Pith. sign in

REVIEW 3 major objections 4 minor 32 references

Distributed Collaborative Inference System in Next-Generation Networks and Communication

T0 review · 3 major / 4 minor · reviewed 2026-08-12 · deepseek-v4-flash

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

Pith's one-line read A cloud-edge-end inference system claims to cut transformer classification latency by up to 17% by pruning low-attention tokens and exiting early, without sacrificing accuracy.

desk verdict Table I kills the paper's headline claim: every early-exit row loses accuracy, and the 17% figure is a post hoc selection. read the letter →

arxiv 2412.12102 v1 pith:JFPCJUKC submitted 2024-11-16 cs.NI cs.AI

Chuan Zhang , Xixi Zheng , Xiaolong Tao , Chenfei Hu , Weiting Zhang , Liehuang Zhu This is my paper · ORCID
classification cs.NIcs.AI
keywords collaborativeinferenceearlyexitattention-basedpruningtaskoffloadingtransformer6Gnetworkslatencyoptimizationsentimentanalysis
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 proposes a three-level 'cloud-edge-end' inference system for transformer-based classifiers in mobile networks. The central claim is that combining confidence-based task offloading, attention-based pruning of input text, an early-exit mechanism, and weighted aggregation of model outputs reduces inference time by up to 17% compared with existing work while preserving inference accuracy. The authors test the system on IMDB sentiment analysis with three BERT-family models and recommend a confidence threshold of 0.8 and an early-exit difference threshold of 0.00001 as the configuration that best balances speed and accuracy. If the claim holds, resource-limited devices could serve many requests locally and send only the hard, low-confidence cases to edge or cloud, usually in pruned form.

What carries the argument

The load-bearing mechanism is the reuse of a small model's attention weights as a token-importance signal for pruning. Equation (10) retains token $y$ only if its accumulated attention importance exceeds $\alpha$ times the sentence-average importance; the pruned raw text is then re-tokenized by the next model under three prefix/suffix transfer rules. This is paired with a temperature-scaled softmax confidence, Equation (6), which drives the probabilistic offloading decision in Equation (7), and a patience-based early exit, Equations (13)--(14), that stops a model when adjacent-layer softmax differences stay below $\tau$ for $p$ consecutive layers. Together these modules turn a cascade of separately trained transformers into one latency-adaptive pipeline whose final output is a weighted ensemble of all models that processed the task.

What would settle it

A decisive test is to re-run the IMDB experiment at the recommended configuration and vary the pruning coefficient $\alpha$ in Equation (10) from 0 to 1 while holding the early-exit and confidence thresholds fixed; if accuracy drops sharply as soon as $\alpha$ exceeds a small value, then pruning is removing tokens the larger model depends on. The same experiment should report accuracy and time for the early-exit-disabled baseline, so the 'no accuracy sacrifice' part of the 17% claim can be checked directly.

Watch

Extended reading notes

Core claim

The paper's central discovery is that the intermediate outputs of a small local transformer—attention weights and softmax confidences—can be reused as cheap control signals for a distributed inference pipeline. Low-confidence samples are probabilistically offloaded to larger models; before transmission, tokens whose accumulated attention importance falls below a threshold are pruned from the raw text; each model may stop early when consecutive layers give nearly identical softmax results; and every model that processed the task contributes to an accuracy-weighted final answer. The paper reports that on IMDB this combination reduces end-to-end inference time by up to 17% compared with the Tabi baseline while keeping accuracy within the system's target, with the recommended settings (confidence threshold 0.8, early-exit threshold 0.00001) yielding 0.9012 accuracy and 497.71 ms.

Load-bearing premise

The load-bearing premise is that the words a small model attends to least are safe to delete from the input before a larger, differently tokenized model reads it, so pruning never removes information the larger model needs for a correct answer.

Editorial extensions

If this is right

  • Latency-critical mobile applications such as real-time sentiment analysis could run largely on the user device, offloading only low-confidence cases, so bandwidth use and cloud load drop.
  • Because attention cost grows quadratically with sequence length, pruning even a modest fraction of tokens yields disproportionately large compute savings in the higher-level models.
  • The early-exit rule lets easy examples finish in early transformer layers, lowering average per-request latency without retraining or changing model weights.
  • The weighted ensemble lets weak local results contribute to the final answer, so the system degrades gracefully when the cloud link is slow: the user still receives a local answer, just with less influence from stronger models.

Reading between the lines

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

  • If the attention-pruning transfer rules hold across tokenizers, the same raw-text interface could connect models of different families, such as a BERT-class model at the edge and a decoder-only LLM in the cloud, without retraining either.
  • The paper does not isolate how much of the latency gain comes from pruning versus early exit; rerunning the system with pruning disabled would measure each component's contribution separately.
  • The probabilistic offloading curve implies a network-aware controller could adjust the offloading probability in real time: when transmission is expensive, raising the scaling parameter would keep more high-confidence tasks local, trading a small accuracy loss for lower total latency.
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

3 major / 4 minor

Summary. The paper proposes a three-level cloud-edge-end collaborative inference system for transformer-based discriminative tasks. The system combines confidence-based probabilistic offloading, attention-based input pruning across models with different tokenizers, weighted ensembling of intermediate results, and an early-exit mechanism for individual models. The authors state in the abstract, introduction, and Section IV-B3 that the system reduces inference time by up to 17% without sacrificing inference accuracy. The evaluation uses three BERT-family models on the IMDB sentiment analysis task and reports inference accuracy and time under combinations of confidence thresholds and early-exit thresholds.

Significance. The system concept addresses a real problem: coordinating resource-constrained end devices with edge and cloud models in next-generation networks. The authors build a prototype, incorporate real network communication delays, and provide a parameterized framework with Eqs. (1)-(14). The claimed 17% latency reduction with no accuracy loss would be practically valuable if supported. However, the paper's own experimental results contradict this central claim, and the pruning and tokenizer-transfer mechanisms are not empirically validated. With those issues, the contribution as presented does not meet the bar for publication.

major comments (3)
  1. The central quantitative claim is contradicted by the paper's own data. The recommended configuration in Section IV-B3 is confidence threshold 0.8 with early-exit threshold 0.00001; Table I reports 497.71 ms and accuracy 0.9012 for this row. The corresponding early-exit-disabled baseline (confidence 0.8, tau = 0), which the paper identifies with reproducing Tabi, is 523.93 ms and accuracy 0.9334. This is a 5.0% time reduction and a 3.22-percentage-point accuracy drop, not a 17% reduction with no accuracy loss. The high-accuracy configuration (confidence 0.9, tau = 0.00001) is 502.98 ms versus the baseline 534.7 ms (6.9% faster) but accuracy drops from 0.9685 to 0.9195. No row in Table I achieves both a 17% speedup and maintained accuracy; the largest speedup (tau = 0.01, confidence 0.7) is 22% faster but loses 20.6 accuracy points. The abstract and Section I must be corrected, and the Section IV-B3 recommendation must be reconciled with the numbers in Table I.
  2. The pruning coefficient alpha in Eq. (10) is never reported, and the three cross-tokenizer rules in Section III-B are not evaluated in any experiment. The accuracy-preservation argument for the whole system depends on this mechanism: pruning must remove only tokens that the downstream model does not need. Without reporting alpha, without a sensitivity analysis over alpha, and without testing the tokenizer-transfer rules, the claimed 'without sacrificing inference accuracy' is unverified for the pruning component. If alpha removes tokens the larger model relies on, the accuracy loss would be larger than reported and the offloading benefit would shrink.
  3. The headline performance figure appears to be selected post hoc from the grid of hyperparameters. The offloading probability uses the temperature T, threshold t, and scaling parameter k (Eqs. (6)-(8)); the early exit uses threshold tau and patience counter p (Eqs. (13)-(14)). These hyperparameters are tuned on the IMDB evaluation task, and the 'optimal' configuration is chosen from the grid reported in Table I. The paper does not separate validation from test, does not provide a sensitivity analysis over these hyperparameters, and does not show that the '17%' figure generalizes to other tasks, datasets, or network conditions. This makes the central quantitative claim a fitted outcome rather than a predictive result.
minor comments (4)
  1. The sentence 'As the early exit threshold increases (i.e., as tau decreases), the inference time becomes shorter' is internally contradictory: increasing the threshold would, according to Eq. (14), make exits easier and reduce time, but the stated direction in the paper is the opposite. The intended relationship should be clarified.
  2. The axis labels in Fig. 4 are rendered as garbled characters in the provided version; please ensure the final figure has readable axis labels and a caption that explains the two axes.
  3. The claim that the confidence threshold 0.8 and early-exit threshold 0.00001 'reduces inference time by approximately 17%' is not derivable from Table I, as the baseline is not defined. If the baseline is the early-exit-disabled row with the same confidence threshold, the reduction is 5.0%; if it is the best accuracy row, the reduction is 6.9%. Please state explicitly what the 17% refers to and correct the arithmetic.
  4. The ensemble weights w_i are said to be determined by 'the properties of the model itself during pre-profiling,' but the exact procedure is not described and the weights are not reported. A short statement of how these weights are computed would improve reproducibility.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity; the paper is an empirical systems integration whose components are cited external mechanisms, and the headline 17% claim is contradicted by its own Table I rather than derived from it.

full rationale

The paper does not claim a first-principles derivation; it assembles known mechanisms (Tabi-style offloading [2], temperature-scaled confidence [9], attention-based pruning [11,12], PABEE-style early exit [3], weighted ensembling [6,7]) and reports experiments. The load-bearing equations (6)-(8), (10), (13)-(14) are design heuristics with tunable hyperparameters, not identities that make the reported latency or accuracy true by construction. The only self-citation is reference [31] (a co-author's edge-cloud scheduling work) in a background list, and it is not load-bearing. The abstract and Section IV-B3 claim "up to 17%" time reduction "without sacrificing accuracy," but Table I's recommended configuration (t=0.8, tau=0.00001) shows 497.71 ms and accuracy 0.9012 versus the tau=0 baseline of 523.93 ms and 0.9334, a 5.0% speedup with a 3.22-point accuracy drop; no row simultaneously achieves 17% and preserves accuracy. That is a correctness/empirical-support problem, not circularity: the claim is contradicted by the paper's own data, not equivalent to its inputs by construction. Hyperparameters are acknowledged as tunable, but tuning on the evaluation set is a methodological limitation, not a self-definitional reduction. The un-reported pruning coefficient alpha in Eq. (10) and un-evaluated cross-tokenizer rules in Section III-B are missing-support issues, not circularity.

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

The system's performance rests on several adjustable hyperparameters (T, t, k, alpha, tau, p, w_i) and on domain assumptions about hierarchy structure, confidence calibration, attention-based pruning, ensembling, and simplified communication latency. No new physical or conceptual entities are postulated; all components are algorithmic modules assembled from prior work.

free parameters (7)
  • Temperature scaling parameter T = not reported
    Eq. (6); optimized on the validation set to calibrate softmax confidence.
  • Confidence threshold t = 0.7, 0.8, 0.9
    Eq. (7); grid values tested on the IMDB task and used to select the best configuration.
  • Offloading probability scaling parameter k = not reported
    Eq. (7); shapes the S-curve for probabilistic offloading; no fitted or chosen value given.
  • Pruning coefficient alpha = not reported
    Eq. (10); sets token retention threshold as alpha times average importance; no value or sensitivity analysis reported.
  • Early exit difference threshold tau = 0.01, 0.0001, 0.00001
    Eqs. (13)-(14); grid values used in the experiments; the paper also calls tau the early exit parameter.
  • Early exit patience counter p = not reported
    Eq. (14); number of consecutive low-difference layers needed before exit; never specified.
  • Ensemble weights w_i = not reported
    Eq. (11); weights sum to 1 and are determined by the properties of the model itself during pre-profiling, but exact values are not given.
assumptions (5)
  • domain assumption A three-level end-edge-cloud hierarchy with monotonically increasing model capability covers next-generation mobile network inference scenarios
    Section II states 'the network is ideally divided into a three-level structure' and that models are matched to device capabilities; this underlies the entire deployment model.
  • domain assumption Temperature-scaled softmax confidence, calibrated on a validation set, reliably drives offloading decisions
    Section III-A uses Eq. (6) with a temperature T optimized on the validation set, citing calibration literature; no evidence is given that calibration transfers across the three BERT models used.
  • domain assumption Attention weights summed over layers indicate token importance, and pruning low-importance tokens preserves accuracy
    Section III-B, Eq. (10) prunes tokens below alpha times average importance, relying on analyses of BERT attention; the paper gives no sensitivity analysis for alpha.
  • domain assumption Weighted averaging of softmax outputs from small and large models improves accuracy over the largest model alone
    Section III-C, Eq. (11) combines outputs with weights set by pre-profiled accuracy, citing ensemble literature; no IMDB-specific evidence is presented.
  • domain assumption Communication latency is proportional to payload size divided by a fixed link rate, with no queueing or protocol overhead
    Eq. (2) models L_tra = size(x)/S_j; the evaluation 'incorporates communication delays from a real network' but the exact simulator is not described.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Distributed Collaborative Inference System in Next-Generation Networks and Communication." pith.science (2026). https://pith.science/paper/JFPCJUKC

@misc{pith2026241212102,
  author       = {Pith},
  title        = {Pith review of: Distributed Collaborative Inference System in Next-Generation Networks and Communication},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/JFPCJUKC}},
  note         = {Machine review of arXiv:2412.12102}
}
read the original abstract

With the rapid advancement of artificial intelligence, generative artificial intelligence (GAI) has taken a leading role in transforming data processing methods. However, the high computational demands of GAI present challenges for devices with limited resources. As we move towards the sixth generation of mobile networks (6G), the higher data rates and improved energy efficiency of 6G create a need for more efficient data processing in GAI. Traditional GAI, however, shows its limitations in meeting these demands. To address these challenges, we introduce a multi-level collaborative inference system designed for next-generation networks and communication. Our proposed system features a deployment strategy that assigns models of varying sizes to devices at different network layers. Then, we design a task offloading strategy to optimise both efficiency and latency. Furthermore, a modified early exit mechanism is implemented to enhance the inference process for single models. Experimental results demonstrate that our system effectively reduces inference latency while maintaining high-quality output. Specifically, compared to existing work, our system can reduce inference time by up to 17% without sacrificing the inference accuracy.

Figures

Figures reproduced from arXiv: 2412.12102 by the authors.

Figure 1
Figure 1. Examples of services provided by multi-level [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 3
Figure 3. Accelerating of single model inference. of the task statement to expedite the inference process of the higher-level model by pruning unnecessary words from the input data. This approach helps offset the excess latency overhead caused by multiple inference. To some extent, it also reduces the communication overhead. Specifically, we record the attention values at each level of the inference process and sum them to ga… view at source ↗
Figure 4
Figure 4. Confidence vs. offloading probability curve. [PITH_FULL_IMAGE:figures/full_fig_p005_4.png] view at source ↗
Figures from the paper (5 more)
Figure 5
Figure 5. Figure 5: Simple structure of the transformer encoder. [PITH_FULL_IMAGE:figures/full_fig_p005_5.png]
Figure 6
Figure 6. Figure 6: Examples of attention weights. Darker colours [PITH_FULL_IMAGE:figures/full_fig_p006_6.png]
Figure 7
Figure 7. Figure 7: Optimized early exit. is immediately utilized as the final result. This approach obvi￾ates the need for further computation in the remaining layers, thereby significantly reducing the wastage of computational resources. We refer to previous research on early exit metho…
Figure 8
Figure 8. Figure 8: Comparison of time. 0 0.01 0.0001 0.00001 Early Exit Parameters 0 20 40 60 80 100 Accuracy(%) Threshold 0.9 Threshold 0.8 Threshold 0.7 [PITH_FULL_IMAGE:figures/full_fig_p008_8.png]
Figure 9
Figure 9. Figure 9: Comparison of accuracy. robustness of the early exit mechanism in optimizing compu￾tational efficiency without a substantial loss in accuracy. From [PITH_FULL_IMAGE:figures/full_fig_p008_9.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

32 extracted references · 19 canonical work pages

  1. [1]

    Cocktail: A multidimensional optimization for model serving in cloud,

    J. R. Gunasekaran, C. S. Mishra, P. Thinakaran, B. Sharma, M. T. Kandemir, and C. R. Das, “Cocktail: A multidimensional optimization for model serving in cloud,” in 19th USENIX Symposium on Networked Systems Design and Implementation (NSDI 22) , 2022, pp. 1041–1057

    work page 2022

  2. [2]

    Tabi: An efficient multi- level inference system for large language models,

    Y . Wang, K. Chen, H. Tan, and K. Guo, “Tabi: An efficient multi- level inference system for large language models,” in Proceedings of the Eighteenth European Conference on Computer Systems , 2023, pp. 233–248

    2023

  3. [3]

    Bert loses patience: Fast and robust inference with early exit,

    W. Zhou, C. Xu, T. Ge, J. McAuley, K. Xu, and F. Wei, “Bert loses patience: Fast and robust inference with early exit,” Advances in Neural Information Processing Systems , vol. 33, pp. 18 330–18 341, 2020

    2020

  4. [4]

    Accelerating inference in large lan- guage models with a unified layer skipping strategy,

    Y . Liu, F. Meng, and J. Zhou, “Accelerating inference in large lan- guage models with a unified layer skipping strategy,” arXiv preprint arXiv:2404.06954, 2024

    arXiv 2024

  5. [5]

    Learning layer-skippable inference network,

    Y .-G. Jiang, C. Cheng, H. Lin, and Y . Fu, “Learning layer-skippable inference network,” IEEE Transactions on Image Processing , vol. 29, pp. 8747–8759, 2020

    work page 2020

  6. [6]

    Neural network ensembles, cross validation, and active learning,

    A. Krogh and J. Vedelsby, “Neural network ensembles, cross validation, and active learning,” Advances in neural information processing systems, vol. 7, 1994

    work page 1994

  7. [7]

    An empirical evaluation of bagging and boosting,

    R. Maclin and D. Opitz, “An empirical evaluation of bagging and boosting,” AAAI/IAAI, vol. 1997, pp. 546–551, 1997

    work page 1997

  8. [8]

    Calibration of pre-trained transformers,

    S. Desai and G. Durrett, “Calibration of pre-trained transformers,” arXiv preprint arXiv:2003.07892, 2020

    arXiv 2003

Show all 32 references
  1. [9]

    On calibration of modern neural networks,

    C. Guo, G. Pleiss, Y . Sun, and K. Q. Weinberger, “On calibration of modern neural networks,” in International conference on machine learning. PMLR, 2017, pp. 1321–1330

    2017

  2. [10]

    Branchynet: Fast inference via early exiting from deep neural networks,

    S. Teerapittayanon, B. McDanel, and H.-T. Kung, “Branchynet: Fast inference via early exiting from deep neural networks,” in 2016 23rd international conference on pattern recognition (ICPR) . IEEE, 2016, pp. 2464–2469

    2016

  3. [11]

    What does bert look at? an analysis of bert’s attention,

    K. Clark, U. Khandelwal, O. Levy, and C. D. Manning, “What does bert look at? an analysis of bert’s attention,” arXiv preprint arXiv:1906.04341, 2019

    1906 arXiv

  4. [12]

    Power-bert: Accelerating bert inference via progres- sive word-vector elimination,

    S. Goyal, A. R. Choudhury, S. Raje, V . Chakaravarthy, Y . Sabharwal, and A. Verma, “Power-bert: Accelerating bert inference via progres- sive word-vector elimination,” in International Conference on Machine Learning. PMLR, 2020, pp. 3690–3699

    2020

  5. [13]

    A systematic review of social media- based sentiment analysis: Emerging trends and challenges,

    Q. A. Xu, V . Chang, and C. Jayne, “A systematic review of social media- based sentiment analysis: Emerging trends and challenges,” Decision Analytics Journal, vol. 3, p. 100073, 2022

    2022

  6. [14]

    A systematic survey on explainable ai applied to fake news detection,

    A. Athira, S. M. Kumar, and A. M. Chacko, “A systematic survey on explainable ai applied to fake news detection,” Engineering Applications of Artificial Intelligence , vol. 122, p. 106087, 2023

    2023

  7. [15]

    A new chatbot for customer service on social media,

    A. Xu, Z. Liu, Y . Guo, V . Sinha, and R. Akkiraju, “A new chatbot for customer service on social media,” in Proceedings of the 2017 CHI conference on human factors in computing systems , 2017, pp. 3506– 3510

    2017

  8. [16]

    The roadmap to 6g: Ai empowered wireless networks,

    K. B. Letaief, W. Chen, Y . Shi, J. Zhang, and Y .-J. A. Zhang, “The roadmap to 6g: Ai empowered wireless networks,” IEEE communica- tions magazine, vol. 57, no. 8, pp. 84–90, 2019

    2019

  9. [17]

    Artificial-intelligence-enabled intelligent 6g networks,

    H. Yang, A. Alphones, Z. Xiong, D. Niyato, J. Zhao, and K. Wu, “Artificial-intelligence-enabled intelligent 6g networks,” IEEE network, vol. 34, no. 6, pp. 272–280, 2020

    2020

  10. [18]

    Machine learning in business management using customer behavior analysis using 6g technology,

    A. B. Prasad, K. Abraham, S. Tripathy, N. Goyal et al. , “Machine learning in business management using customer behavior analysis using 6g technology,” Journal of Pharmaceutical Negative Results , pp. 399– 410, 2022

    2022

  11. [19]

    6g comprehensive intelligence: network operations and optimization based on large language models,

    S. Long, F. Tang, Y . Li, T. Tan, Z. Jin, M. Zhao, and N. Kato, “6g comprehensive intelligence: network operations and optimization based on large language models,” arXiv preprint arXiv:2404.18373 , 2024

    2024 arXiv

  12. [20]

    Flexnn: Efficient and adaptive dnn inference on memory-constrained edge devices,

    X. Li, Y . Li, Y . Li, T. Cao, and Y . Liu, “Flexnn: Efficient and adaptive dnn inference on memory-constrained edge devices,” in Proceedings of the 30th Annual International Conference on Mobile Computing and Networking, 2024, pp. 709–723

    2024

  13. [21]

    Cmix-nn: Mixed low-precision cnn library for memory-constrained edge devices,

    A. Capotondi, M. Rusci, M. Fariselli, and L. Benini, “Cmix-nn: Mixed low-precision cnn library for memory-constrained edge devices,” IEEE Transactions on Circuits and Systems II: Express Briefs , vol. 67, no. 5, pp. 871–875, 2020

    2020

  14. [22]

    Model-distributed dnn training for memory-constrained edge computing devices,

    P. Li, H. Seferoglu, V . R. Dasari, and E. Koyuncu, “Model-distributed dnn training for memory-constrained edge computing devices,” in 2021 IEEE International Symposium on Local and Metropolitan Area Net- works (LANMAN). IEEE, 2021, pp. 1–6

    2021

  15. [23]

    Auto-split: A general framework of collaborative edge-cloud ai,

    A. Banitalebi-Dehkordi, N. Vedula, J. Pei, F. Xia, L. Wang, and Y . Zhang, “Auto-split: A general framework of collaborative edge-cloud ai,” in Proceedings of the 27th ACM SIGKDD Conference on Knowledge Discovery & Data Mining , 2021, pp. 2543–2553

    2021

  16. [24]

    Edge-cloud polarization and collaboration: A comprehensive survey for ai,

    J. Yao, S. Zhang, Y . Yao, F. Wang, J. Ma, J. Zhang, Y . Chu, L. Ji, K. Jia, T. Shen et al. , “Edge-cloud polarization and collaboration: A comprehensive survey for ai,” IEEE Transactions on Knowledge and Data Engineering, vol. 35, no. 7, pp. 6866–6886, 2022

    2022

  17. [25]

    Deadline-based dynamic resource allocation and provisioning algorithms in fog-cloud environment,

    R. K. Naha, S. Garg, A. Chan, and S. K. Battula, “Deadline-based dynamic resource allocation and provisioning algorithms in fog-cloud environment,” Future Generation Computer Systems, vol. 104, pp. 131– 141, 2020

    2020

  18. [26]

    Analytics-as-a-service in a multi-cloud environment through semantically-enabled hierarchical data processing,

    P. P. Jayaraman, C. Perera, D. Georgakopoulos, S. Dustdar, D. Thakker, and R. Ranjan, “Analytics-as-a-service in a multi-cloud environment through semantically-enabled hierarchical data processing,” Software: Practice and Experience , vol. 47, no. 8, pp. 1139–1156, 2017

    2017

  19. [27]

    Trends in ai inference energy consumption: Beyond the performance-vs-parameter laws of deep learning,

    R. Desislavov, F. Mart ´ınez-Plumed, and J. Hern ´andez-Orallo, “Trends in ai inference energy consumption: Beyond the performance-vs-parameter laws of deep learning,” Sustainable Computing: Informatics and Sys- tems, vol. 38, p. 100857, 2023

    2023

  20. [28]

    Green ai,

    R. Schwartz, J. Dodge, N. A. Smith, and O. Etzioni, “Green ai,” Communications of the ACM , vol. 63, no. 12, pp. 54–63, 2020

    2020

  21. [29]

    Green ai: Do deep learning frameworks have different costs?

    S. Georgiou, M. Kechagia, T. Sharma, F. Sarro, and Y . Zou, “Green ai: Do deep learning frameworks have different costs?” in Proceedings of the 44th International Conference on Software Engineering , 2022, pp. 1082–1094

    2022

  22. [30]

    Fake news detection using machine learning approaches,

    Z. Khanam, B. Alwasel, H. Sirafi, and M. Rashid, “Fake news detection using machine learning approaches,” in IOP conference series: materials science and engineering , vol. 1099, no. 1. IOP Publishing, 2021, p. 012040

    2021

  23. [31]

    Edgeop- timizer: A programmable containerized scheduler of time-critical tasks in kubernetes-based edge-cloud clusters,

    Y . Qiao, S. Shen, C. Zhang, W. Wang, T. Qiu, and X. Wang, “Edgeop- timizer: A programmable containerized scheduler of time-critical tasks in kubernetes-based edge-cloud clusters,” Future Generation Computer Systems, vol. 156, pp. 221–230, 2024

    2024

  24. [32]

    A socialized learning-based scheduling framework in intricate edge clouds,

    Y . Zhao, Z. Wang, C. Qiu, Q. He, D. Niyato, X. Wang, X. Wang, and Q. Hu, “A socialized learning-based scheduling framework in intricate edge clouds,” IEEE Transactions on Services Computing , 2024

    2024

Pith tools

Reviewed August 12, 2026 · model on record in the stance chip above.