Pith. sign in

REVIEW 5 major objections 5 minor 25 references

Adaptive, Efficient and Fair Resource Allocation in Cloud Datacenters leveraging Weighted A3C Deep Reinforcement Learning

T0 review · 5 major / 5 minor · reviewed 2026-08-07 · deepseek-v4-flash

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

Pith's one-line read A weighted A3C scheduler beats heuristic and deep-RL baselines on latency, energy, and dismissals in cloud simulations.

desk verdict A reasonable weighted-reward A3C extension, but the evaluation is single-run, internally inconsistent, and the headline outperformance claim is not supported. read the letter →

arxiv 2506.00929 v1 pith:ZHUEGENA submitted 2025-06-01 cs.DC

Suchi Kumari , Dhruv Mishra This is my paper
classification cs.DC
keywords cloudresourceallocationdeepreinforcementlearningA3Cjobschedulingfairnesspriorityenergyefficiencyactor-critic
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

This paper claims that a deep reinforcement learning scheduler for cloud datacenters can serve high-priority jobs without starving low-priority ones while also cutting latency, energy use, and job rejections. The proposed Weighted A3C (WA3C) model extends the A3C actor-critic algorithm with a five-term reward that combines quality-of-service, energy, priority satisfaction, fairness, and dismissal penalties, plus a priority-weighted softmax that biases action selection toward urgent jobs. In simulations with synthetic workloads and Google cluster-usage traces, WA3C reports lower average latency, lower energy consumption, a lower job dismissal rate, and higher total reward than Round Robin, Shortest Job First, Longest Job First, Tetris, Random, standard A3C, Deep Q-Learning, and Policy Gradient. If correct, this gives cloud operators a scheduler that adapts online to workload shifts without retraining from scratch. The paper's own discussion acknowledges that deep-RL training is costly, reward shaping requires tuning, and integration with existing orchestration frameworks is non-trivial.

What carries the argument

The load-bearing object is the composite reward function of Eq. (1), which forces the agent to optimize latency, energy, priority satisfaction, fairness, and dismissal avoidance simultaneously; without it the scheduler reduces to plain A3C. The second piece is the priority-weighted softmax of Eq. (10), which injects job urgency directly into action selection rather than relying only on learned Q-values. The third is the asynchronous worker architecture, in which each VM runs a worker that accumulates gradients and periodically synchronizes with the global actor and critic, providing scalability and online adaptation. All results are generated from this combination together with the hand-calibrated energy model of Eqs. (5)–(6), which converts CPU utilization, CPI, and MAPI into a per-job energy estimate.

What would settle it

Measure actual power draw of a cluster running WA3C and the baselines on the Google cluster-usage trace, then compare measured kWh against the simulated values from Eqs. (4)–(6); if the measured ranking differs from the simulated one, the paper's energy-advantage claim is refuted.

Watch

Extended reading notes

Core claim

The central claim is that adding job priority and fairness to the reward signal of an asynchronous actor-critic learner transforms the scheduler: WA3C maximizes $R_t = w_1 R^{QoS}_t + w_2 R^E_t + w_3 R^P_t + w_4 R^F_t + w_5 R^D_t$, with the penalty terms shaping behaviour so that urgent jobs are served promptly, utilization variance is penalized, and dismissals are discouraged. Action selection then uses the priority-weighted softmax $\pi(a_t=j) \propto \exp(Q(s_t,j) + \beta P_j)$, and $n$ asynchronous workers, one per VM, push gradients to a shared global actor and critic. The authors report that this configuration converges to better cumulative rewards, lower normalized latency, and lower dismissal rates than standard A3C, DQL, and Policy Gradient, and outperforms non-RL schedulers on every metric across all tested system loads. The simulations use both synthetic 1000-job traces and the Google cluster-usage dataset.

Load-bearing premise

The energy and fairness results come from a hand-set simulation model—power drawn as a linear ramp in CPU utilization and per-job constants $z=0.3$ and $\theta_j=0.2$, plus hand-chosen reward weights $w_1$ through $w_5$—rather than from measured power readings, so the reported energy and fairness advantages may not transfer to real hardware if these constants are wrong.

Editorial extensions

If this is right

  • If the reported results hold, cloud schedulers can lower average latency and energy use while respecting job priorities, without retraining from scratch when workload mix shifts.
  • Operators can rebalance scheduler behavior by changing the five reward weights, e.g., increasing $w_4$ to strengthen fairness in multi-tenant systems.
  • The priority-weighted softmax is a generic mechanism for injecting urgency into DRL schedulers beyond clouds, such as edge or serverless platforms.
  • The worker-per-VM design ties scalability to cluster size, so larger deployments would simply add more asynchronous workers.
  • The reported curves indicate WA3C's advantage persists as average system load rises from 0.4 to 2.8, implying the policy generalizes to congested conditions.

Reading between the lines

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

  • A reader should double-check the sensitivity analysis: Section IV-C calls the discount factor $\beta$ and the learning rate $\gamma$, whereas the model definition in Sections III-A and III-C sets $\gamma=0.95$ as the discount factor and $\beta=2.0$ as the priority weight; if the labels were swapped in the experiments, the reported best hyperparameters do not describe the model as specified.
  • A natural next test is to replace the hand-set energy model with measured power traces, turning the simulated kWh savings into verifiable operational savings.
  • The fairness term penalizes variance in resource utilization; for workloads with heterogeneous resource demands, a size-aware fairness metric would be needed to avoid penalizing legitimate differences.
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

5 major / 5 minor

Summary. The paper proposes Weighted A3C (WA3C), an extension of the A3C deep reinforcement learning framework for cloud job scheduling. The contribution is a composite reward function that linearly combines QoS, energy, priority, fairness, and dismissal-penalty terms, together with a priority-weighted softmax action-selection mechanism. The authors evaluate WA3C against non-RL heuristics (RR, SJF, LJF, Tetris, Random) and DRL baselines (A3C, DQL, PG) on synthetic workloads, claiming that WA3C consistently outperforms all baselines in average job latency, energy consumption, job dismissal rate, and total reward. The paper also reports a sensitivity analysis of the discount factor and learning rate.

Significance. If the headline claim were reliably established, WA3C would be a practically relevant contribution to multi-objective cloud scheduling: the idea of injecting job priority and fairness directly into a weighted A3C reward is timely, and the asynchronous worker design is a reasonable starting point. However, the manuscript does not provide reproducible code, data, seeds, error bars, or significance tests, and it contains multiple internal inconsistencies in notation, units, and reported reward values. Because the entire contribution is empirical, these issues prevent the reader from verifying the central claim; the significance of the work cannot currently be assessed.

major comments (5)
  1. [Section III-C, Algorithm 1, Section IV-C] The symbol β is overloaded with at least three different meanings. Eq. (10) defines β=2.0 as the priority scaling factor in the softmax action selection; Algorithm 1 lines 16 and 18 use β in the advantage and TD-error formulas as a discount-like coefficient; and Section IV-C treats β as the discount factor varied over {0.6, 0.7, 0.8, 0.9} while γ, defined as the discount factor γ=0.95 in Section III-A, is called the learning rate. As a result, the sensitivity analysis in Figures 3 and 4 cannot be mapped to the algorithm as written, and the reported A3C-style updates are not well-defined.
  2. [Section IV-D and Fig. 5(d)] The 'Total Reward per Episode' panel for non-RL baselines is semantically incoherent. RR, SJF, LJF, Tetris, and Random are heuristics that do not optimize the reward in Eq. (1), and no reward formula is provided for them, so plotting their 'total reward' has no clear meaning. Additionally, the text says the x-axis is average system load (0.4 to 2.8), while the caption and paragraph refer to 'training epochs'; this mismatch makes the panel uninterpretable as evidence.
  3. [Section IV (entire evaluation)] The load-bearing claim that WA3C 'consistently outperforms' baselines is supported only by single simulation curves. No code, data, random seeds, number of repeated runs, confidence intervals, or significance tests are reported. Because the evaluation is the only evidence for the paper's central claim, the current results are not independently checkable and cannot establish consistent outperformance.
  4. [Section III-B.2, Section IV-D, Section IV-E] The energy model in Eqs. (4)-(6) relies on hand-set constants (z=0.3, θ_j=0.2) and is not validated against any measured power data. The reported units are also inconsistent: Section IV-D reports energy in kWh, whereas Eq. (5) with W in Watts and τ in seconds yields Joules, and Section IV-E mixes 'kW' and 'kWh' within the same sentence. The energy comparisons may therefore reflect modeling assumptions rather than real datacenter behavior.
  5. [Abstract, Section III-B.4, Section V-A] Fairness is presented as a key contribution, but no experimental result reports a fairness metric. Eq. (8) defines the fairness reward via the variance of resource utilization, while Section V-A states that fairness is measured using 'Jain's Index'; these are different quantities. The fairness component is never isolated, reported, or validated, so the 'fair' part of the headline claim is unsupported.
minor comments (5)
  1. [Section IV-A] The text says real-world Google Cluster Workload traces were 'incorporated,' but the results section reports only synthetic job traces; please clarify which experiments use which dataset.
  2. [Section V-A] The discussion refers to 'a tunable parameter α' for priority in the softmax policy, while Eq. (10) and Section III-C use β=2.0; the notation should be consistent.
  3. [Algorithm 2] Line 7 writes the critic gradient as ∇(R_i−V(s_i))^2, but the squared TD error is not defined with the correct sign or factor, and R_i is not defined (immediate reward versus discounted return).
  4. [Fig. 6(b)] The energy axis mixes units ('12 kW' and '10 kW h'); please standardize to a single unit, e.g., energy per job in kWh or Joules.
  5. [References] Reference [21] is cited as the source of the A2C model, but the cited Sutton and Barto book introduces actor-critic methods more generally; the attribution should be corrected.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: WA3C is an empirically evaluated algorithm; its reward-aligned metrics and self-citations are not derivation-level circular steps.

full rationale

The paper proposes an RL-based scheduling algorithm and evaluates it in simulation; it does not derive a prediction from fitted constants or import a load-bearing uniqueness theorem from its own prior work. The reward function in Eq. (1) linearly combines latency, energy, priority, fairness, and dismissal components (Eqs. 2-9), and the reported performance metrics in Figs. 5-6 are the same simulated quantities that the reward directly optimizes. This is objective-aligned evaluation rather than circularity: the paper hand-sets weights and reports the resulting simulated trajectories; it never fits those quantities to a subset of data and then presents them as an independent prediction. The priority-weighted softmax in Eq. (10) incorporates job priority P_j by design, so a tendency to favor high-priority jobs is a design consequence, not a hidden equivalence. The only self-citation, reference [8], appears in the related-work discussion and is not load-bearing. There are serious internal inconsistencies that undermine reproducibility and validity, including β defined as a priority scaling factor in Eq. (10) but treated as a discount factor in Section IV-C, γ defined as a discount factor in Section III-A but called a learning rate in Section IV-C, the use of β in the TD error in Algorithm 1 while Section III uses γ, and Fig. 5(d) assigning reward curves to non-RL heuristics that have no defined reward function. These are correctness and evidence-quality problems, not circularity under the specified rubric, because no load-bearing claim reduces to its own input by construction. Accordingly, no circular step is flagged and the circularity score is 0.

Assumptions & free parameters 10 free parameters · 5 assumptions · 0 invented entities

The central claim rests on a hand-set reward weight vector, several hand-tuned constants in the energy and latency models, and a simulator whose behavior is never validated against physical measurements or a deployed scheduler. These choices, rather than external data, determine much of the reported performance.

free parameters (10)
  • Reward weights w1-w5 = 0.25, 0.2, 0.25, 0.15, 0.15
    Eq. (1) hand-set; no sensitivity analysis is reported for these values, and they control the entire tradeoff.
  • Priority scaling beta = 2.0
    Eq. (10); set by hand, not swept.
  • Discount factor = 0.95 stated; 0.9 used in tuning
    Section III-A sets gamma=0.95, but Section IV-C treats beta as discount and chooses beta=0.9; notation and value inconsistent.
  • Learning rate = 0.01 chosen from {0.1, 0.01, 0.001, 0.0001}
    Section IV-C grid search; no schedule or warmup specified.
  • z (memory vs compute weight) = 0.3
    Eq. (6), hand-set coefficient in energy model.
  • theta_j (interference factor) = 0.2
    Section III-B.1, set to 0.2 based on empirical tuning.
  • mu (dismissal penalty constant) = 0.5
    Eq. (9).
  • alpha (energy penalty scaling) = not reported
    Eq. (4) uses alpha but no value is given.
  • lambda (fairness regularization) = not reported
    Eq. (8) uses lambda but no value is given.
  • T_max (overload dismissal threshold) = not reported
    Algorithm 3 uses T_max but the threshold is undefined.
assumptions (5)
  • domain assumption Cloud job scheduling can be captured by an MDP with state st, action select job j, transition P, and discount factor gamma.
    Section III-A formulates the problem as an MDP tuple; the fidelity of this abstraction to real cloud schedulers is assumed.
  • ad hoc to paper The linear power model in Eq. (5) with instruction-dependent coefficient iota (Eq. 6) approximates datacenter energy consumption.
    Section III-B.2; no real power measurements or validation against an actual cluster are provided.
  • ad hoc to paper Execution time is T_exec = C_j / (f * (1 - theta_j)) with theta_j = 0.2.
    Section III-B.1; the interference value is set by hand.
  • domain assumption Synthetic 1000-job traces and Google Cluster trace subsets are representative of production datacenter workloads.
    Section IV-A; no preprocessing details or trace filtering criteria are given.
  • standard math Standard RL convergence assumptions for RMSProp, TD error, and asynchronous parameter updates hold.
    Section III-D and Algorithm 2 rely on standard stochastic approximation; no convergence proof is given.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Adaptive, Efficient and Fair Resource Allocation in Cloud Datacenters leveraging Weighted A3C Deep Reinforcement Learning." pith.science (2026). https://pith.science/paper/ZHUEGENA

@misc{pith2026250600929,
  author       = {Pith},
  title        = {Pith review of: Adaptive, Efficient and Fair Resource Allocation in Cloud Datacenters leveraging Weighted A3C Deep Reinforcement Learning},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/ZHUEGENA}},
  note         = {Machine review of arXiv:2506.00929}
}
read the original abstract

Cloud data centres demand adaptive, efficient, and fair resource allocation techniques due to heterogeneous workloads with varying priorities. However, most existing approaches struggle to cope with dynamic traffic patterns, often resulting in suboptimal fairness, increased latency, and higher energy consumption. To overcome these limitations, we propose a novel method called Weighted Actor-Critic Deep Reinforcement Learning (WA3C). Unlike static rule-based schedulers, WA3C continuously learns from the environment, making it resilient to changing workload patterns and system dynamics. Furthermore, the algorithm incorporates a multi-objective reward structure that balances trade-offs among latency, throughput, energy consumption, and fairness. This adaptability makes WA3C well-suited for modern multi-tenant cloud infrastructures, where diverse applications often compete for limited resources. WA3C also supports online learning, allowing it to adapt in real time to shifting workload compositions without the need for retraining from scratch. The model's architecture is designed to be lightweight and scalable, ensuring feasibility even in large-scale deployments. Additionally, WA3C introduces a priority-aware advantage estimator that better captures the urgency of tasks, enhancing scheduling precision. As a result, WA3C achieves more effective convergence, lower latency, and balanced resource allocation among jobs. Extensive experiments using synthetic job traces demonstrate that WA3C consistently outperforms both traditional and reinforcement learning-based baselines, highlighting its potential for real-world deployment in large-scale cloud systems.

Figures

Figures reproduced from arXiv: 2506.00929 by the authors.

Figure 1
Figure 1. n-Workers collaboration in WA3C model [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 2
Figure 2. Detailed Two Worker collaboration in WA3C model [PITH_FULL_IMAGE:figures/full_fig_p005_2.png] view at source ↗
Figure 3
Figure 3. Sensitivity of the proposed WA3C model to the varying Discount Rates 2) Effect of Learning Rate (γ) on WA3C framework: In [PITH_FULL_IMAGE:figures/full_fig_p008_3.png] view at source ↗
Figures from the paper (4 more)
Figure 4
Figure 4. Figure 4: Sensitivity to varying Learning Rates From Figs. 3 and 4, we can say that WA3C is most effective when configured with a higher discount factor (β = 0.9) and a moderate learning rate (γ = 0.01). These values ensure that the agent optimally balances short- and long-term …
Figure 5
Figure 5. Figure 5: Comparison with non-Reinforcement Learning models considering (a) Average Job Latency, (b) Average Energy Consumption, (c) Job Dismissal [PITH_FULL_IMAGE:figures/full_fig_p009_5.png]
Figure 6
Figure 6. Figure 6: Comparison with Deep Reinforcement Learning models considering (a) Normalized Job Latency, (b) Average Energy Consumption, (c) Job Dismissal [PITH_FULL_IMAGE:figures/full_fig_p010_6.png]
Figure 7
Figure 7. Figure 7: Flow of execution and collaboration between agents in the [PITH_FULL_IMAGE:figures/full_fig_p011_7.png]

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

25 extracted references · 23 canonical work pages

  1. [1]

    Dynamic resource allocation for cloud computing using reinforcement learning,

    J. Xu, J. Li, and J. Lin, “Dynamic resource allocation for cloud computing using reinforcement learning,” Journal of Grid Computing , vol. 11, no. 3, pp. 429–445, 2013

    work page 2013

  2. [2]

    Resource man- agement with deep reinforcement learning,

    H. Mao, M. Alizadeh, I. Menache, and S. Kandula, “Resource man- agement with deep reinforcement learning,” in Proceedings of the 15th ACM Workshop on Hot Topics in Networks , 2016

    work page 2016

  3. [3]

    A2c-drl: Dynamic scheduling for stochastic edge-cloud environments using a2c and deep reinforcement learning,

    J. Lu, J. Yang, S. Li, Y . Li, W. Jiang, J. Dai, and J. Hu, “A2c-drl: Dynamic scheduling for stochastic edge-cloud environments using a2c and deep reinforcement learning,” IEEE Internet of Things Journal , 2024

    work page 2024

  4. [4]

    Asynchronous methods for deep reinforcement learning,

    V . Mnih, A. P. Badia, M. Mirza, A. Graves, T. Lillicrap, T. Harley, D. Silver, and K. Kavukcuoglu, “Asynchronous methods for deep reinforcement learning,” Proceedings of ICML , 2016

    work page 2016

  5. [5]

    Adaptive and efficient resource allocation in cloud datacenters using actor-critic deep reinforcement learning,

    Z. Chen, J. Hu, G. Min, C. Luo, and T. El-Ghazawi, “Adaptive and efficient resource allocation in cloud datacenters using actor-critic deep reinforcement learning,” IEEE Transactions on Parallel and Distributed Systems, vol. 33, no. 8, pp. 1911–1923, 2021

    work page 1911

  6. [6]

    Balancing energy- efficiency and service quality for cloud applications,

    U. Sharma, P. Shenoy, S. Sahu, and A. Shaikh, “Balancing energy- efficiency and service quality for cloud applications,” Proceedings of ACM SIGMETRICS, pp. 3–14, 2011

    work page 2011

  7. [7]

    Dynamic vm placement and migration using machine learning in cloud,

    R. Ghosh and V . Naik, “Dynamic vm placement and migration using machine learning in cloud,” International Conference on Cloud Com- puting, pp. 1–8, 2015

    work page 2015

  8. [8]

    Enhancing Cloud Task Scheduling Using a Hybrid Particle Swarm and Grey Wolf Optimization Approach

    R. Prasad, A. Roy, and S. Kumari, “Enhancing cloud task scheduling using a hybrid particle swarm and grey wolf optimization approach,” arXiv preprint arXiv:2505.15171 , 2025

    work page Pith review arXiv 2025

Show all 25 references
  1. [9]

    A game-theoretic method of fair resource allocation for cloud computing services,

    G. Wei, A. V . Vasilakos, Y . Zheng, and N. Xiong, “A game-theoretic method of fair resource allocation for cloud computing services,” The journal of supercomputing , vol. 54, pp. 252–269, 2010

    2010

  2. [10]

    Resource allocation mechanisms and approaches on the internet of things,

    Z. Ghanbari, N. Jafari Navimipour, M. Hosseinzadeh, and A. Darwesh, “Resource allocation mechanisms and approaches on the internet of things,” Cluster Computing, vol. 22, no. 4, pp. 1253–1282, 2019

    2019

  3. [11]

    Dynamic resource allocation method based on symbiotic organism search algorithm in cloud computing,

    A. Belgacem, K. Beghdad-Bey, and H. Nacer, “Dynamic resource allocation method based on symbiotic organism search algorithm in cloud computing,” IEEE Transactions on Cloud Computing , vol. 10, no. 3, pp. 1714–1725, 2020

    2020

  4. [12]

    Sg-pbfs: Shortest gap-priority based fair scheduling technique for job scheduling in cloud environment,

    S. A. Murad, Z. R. M. Azmi, A. J. M. Muzahid, M. K. B. Bhuiyan, M. Saib, N. Rahimi, N. J. Prottasha, and A. K. Bairagi, “Sg-pbfs: Shortest gap-priority based fair scheduling technique for job scheduling in cloud environment,” Future Generation Computer Systems , vol. 150, pp. ...

    2024

  5. [13]

    Learning-based resource provisioning for cloud applications,

    W. Shi and Y . Hong, “Learning-based resource provisioning for cloud applications,” IEEE Transactions on Parallel and Distributed Systems , vol. 22, no. 12, pp. 2035–2042, 2011

    2011

  6. [14]

    Integrated deep learning method for workload and resource prediction in cloud systems,

    J. Bi, S. Li, H. Yuan, and M. Zhou, “Integrated deep learning method for workload and resource prediction in cloud systems,” Neurocomputing, vol. 424, pp. 35–48, 2021

    2021

  7. [15]

    Resource overbooking and application profiling in shared hosting platforms,

    B. Urgaonkar, P. Shenoy, A. Chandra, P. Goyal, and T. Wood, “Resource overbooking and application profiling in shared hosting platforms,” in OSDI, 2005, pp. 239–254

    2005

  8. [16]

    Optimal re- source allocation of cloud-based spark applications,

    M. Lattuada, E. Barbierato, E. Gianniti, and D. Ardagna, “Optimal re- source allocation of cloud-based spark applications,” IEEE Transactions on Cloud Computing , vol. 10, no. 2, pp. 1301–1316, 2020

    2020

  9. [17]

    Resource provi- sioning using workload clustering in cloud computing environment: a hybrid approach,

    A. Shahidinejad, M. Ghobaei-Arani, and M. Masdari, “Resource provi- sioning using workload clustering in cloud computing environment: a hybrid approach,” Cluster Computing, vol. 24, no. 1, pp. 319–342, 2021

    2021

  10. [18]

    An energy aware resource allocation based on combination of cnn and gru for virtual machine selection,

    Z. Khodaverdian, H. Sadr, S. A. Edalatpanah, and M. Nazari, “An energy aware resource allocation based on combination of cnn and gru for virtual machine selection,” Multimedia tools and applications , vol. 83, no. 9, pp. 25 769–25 796, 2024

    2024

  11. [19]

    Machine and deep learning for resource allocation in multi-access edge computing: A survey,

    H. Djigal, J. Xu, L. Liu, and Y . Zhang, “Machine and deep learning for resource allocation in multi-access edge computing: A survey,” IEEE Communications Surveys & Tutorials , vol. 24, no. 4, pp. 2449–2494, 2022

    2022

  12. [20]

    Aquatope: Qos-and-uncertainty- aware resource management for multi-stage serverless workflows,

    Z. Zhou, Y . Zhang, and C. Delimitrou, “Aquatope: Qos-and-uncertainty- aware resource management for multi-stage serverless workflows,” in Proceedings of the 28th ACM International Conference on Architectural Support for Programming Languages and Operating Systems, Volume 1, 2...

    2022

  13. [21]

    R. S. Sutton, A. G. Barto et al., Reinforcement learning: An introduction. MIT press Cambridge, 1998, vol. 1, no. 1

    1998

  14. [22]

    Resource allocation with workload-time windows for cloud-based software ser- vices: a deep reinforcement learning approach,

    X. Chen, L. Yang, Z. Chen, G. Min, X. Zheng, and C. Rong, “Resource allocation with workload-time windows for cloud-based software ser- vices: a deep reinforcement learning approach,” IEEE Transactions on Cloud Computing, vol. 11, no. 2, pp. 1871–1885, 2022

    2022

  15. [23]

    Deep reinforcement learning-based methods for resource scheduling in cloud computing: A review and future directions,

    G. Zhou, W. Tian, R. Buyya, R. Xue, and L. Song, “Deep reinforcement learning-based methods for resource scheduling in cloud computing: A review and future directions,” Artificial Intelligence Review , vol. 57, no. 5, p. 124, 2024

    2024

  16. [24]

    A sufficient condition for convergences of adam and rmsprop,

    F. Zou, L. Shen, Z. Jie, W. Zhang, and W. Liu, “A sufficient condition for convergences of adam and rmsprop,” in Proceedings of the IEEE/CVF Conference on computer vision and pattern recognition , 2019, pp. 11 127–11 135

    2019

  17. [25]

    Google cluster-usage traces: format+ schema,

    C. Reiss, J. Wilkes, and J. M. Hellerstein, “Google cluster-usage traces: format+ schema,” in Google Inc. Technical Report , 2011. [Online]. Available: https://github.com/google/cluster-data

    2011

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.