constrained multi-objective proximal policy optimization (C-MOPPO) algorithm, which converts the NP-hard online multi-objective problem into a MOMDP and jointly optimizes device mode selections along with communication and computation resource allocations.

If this is right

  • Maximizes inference accuracy while minimizing latency and energy consumption through joint optimization of training and inference.
  • Achieves well-balanced trade-offs among accuracy, latency, and energy objectives.
  • Significantly outperforms baseline methods under various system configurations.
  • Supports online management of mode selections, communication, and computation resources on edge devices.

Where Pith is reading between the lines

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

  • The MOMDP formulation with freshness metrics could transfer to time-sensitive distributed inference tasks outside federated learning.
  • Hardware-in-the-loop testing would reveal whether the queue conversion holds when real network delays replace the modeled tandem queues.
  • Similar constrained multi-objective reinforcement learning could address conflicting goals in other resource allocation problems such as task offloading.
  • Adding explicit network topology variables to the state space might further tighten the accuracy model.

Load-bearing premise

The tandem-queue-inspired conversion mechanism that bridges inference requests and training data, together with the inclusion of data and model freshness in the accuracy formulation, accurately models temporal dynamics in real-world federated edge environments.

What would settle it

Physical deployment measurements in which observed accuracy-latency-energy triples deviate substantially from the Pareto front produced by C-MOPPO under the tandem-queue model.

Figures

Figures reproduced from arXiv: 2605.25916 by Chao Yang, Haoran Gao, Jun Cai, Zhen Li.

Figure 1
Figure 1. Figure 1: Illustration of the FEEL system. mode process requests using the stored model and return the results to users for evaluation. The completed requests, along with user feedback, are then converted into labeled training samples that are temporarily stored and utilized in the subsequent training [26]. After training, these samples are discarded, thereby ensuring the storage constraints in practical deployments… view at source ↗
Figure 2
Figure 2. Figure 2: Overview of C-MOPPO. B. C-MOPPO Algorithm As illustrated in [PITH_FULL_IMAGE:figures/full_fig_p007_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Pareto fronts obtained by C-MOPPO, PG-MORL, MOEA/D, S-PPO, and R+PPO algorithms. [PITH_FULL_IMAGE:figures/full_fig_p012_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Performance comparison with different numbers of [PITH_FULL_IMAGE:figures/full_fig_p013_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Performance comparison with different average re [PITH_FULL_IMAGE:figures/full_fig_p013_5.png] view at source ↗
Figure 8
Figure 8. Figure 8: Comparison of la￾tency distributions across different algorithms. flatter curves with severe long-tail effects, indicating the random mode selection frequently forces inference requests to wait in training mode, causing severe latency degradation. By strategically switching modes, C-MOPPO avoids such delays and provides deterministic latency bounds, ensuring the strict QoS constraints. E. Evaluation on Sta… view at source ↗
read the original abstract

Federated edge learning (FEEL) has recently emerged as a promising paradigm for achieving edge intelligence (EI) via enabling collaborative model training across edge devices while protecting data privacy. In this paper, we put forth an online optimization framework that jointly manages federated training and inference on resource-constrained edge devices. We introduce a tandem-queue-inspired conversion mechanism that bridges inference requests and training data, and further incorporate both data and model freshness into the accuracy formulation to capture temporal dynamics in real-world environments. To maximize inference accuracy while minimizing latency and energy consumption, the mode selections, communication, and computation resource allocations of edge devices are jointly optimized. We formulate this optimization as a multi-objective optimization problem, which is NP-hard and further complicated by the online setting. To address these challenges, we transform the problem into a multi-objective Markov decision process (MOMDP) and develop a \underline{c}onstrained \underline{m}ulti-\underline{o}bjective \underline{p}roximal \underline{p}olicy \underline{o}ptimization (C-MOPPO) algorithm. Specifically, C-MOPPO first learns a set of policies with different preferences across three objectives, then leverages constrained policy optimization to enrich the Pareto front and obtain high-quality, dense solutions. Extensive experiments demonstrate that C-MOPPO achieves well-balanced trade-offs among objectives and significantly outperforms baselines under various system configurations.

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 paper proposes an online joint optimization framework for federated training and inference on resource-constrained edge devices. It introduces a tandem-queue-inspired conversion mechanism to link inference requests with training data, augments the accuracy objective with data and model freshness terms to capture temporal dynamics, formulates the resulting multi-objective problem as an MOMDP, and develops the C-MOPPO algorithm that first learns preference-parameterized policies and then applies constrained policy optimization to enrich the Pareto front. The central claim is that C-MOPPO produces well-balanced trade-offs and significantly outperforms baselines across system configurations.

Significance. If the modeling assumptions prove faithful to real edge request and staleness statistics, the work would supply a concrete, online multi-objective RL method for balancing accuracy, latency, and energy in federated edge learning, extending constrained policy optimization to dense Pareto fronts in this domain.

major comments (2)
  1. [Problem formulation and accuracy model] The tandem-queue conversion mechanism and the freshness-augmented accuracy formulation (introduced in the problem statement and used to define the MOMDP state/reward) are load-bearing for the performance claim; without empirical grounding against real federated edge traces or request-arrival statistics, the learned policies risk being artifacts of the synthetic dynamics rather than evidence of practical superiority.
  2. [Abstract and experimental evaluation] The abstract asserts that 'extensive experiments demonstrate' outperformance and well-balanced trade-offs, yet the provided description contains no quantitative metrics, baseline specifications, ablation results, or error bars; this absence prevents verification of the headline claim that C-MOPPO enriches the Pareto front under varied configurations.
minor comments (2)
  1. Define all acronyms (MOMDP, C-MOPPO, FEEL, EI) on first use and ensure consistent notation between the MOMDP tuple and the subsequent C-MOPPO description.
  2. [C-MOPPO algorithm] Clarify whether the constrained policy optimization step in C-MOPPO enforces hard constraints on latency/energy or uses Lagrangian relaxation, and state the exact form of the constraint violation penalty.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the constructive feedback. We address each major comment below and indicate where revisions will be incorporated.

read point-by-point responses

  1. Referee: [Problem formulation and accuracy model] The tandem-queue conversion mechanism and the freshness-augmented accuracy formulation (introduced in the problem statement and used to define the MOMDP state/reward) are load-bearing for the performance claim; without empirical grounding against real federated edge traces or request-arrival statistics, the learned policies risk being artifacts of the synthetic dynamics rather than evidence of practical superiority.

    Authors: We agree that real-world traces would provide stronger validation. The current evaluation uses synthetic request arrivals and staleness dynamics parameterized from established edge-computing literature, with extensive sweeps across system configurations to test robustness. In revision we will add an explicit subsection on modeling assumptions, their literature basis, and a dedicated limitations paragraph outlining the need for future real-trace validation. This is a partial revision that clarifies rather than alters the technical claims. revision: partial

  2. Referee: [Abstract and experimental evaluation] The abstract asserts that 'extensive experiments demonstrate' outperformance and well-balanced trade-offs, yet the provided description contains no quantitative metrics, baseline specifications, ablation results, or error bars; this absence prevents verification of the headline claim that C-MOPPO enriches the Pareto front under varied configurations.

    Authors: The full manuscript (Section V) reports quantitative results with baselines (single-objective PPO, random, greedy), ablations on the tandem-queue and freshness terms, multiple-run error bars, and Pareto-front density metrics across configurations. To address the abstract's brevity we will revise it to include concise quantitative highlights (e.g., average improvement ranges and key trade-off metrics). This change will be incorporated. revision: yes

Circularity Check

0 steps flagged

No circularity: MOMDP and C-MOPPO derived from stated problem without reduction to inputs

full rationale

The paper states the FEEL joint optimization problem, introduces a tandem-queue conversion and freshness-augmented accuracy model as modeling choices, casts the problem as an MOMDP, and applies constrained multi-objective PPO (C-MOPPO) to solve it. None of these steps reduce by construction to fitted parameters, self-definitions, or self-citations; the algorithm is a standard extension of PPO to the multi-objective constrained setting. The performance claims rest on experimental evaluation rather than any equation that equates outputs back to the modeling assumptions. This is the normal case of an independent derivation.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

The framework rests on standard assumptions from federated learning and multi-objective RL; the tandem-queue conversion and freshness terms are introduced without independent empirical grounding in the abstract.

pith-pipeline@v0.9.1-grok · 5793 in / 1190 out tokens · 31637 ms · 2026-06-29T22:31:41.521337+00:00 · methodology

discussion (0)

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

Reference graph

Works this paper leans on

49 extracted references · 4 canonical work pages · 3 internal anchors

  1. [1]

    Federated learning for internet of things: A comprehensive survey,

    D. C. Nguyen, M. Ding, P. N. Pathirana, A. Seneviratne, J. Li, and H. Vincent Poor, “Federated learning for internet of things: A comprehensive survey,”IEEE Communications Surveys & Tutorials, vol. 23, no. 3, pp. 1622–1658, 2021

    2021

  2. [2]

    Federated learning-empowered ai-generated content in wireless net- works,

    X. Huang, P. Li, H. Du, J. Kang, D. Niyato, D. I. Kim, and Y . Wu, “Federated learning-empowered ai-generated content in wireless net- works,”IEEE Network, vol. 38, no. 5, pp. 304–313, 2024

    2024

  3. [3]

    Federated edge learning for 6g: Foundations, methodologies, and applications,

    M. Tao, Y . Zhou, Y . Shi, J. Lu, S. Cui, J. Lu, and K. B. Letaief, “Federated edge learning for 6g: Foundations, methodologies, and applications,”Proceedings of the IEEE, pp. 1–39, 2024

    2024

  4. [4]

    Federated learning in mobile edge networks: A comprehensive survey,

    W. Y . B. Lim, N. C. Luong, D. T. Hoang, Y . Jiao, Y .-C. Liang, Q. Yang, D. Niyato, and C. Miao, “Federated learning in mobile edge networks: A comprehensive survey,”IEEE Communications Surveys & Tutorials, vol. 22, no. 3, pp. 2031–2063, 2020

    2031

  5. [5]

    Combining federated learning and edge computing toward ubiquitous intelligence in 6g network: Challenges, recent advances, and future directions,

    Q. Duan, J. Huang, S. Hu, R. Deng, Z. Lu, and S. Yu, “Combining federated learning and edge computing toward ubiquitous intelligence in 6g network: Challenges, recent advances, and future directions,” IEEE Communications Surveys & Tutorials, vol. 25, no. 4, pp. 2892– 2950, 2023

    2023

  6. [6]

    Scheduling for cellular federated edge learning with importance and channel awareness,

    J. Ren, Y . He, D. Wen, G. Yu, K. Huang, and D. Guo, “Scheduling for cellular federated edge learning with importance and channel awareness,”IEEE Transactions on Wireless Communications, vol. 19, no. 11, pp. 7690–7703, 2020

    2020

  7. [7]

    Dynamic scheduling for over-the-air federated edge learning with energy constraints,

    Y . Sun, S. Zhou, Z. Niu, and D. G ¨und¨uz, “Dynamic scheduling for over-the-air federated edge learning with energy constraints,”IEEE Journal on Selected Areas in Communications, vol. 40, no. 1, pp. 227–242, 2022

    2022

  8. [8]

    Optimized power control design for over-the-air federated edge learning,

    X. Cao, G. Zhu, J. Xu, Z. Wang, and S. Cui, “Optimized power control design for over-the-air federated edge learning,”IEEE Journal on Selected Areas in Communications, vol. 40, no. 1, pp. 342–358, 2022

    2022

  9. [9]

    Federated learning while providing model as a service: Joint training and inference optimiza- tion,

    P. Han, S. Wang, Y . Jiao, and J. Huang, “Federated learning while providing model as a service: Joint training and inference optimiza- tion,” inIEEE INFOCOM 2024 - IEEE Conference on Computer Communications, 2024, pp. 631–640

    2024

  10. [10]

    Efficient coordination of federated learning and inference offloading at the edge: A proactive optimization paradigm,

    K. Luo, K. Zhao, T. Ouyang, X. Zhang, Z. Zhou, H. Wang, and X. Chen, “Efficient coordination of federated learning and inference offloading at the edge: A proactive optimization paradigm,”IEEE Transactions on Mobile Computing, vol. 24, no. 1, pp. 407–421, 2025

    2025

  11. [12]

    Fidelity-aware inference services in dt-assisted edge computing via service model retraining,

    X. Ai, W. Liang, Y . Zhang, and W. Xu, “Fidelity-aware inference services in dt-assisted edge computing via service model retraining,” IEEE Transactions on Services Computing, vol. 18, no. 4, pp. 2089– 2102, 2025

    2089

  12. [13]

    Toward dynamic resource allocation and client scheduling in hierarchical federated learning: A two-phase deep reinforcement learning approach,

    X. Chen, Z. Li, W. Ni, X. Wang, S. Zhang, Y . Sun, S. Xu, and Q. Pei, “Toward dynamic resource allocation and client scheduling in hierarchical federated learning: A two-phase deep reinforcement learning approach,”IEEE Transactions on Communications, vol. 72, no. 12, pp. 7798–7813, 2024

    2024

  13. [14]

    Timely communications for remote inference,

    M. K. C. Shisher, Y . Sun, and I.-H. Hou, “Timely communications for remote inference,”IEEE/ACM Transactions on Networking, vol. 32, no. 5, pp. 3824–3839, 2024

    2024

  14. [15]

    Aoi-aware inference services in edge computing via digital twin network slicing,

    Y . Zhang, W. Liang, Z. Xu, W. Xu, and M. Chen, “Aoi-aware inference services in edge computing via digital twin network slicing,” IEEE Transactions on Services Computing, vol. 17, no. 6, pp. 3154– 3170, 2024

    2024

  15. [16]

    Decomposition and meta-drl based multi-objective optimization for 16 asynchronous federated learning in 6g-satellite systems,

    Y . Zhou, L. Lei, X. Zhao, L. You, Y . Sun, and S. Chatzinotas, “Decomposition and meta-drl based multi-objective optimization for 16 asynchronous federated learning in 6g-satellite systems,”IEEE Jour- nal on Selected Areas in Communications, vol. 42, no. 5, 2024

    2024

  16. [17]

    Joint class-balanced client selection and bandwidth allocation for cost- efficient federated learning in mobile edge computing networks,

    J. Tang, X. Li, H. Li, P. Li, X. Wang, and V . C. M. Leung, “Joint class-balanced client selection and bandwidth allocation for cost- efficient federated learning in mobile edge computing networks,” IEEE Transactions on Mobile Computing, vol. 24, no. 7, pp. 5681– 5698, 2025

    2025

  17. [18]

    Joint optimization of device selection and resource allocation for multiple federations in federated edge learning,

    S. Fu, F. Dong, D. Shen, J. Zhang, Z. Huang, and Q. He, “Joint optimization of device selection and resource allocation for multiple federations in federated edge learning,”IEEE Transactions on Ser- vices Computing, vol. 17, no. 1, pp. 251–262, 2024

    2024

  18. [19]

    Heterogeneous com- putation and resource allocation for wireless powered federated edge learning systems,

    J. Feng, W. Zhang, Q. Pei, J. Wu, and X. Lin, “Heterogeneous com- putation and resource allocation for wireless powered federated edge learning systems,”IEEE Transactions on Communications, vol. 70, no. 5, pp. 3220–3233, 2022

    2022

  19. [20]

    Joint task offloading and resource allocation for quality-aware edge-assisted machine learning task inference,

    W. Fan, Z. Chen, Z. Hao, F. Wu, and Y . Liu, “Joint task offloading and resource allocation for quality-aware edge-assisted machine learning task inference,”IEEE Transactions on Vehicular Technology, vol. 72, no. 5, pp. 6739–6752, 2023

    2023

  20. [21]

    Low latency and accuracy-guaranteed dnn inference for uav-assisted iot networks,

    J. Xu, H. Yao, R. Zhang, T. Mai, and M. Guizani, “Low latency and accuracy-guaranteed dnn inference for uav-assisted iot networks,” IEEE Transactions on Cognitive Communications and Networking, pp. 1–1, 2025

    2025

  21. [22]

    Inference load-aware orchestration for hierarchical federated learning,

    A. Lackinger, P. A. Frangoudis, I. ˇCili´c, A. Furutanpey, I. Murturi, I. P. ˇZarko, and S. Dustdar, “Inference load-aware orchestration for hierarchical federated learning,” in2024 IEEE 49th Conference on Local Computer Networks (LCN), 2024, pp. 1–9

    2024

  22. [23]

    Joint computational resource allocation and layer partitioning for federated learning,

    G. Qiang, F. Fang, H. Chen, and X. Wang, “Joint computational resource allocation and layer partitioning for federated learning,” IEEE Internet of Things Journal, pp. 1–1, 2025

    2025

  23. [24]

    Personalized federated deep rein- forcement learning for heterogeneous edge content caching networks,

    Z. Li, T. Li, H. Liu, and T.-T. Chan, “Personalized federated deep rein- forcement learning for heterogeneous edge content caching networks,” in2024 22nd International Symposium on Modeling and Optimization in Mobile, Ad Hoc, and Wireless Networks (WiOpt), 2024, pp. 313– 320

    2024

  24. [25]

    Differentially private federated multi-task learning framework for enhancing human- to-virtual connectivity in human digital twin,

    S. D. Okegbile, J. Cai, H. Zheng, J. Chen, and C. Yi, “Differentially private federated multi-task learning framework for enhancing human- to-virtual connectivity in human digital twin,”IEEE Journal on Selected Areas in Communications, vol. 41, no. 11, pp. 3533–3547, 2023

    2023

  25. [26]

    Federated learning with noisy user feedback,

    R. Sharma, A. Ramakrishna, A. MacLaughlin, A. Rumshisky, J. Ma- jmudar, C. Chung, S. Avestimehr, and R. Gupta, “Federated learning with noisy user feedback,”arXiv preprint arXiv:2205.03092, 2022

    work page arXiv 2022

  26. [27]

    Communication-efficient learning of deep networks from decentralized data,

    B. McMahan, E. Moore, D. Ramage, S. Hampson, and B. A. y Arcas, “Communication-efficient learning of deep networks from decentralized data,” inArtificial intelligence and statistics. PMLR, 2017, pp. 1273–1282

    2017

  27. [28]

    Energy efficient federated learning over wireless communication networks,

    Z. Yang, M. Chen, W. Saad, C. S. Hong, and M. Shikh-Bahaei, “Energy efficient federated learning over wireless communication networks,”IEEE Transactions on Wireless Communications, vol. 20, no. 3, pp. 1935–1949, 2021

    1935

  28. [29]

    Joint device scheduling and resource allocation for latency constrained wireless federated learning,

    W. Shi, S. Zhou, Z. Niu, M. Jiang, and L. Geng, “Joint device scheduling and resource allocation for latency constrained wireless federated learning,”IEEE Transactions on Wireless Communications, vol. 20, no. 1, pp. 453–467, 2021

    2021

  29. [30]

    Resource allocation for multiuser edge inference with batching and early exiting,

    Z. Liu, Q. Lan, and K. Huang, “Resource allocation for multiuser edge inference with batching and early exiting,”IEEE Journal on Selected Areas in Communications, vol. 41, no. 4, pp. 1186–1200, 2023

    2023

  30. [31]

    Efficient privacy auditing in federated learning,

    H. Chang, B. Edwards, A. S. Paul, and R. Shokri, “Efficient privacy auditing in federated learning,” in33rd USENIX Security Symposium (USENIX Security 24), 2024, pp. 307–323

    2024

  31. [32]

    Age of information minimization in uav-enabled iot networks via federated reinforcement learning,

    X. Zhang, H. Xing, Y . Shen, J. Xu, and S. Cui, “Age of information minimization in uav-enabled iot networks via federated reinforcement learning,”IEEE Transactions on Wireless Communications, vol. 24, no. 9, pp. 7923–7939, 2025

    2025

  32. [33]

    Adaptive layer-wise personalized federated deep reinforcement learning for heterogeneous edge caching,

    T. Li, Z. Li, H. Liu, C. Yang, T.-T. Chan, and J. Cai, “Adaptive layer-wise personalized federated deep reinforcement learning for heterogeneous edge caching,”IEEE Transactions on Cognitive Com- munications and Networking, vol. 12, pp. 4532–4546, 2026

    2026

  33. [34]

    Deep Learning Scaling is Predictable, Empirically

    J. Hestness, S. Narang, N. Ardalani, G. Diamos, H. Jun, H. Kianine- jad, M. M. A. Patwary, Y . Yang, and Y . Zhou, “Deep learning scaling is predictable, empirically,”arXiv preprint arXiv:1712.00409, 2017

    work page internal anchor Pith review Pith/arXiv arXiv 2017

  34. [35]

    A learning-based incentive mechanism for federated learning,

    Y . Zhan, P. Li, Z. Qu, D. Zeng, and S. Guo, “A learning-based incentive mechanism for federated learning,”IEEE Internet of Things Journal, vol. 7, no. 7, pp. 6360–6368, 2020

    2020

  35. [36]

    Age and value of information optimization for systems with multi-class updates,

    A. Arafa and R. D. Yates, “Age and value of information optimization for systems with multi-class updates,” inICC 2024 - IEEE Interna- tional Conference on Communications, 2024, pp. 195–200

    2024

  36. [37]

    Joint optimization of model retraining and inference services in dt-assisted edge computing,

    X. Ai, W. Liang, and C. Liu, “Joint optimization of model retraining and inference services in dt-assisted edge computing,”IEEE Trans- actions on Networking, vol. 34, pp. 1804–1819, 2026

    2026

  37. [38]

    Joint age-based client selection and re- source allocation for communication-efficient federated learning over noma networks,

    B. Wu, F. Fang, and X. Wang, “Joint age-based client selection and re- source allocation for communication-efficient federated learning over noma networks,”IEEE Transactions on Communications, vol. 72, no. 1, pp. 179–192, 2024

    2024

  38. [39]

    Proximal Policy Optimization Algorithms

    J. Schulman, F. Wolski, P. Dhariwal, A. Radford, and O. Klimov, “Proximal policy optimization algorithms,”arXiv preprint arXiv:1707.06347, 2017

    work page internal anchor Pith review Pith/arXiv arXiv 2017

  39. [40]

    High-Dimensional Continuous Control Using Generalized Advantage Estimation

    J. Schulman, P. Moritz, S. Levine, M. Jordan, and P. Abbeel, “High- dimensional continuous control using generalized advantage estima- tion,”arXiv preprint arXiv:1506.02438, 2015

    work page internal anchor Pith review Pith/arXiv arXiv 2015

  40. [41]

    A fast and elitist multiobjective genetic algorithm: Nsga-ii,

    K. Deb, A. Pratap, S. Agarwal, and T. Meyarivan, “A fast and elitist multiobjective genetic algorithm: Nsga-ii,”IEEE Transactions on Evolutionary Computation, vol. 6, no. 2, pp. 182–197, 2002

    2002

  41. [42]

    Efficient discovery of pareto front for multi-objective reinforcement learning,

    R. Liu, Y . Pan, L. Xu, L. Song, P. You, Y . Chen, and J. Bian, “Efficient discovery of pareto front for multi-objective reinforcement learning,” inThe Thirteenth International Conference on Learning Representations, 2025

    2025

  42. [43]

    Ipo: Interior-point policy optimization under constraints,

    Y . Liu, J. Ding, and X. Liu, “Ipo: Interior-point policy optimization under constraints,” inProceedings of the AAAI conference on artificial intelligence, vol. 34, no. 04, 2020, pp. 4940–4947

    2020

  43. [44]

    Boyd and L

    S. Boyd and L. Vandenberghe,Convex Optimization. Cambridge University Press, 2004

    2004

  44. [45]

    Collaborative ground-space communications via evolution- ary multi-objective deep reinforcement learning,

    J. Li, G. Sun, Q. Wu, D. Niyato, J. Kang, A. Jamalipour, and V . C. M. Leung, “Collaborative ground-space communications via evolution- ary multi-objective deep reinforcement learning,”IEEE Journal on Selected Areas in Communications, vol. 42, no. 12, pp. 3395–3411, 2024

    2024

  45. [46]

    Mobile communications, computing, and caching resources allocation for diverse services via multi-objective proximal policy optimization,

    Z. Chen, B. Yin, H. Zhu, Y . Li, M. Tao, and W. Zhang, “Mobile communications, computing, and caching resources allocation for diverse services via multi-objective proximal policy optimization,” IEEE Transactions on Communications, vol. 70, no. 7, pp. 4498– 4512, 2022

    2022

  46. [47]

    Prediction- guided multi-objective reinforcement learning for continuous robot control,

    J. Xu, Y . Tian, P. Ma, D. Rus, S. Sueda, and W. Matusik, “Prediction- guided multi-objective reinforcement learning for continuous robot control,” inInternational conference on machine learning, 2020, pp. 10 607–10 616

    2020

  47. [48]

    Moea/d: A multiobjective evolutionary algo- rithm based on decomposition,

    Q. Zhang and H. Li, “Moea/d: A multiobjective evolutionary algo- rithm based on decomposition,”IEEE Transactions on Evolutionary Computation, vol. 11, no. 6, pp. 712–731, 2007

    2007

  48. [49]

    Learning multiple layers of features from tiny images,

    A. Krizhevsky, G. Hintonet al., “Learning multiple layers of features from tiny images,” 2009

    2009

  49. [50]

    On the local cache update rules in streaming federated learning,

    H. Wang, J. Bian, and J. Xu, “On the local cache update rules in streaming federated learning,”IEEE Internet of Things Journal, vol. 11, no. 6, pp. 10 808–10 816, 2024

    2024