Collaborative Inference for Large Models with Task Offloading and Early Exiting

Hongli Xu; Yang Xu; Yunming Liao; Zhiyuan Yao; Zuan Xie

REVIEW 3 major objections 4 minor 50 references

Collaborative Inference for Large Models with Task Offloading and Early Exiting

T0 review · 3 major / 4 minor · reviewed 2026-08-11 · deepseek-v4-flash

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

Pith's one-line read A distributed algorithm called DTO-EE lets each edge node jointly tune its task-offloading probabilities and early-exit confidence thresholds, cutting average response delay by 21-41% while keeping or slightly improving inference accuracy…

desk verdict A useful joint-optimization algorithm for edge inference, but the convergence proof is omitted exactly where it matters and the accuracy claims are overstated. read the letter →

arxiv 2412.08284 v1 pith:3RKO2WW7 submitted 2024-12-11 cs.DC

Zuan Xie , Yang Xu , Hongli Xu , Yunming Liao , Zhiyuan Yao This is my paper · ORCID

classification cs.DC

keywords collaborativeinferencetaskoffloadingearlyexitmodelpartitioningedgecomputingdistributedoptimizationconfidencethresholdM/D/1queue

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

The paper claims that when a large model is split across edge servers and fitted with early-exit branches, the task-offloading strategy and the confidence thresholds are so tightly coupled that optimizing either alone leaves performance on the table. It proposes DTO-EE, a distributed algorithm in which each server only exchanges local messages with its neighbors, estimates a recursive gradient of response delay with respect to offload probabilities, and periodically adjusts confidence thresholds using a precomputed accuracy-ratio table. The claim is that this joint, distributed optimization reaches a better delay-accuracy trade-off than baselines, reducing average response delay by 21-41% and improving accuracy by 1-4%. If true, it provides a practical way to run very large models at the network edge over heterogeneous, changing resources without a central coordinator.

What carries the argument

The key machinery is an M/D/1 processor-sharing queueing formula for per-node computing delay, paired with a recursively defined repulsive factor that carries the marginal cost of an offload decision across all downstream stages. The repulsive factor is computed locally from a receiver's load, transmission rate, and the successor's own repulsive factor; exchanging it through RUR/RUS messages gives each offloader the gradient information it needs to shift probabilities toward the least-expensive receiver. The early-exit confidence threshold is coupled to the same machinery by treating an exit as a virtual offload target, which turns a threshold change into a proportional rescaling of all successor probabilities.

What would settle it

Run DTO-EE's delay prediction against an event-driven simulation in which per-task service times are deterministic but arrival streams at downstream stages come from real trace data rather than Poisson splitting; if predicted average delay diverges materially from simulated delay, the gradient updates rest on a wrong model. A simpler check is to measure the interarrival distribution at the output of an actual M/D/1-PS queue and test whether it is exponential.

Watch

Extended reading notes

Core claim

The central discovery is an explicit coupled relationship between the task-offloading strategy and the early-exit confidence thresholds. Changing a confidence threshold is equivalent to changing the probability of offloading to a virtual exit node, so it scales the offload probabilities to every successor by the same factor. The authors derive a recursive quantity, the repulsive factor, that propagates the marginal delay impact of each offload probability through the whole pipeline, and show that minimizing the penalized delay objective with fixed thresholds is convex, so each offloader can shift probability mass toward the receiver with the smallest repulsive factor using a gradient-descent-style update. On top of this, the confidence threshold of each branch is nudged up or down based on its measured effect on delay (via the coupling) and accuracy (from a one-shot validation table), accepting a change only when it reduces a combined utility. Simulation results on ResNet101/ImageNet and BERT/Tnews show DTO-EE beating four baselines consistently on both delay and accuracy.

Load-bearing premise

Every server's computation time is modeled as an M/D/1 processor-sharing queue whose output traffic is still Poisson, so the same closed-form delay formula can be applied recursively at every layer of the pipeline.

Editorial extensions

If this is right

Each edge node can converge to a load-balanced offloading policy using only neighbor-to-neighbor messages, so the scheme scales to networks where no node has a global view.
Early exits stop being a separate heuristic: threshold choices can be evaluated by their measured delay effect on the whole pipeline, computed from the same gradient machinery.
The performance gap over baselines grows when the system is more resource-constrained or more heterogeneous, suggesting joint tuning matters most exactly where edge inference is hardest.
The reuse-based accuracy-ratio table means thresholds for many branches can be optimized with a single forward pass over a validation set, avoiding repeated model evaluations.
In dynamic environments DTO-EE remains low-delay and stable where game-theoretic and genetic-algorithm baselines degrade, since its updates are local and fast.

Reading between the lines

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

An implicit consequence of the coupling view is that the same virtual-exit rescaling applies to any early-exit DNN pipeline where exit decisions can be modeled as probabilistic branching, not just edge-server offloading.
A testable extension would replace the M/D/1-PS Poisson assumption with trace-driven or bursty arrivals and check whether the gradient updates still yield the reported delay reductions; the current numbers are computed under a queueing model whose output-Poisson property is known to be fragile.
The accuracy gains are small (1-4%) while the delay gains are large (21-41%), so the practical value is likely a latency guarantee with minimal accuracy cost, rather than accuracy as the headline result.
A natural next step is to coordinate threshold changes across all exit branches simultaneously, rather than cyclically updating one branch at a time, which could find better global trade-offs.

Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Desk Editor's Note

read the letter

Xie et al. propose DTO-EE, a distributed algorithm that jointly optimizes task offloading probabilities and early-exit confidence thresholds for partitioned large models at the edge. The genuinely new piece is the coupling analysis in Eqs. (13)-(17): they derive the gradient of response delay with respect to offloading probability, then reuse that machinery to estimate how a threshold adjustment changes delay, and connect both to a utility that trades off delay against accuracy. That joint treatment is absent from the cited literature, as far as I can tell. The gradient derivation is internally consistent, and the algorithm design is sensible: receivers broadcast load and gradient information, offloaders update probabilities via a repulsive-factor rule, and thresholds are adjusted periodically by a hill-climb on the estimated utility change. The simulation study is reasonably thorough, with two models, two datasets, four baselines, and dynamic environments.

The soft spots, in order. First, the convergence analysis in Section 3.5 is incomplete exactly where it is load-bearing. Lemma 1 shows only that, at a non-fixed point, the update gives a descent direction. The text then asserts that because R is differentiable in the step size τp, there exists τp with R(P^{t+1}) < R(P^t), and concludes monotone decrease and convergence, with the proof omitted. That inference does not go through: existence of a per-point step size does not imply a single fixed step works for all rounds, and the Jacobi-style simultaneous update changes the gradient every round. The threshold update in Algorithm 3 also has no convergence or non-cycling guarantee. Since joint optimization is the central claim, this needs a real proof or an explicit statement that convergence is only empirical.

Second, the abstract claims accuracy improvement of 1%-4%, but the Bert results in Section 4.2 report gains of 1.1‰-5.2‰ (per mille). The ResNet numbers roughly support 1-2%, so the abstract overstates the Bert results by an order of magnitude. That should be corrected.

The reader's weakest assumption about the M/D/1-PS output process not being Poisson does not survive contact with queueing theory: M/D/1-PS is quasi-reversible, so its stationary departure process is Poisson, and the recursive delay formula is consistent. I would not flag that as a flaw.

Finally, no code or data is released, and the experiments are simulations of the analytical model rather than physical deployments. Code release would materially strengthen the paper.

Bottom line: useful method paper for the edge-inference subfield, with a novel joint-optimization angle and honest reporting of its theoretical gap. It deserves a serious referee, but the referee should demand the missing convergence proof, corrected accuracy claims, and code release. I would not cite it in its current form.

Referee Report

3 major / 4 minor

Summary. The paper proposes DTO-EE, a distributed algorithm for collaborative inference of large models partitioned across edge servers, combining task offloading with early-exit confidence-threshold adaptation. The authors model the system as a pipelined network of M/D/1-PS queues, formulate a utility minimization problem that trades off response delay and inference accuracy, and derive a gradient-based offloading update (DTO-R/DTO-O) plus a discrete threshold hill-climb (Algorithm 3). Simulation experiments on ResNet101/ImageNet and BERT/Tnews report average delay reductions of 21%-41% and accuracy improvements over four baselines, and Section 3.5 claims convergence of the offloading strategy to an optimum.

Significance. If the reported results are reproducible, DTO-EE addresses a timely and important problem: deploying large models on heterogeneous edge resources with early-exit inference. The paper has several strengths: the recursive use of M/D/1-PS delay analysis is justified, since M/G/1-PS queues are quasi-reversible and hence have Poisson output; the reuse-based accuracy-ratio table for evaluating confidence thresholds with a one-shot validation pass is pragmatic and novel; and the distributed message-passing design is consistent with the edge-computing setting. The experimental evaluation is broad, including dynamic environments and resource scaling. The main weakness is that the theoretical convergence claims in Sections 3.4 and 3.5 are not supported by the provided proofs, which is load-bearing for the paper's central assertion of joint optimization.

major comments (3)

[Section 3.5] Lemma 1 and the following paragraph establish only that, at a non-fixed point P^t, the update (19) gives a descent direction, i.e., ⟨∇R(P^t), Γ(P^t)-P^t⟩<0. The text then asserts that, since R is differentiable, there exists τ_p with R(P^{t+1})<R(P^t), and concludes that R(P^t) decreases monotonically and converges, adding that the detailed proof is omitted. This inference is not valid as stated: the existence of a step size for each individual P^t does not guarantee that the fixed τ_p used in Algorithm 2 works for all rounds, and the simultaneous Jacobi-style update changes the gradient at every step. Furthermore, convergence of the scalar sequence R(P^t) does not imply convergence of P^t to an optimum of P2. Because the opening of Section 3.5 claims the offloading strategy 'can converge to an optimum,' this is a load-bearing gap. Please supply a rigorous proof (e.g., using Lipschitz smoothness of R with a uniform step size) or explicitly restrict the claim to local descent.
[Sections 3.4 and 3.5] Algorithm 3's confidence-threshold update lacks any convergence or optimality analysis. The decision rule (18) is based on the first-order delay approximation (17) and on accuracy/remaining-ratio estimates obtained from a validation dataset; no argument shows that the cyclic hill-climbing over discrete thresholds converges to a stationary point of the utility U, nor that the alternating updates of P and C converge jointly to a solution of P1. Since the abstract and introduction credit DTO-EE with jointly optimizing the offloading strategy and the confidence threshold, this missing analysis is central to the paper's contribution. Please add a convergence or optimality-gap result for the joint scheme, or clearly state that the threshold adaptation is a heuristic whose optimality is not guaranteed.
[Abstract and Section 4.2] The claimed accuracy improvement of 1%-4% is not supported by the reported experimental numbers. For BERT on Tnews, the improvements over the four baselines are given as 1.1‰-5.2‰ (Section 4.2 and Fig. 6), and the detailed ResNet101 example at 4.8 tasks/s shows gains of 0.6-2.4 percentage points (Fig. 3); the 0.65× resource case reaches at most 2.4 percentage points. No reported experiment reaches a 4 percentage-point gain. Please correct the abstract and conclusion, or specify the subset of experiments that yields the stated 1%-4% range.

minor comments (4)

[Section 3.5] The text contains a typo: 'there exit τp' should read 'there exists τp'; also, 'differentiable on τp' should be 'differentiable with respect to the offloading probabilities' (or 'along the update direction').
[Figure 6(b)] The y-axis label of Fig. 6(b) reads 'Average Delay (ms)' but should read 'Inference accuracy'.
[Table 1 and Eq. (3)] The task arrival rate of an edge node is denoted by φ in Eq. (3) and Eq. (7) but by ϕ elsewhere (e.g., Table 1, Algorithm 1). Please unify the notation.
[Section 3.1] The 'virtual node' used to explain the effect of early exit on offloading probabilities is an intuitive device but is not defined in the notation; please add a formal definition (e.g., in Table 1 or after Eq. (16)).

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the offloading and threshold updates are derived from the paper's own objective and measured validation statistics, not from the target results.

full rationale

The claimed derivation chain is self-contained. The response-delay model is built from a Poisson-arrival/M-D/1-PS assumption cited to an external queueing textbook (ref [35], Haviv), not to the authors' own results; the offloading update (Eq. 19) is a gradient-descent rule whose descent direction is computed from the paper's own objective R(P) via Eqs. (13)-(16). The confidence-threshold update uses Eq. (18) with the delay impact (Eq. 17) derived from the same gradient, and the accuracy/remaining-ratio table is measured from a validation dataset; the reported inference accuracy is then evaluated on the inference task, so no fitted parameter is renamed as a prediction. The same-group citations ([36] and background references [19],[20],[23],[25]-[27]) are used only for message-exchange conventions, time-slot lengths, and examples of heterogeneity; none carries the load-bearing theoretical argument. The Section 3.5 convergence step (existence of a per-round tau_p does not by itself prove monotone decrease under one fixed tau_p) is a rigor gap, not a circular reduction, because it does not assume the conclusion. The M/D/1-PS output-Poisson property is an externally cited modeling assumption; even if debatable, it is not a self-reference. The abstract's '1%-4%' accuracy claim versus the reported 5.2 per-mille Bert gains is an internal reporting inconsistency, not circularity.

Assumptions & free parameters 7 free parameters · 4 assumptions · 1 invented entities

The central claim rests on several domain assumptions (Poisson arrivals, Poisson output of M/D/1-PS queues, steady state) and a collection of hand-chosen hyperparameters (a, K, epsilon, tau_p, tau_c, m, n). No physical entities are invented; the virtual node is a modeling device. The accuracy-ratio table is an empirical input measured from validation data.

free parameters (7)

a = not specified
Weight coefficient in the utility function U(T,A) (Eq. 9); chosen by hand, no value or sensitivity analysis given.
K = not specified
Penalty factor in Eq. 11, described only as 'sufficiently large'; its value affects the solution of the penalized problem P2.
epsilon = arbitrarily small
Small positive constant in the penalty function (Eq. 11); its exact value is not specified.
tau_p = not specified
Step size for updating the offloading strategy (Eq. 19); convergence is claimed for some tau_p in (0,1], but the algorithm uses a fixed value.
tau_c = not specified
Step size for adjusting the confidence threshold in Algorithm 3.
m = not specified
Update frequency for confidence thresholds in Algorithm 3.
n = not specified
Total number of communication rounds in Algorithm 3; affects the quality of the final solution.

assumptions (4)

domain assumption Task arrivals at end devices follow independent Poisson processes (Section 2.3).
Used to justify the splitting property and M/D/1-PS queueing model.
domain assumption The output process of an M/D/1-PS queue is also a Poisson process (Section 2.3, citing [35]).
Load-bearing for modeling every edge server as an independent M/D/1-PS queue; this property is not generally true for M/D/1 queues.
domain assumption The system is in steady state, so the arrival rate at each edge server is balanced by the departure rate (Eq. 3).
Needed for the delay formulas and for the recursive gradient computation.
domain assumption The accuracy-ratio table obtained from a validation dataset is representative of the real task distribution (Section 3.1).
The threshold update relies on this table to estimate accuracy changes and remaining ratios.

invented entities (1)

Virtual node for early exit
purpose: Models early exit as offloading a task to a node that performs no further computation, enabling the coupling analysis between thresholds and offloading strategies (Section 3.1).
This is a conceptual abstraction, not a physical entity. It is used to derive Eq. 17 for the delay change caused by adjusting a confidence threshold.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Collaborative Inference for Large Models with Task Offloading and Early Exiting." pith.science (2026). https://pith.science/paper/3RKO2WW7

@misc{pith2026241208284,
  author       = {Pith},
  title        = {Pith review of: Collaborative Inference for Large Models with Task Offloading and Early Exiting},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/3RKO2WW7}},
  note         = {Machine review of arXiv:2412.08284}
}

read the original abstract

In 5G smart cities, edge computing is employed to provide nearby computing services for end devices, and the large-scale models (e.g., GPT and LLaMA) can be deployed at the network edge to boost the service quality. However, due to the constraints of memory size and computing capacity, it is difficult to run these large-scale models on a single edge node. To meet the resource constraints, a large-scale model can be partitioned into multiple sub-models and deployed across multiple edge nodes. Then tasks are offloaded to the edge nodes for collaborative inference. Additionally, we incorporate the early exit mechanism to further accelerate inference. However, the heterogeneous system and dynamic environment will significantly affect the inference efficiency. To address these challenges, we theoretically analyze the coupled relationship between task offloading strategy and confidence thresholds, and develop a distributed algorithm, termed DTO-EE, based on the coupled relationship and convex optimization. DTO-EE enables each edge node to jointly optimize its offloading strategy and the confidence threshold, so as to achieve a promising trade-off between response delay and inference accuracy. The experimental results show that DTO-EE can reduce the average response delay by 21%-41% and improve the inference accuracy by 1%-4%, compared to the baselines.

Figures

Figures reproduced from arXiv: 2412.08284 by the authors.

**Figure 2.** The overview of DTO-EE. T = 1 Φ X e h j ∈S ( λ h j µ h j − λ h j + X e h−1 i ∈V h j φ h−1 i,j T cm,h−1 i,j ) (8) To trade off the average response delay T and inference accuracy A, we introduce a utility function U(T, A) to evaluate system performance as: U(T, A) = aT − (1 − a) (A − Amin) (Amax − Amin) (9) where a is a weight coefficient used to balance the performance of inference delay and accuracy, and the accura… view at source ↗

**Figure 3.** Inference performance of algorithms given different task arrival [PITH_FULL_IMAGE:figures/full_fig_p007_3.png] view at source ↗

Figures from the paper (4 more)

**Figure 4.** Figure 4: Inference performance of algorithms given different task arrival [PITH_FULL_IMAGE:figures/full_fig_p007_4.png]

**Figure 5.** Figure 5: Inference performance with varying average computing re [PITH_FULL_IMAGE:figures/full_fig_p008_5.png]

**Figure 8.** Figure 8: Inference performance in the dynamic environment for Bert on [PITH_FULL_IMAGE:figures/full_fig_p008_8.png]

**Figure 9.** Figure 9: Effect of dynamic thresholds on ImageNet. [PITH_FULL_IMAGE:figures/full_fig_p009_9.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

50 extracted references · 34 canonical work pages

[1]

Global smart street lighting market report,

“Global smart street lighting market report,” https://www.businesswire.com/news/home/20231219038618/, 2023

work page arXiv 2023
[2]

An implementation of efficient smart street lights with crime and accident monitoring: A review,

C. T. Yong, C. V . Hao, B. Ruskhan, S. K. Y . Lim, T. G. Boon, T. S. Wei, and S. B. I. A. Shah, “An implementation of efficient smart street lights with crime and accident monitoring: A review,” Journal of Survey in Fisheries Sciences, pp. 287–305, 2023

work page 2023
[3]

State-of-the-art in smart streetlight systems: a review,

M. Mahoor, Z. S. Hosseini, A. Khodaei, A. Paaso, and D. Kushner, “State-of-the-art in smart streetlight systems: a review,”IET Smart Cities, vol. 2, no. 1, pp. 24–33, 2020

work page 2020
[4]

Edge computing: Vision and challenges,

W. Shi, J. Cao, Q. Zhang, Y . Li, and L. Xu, “Edge computing: Vision and challenges,” IEEE internet of things journal , vol. 3, no. 5, pp. 637–646, 2016

2016
[5]

Improving language understanding by generative pre-training,

A. Radford, K. Narasimhan, T. Salimans, I. Sutskever et al., “Improving language understanding by generative pre-training,” 2018

2018
[6]

Llama: Open and efficient foundation language models,

H. Touvron, T. Lavril, G. Izacard, X. Martinet, M.-A. Lachaux, T. Lacroix, B. Rozi `ere, N. Goyal, E. Hambro, F. Azhar et al. , “Llama: Open and efficient foundation language models,” arXiv preprint arXiv:2302.13971, 2023

arXiv 2023
[7]

Large language models are zero-shot time series forecasters,

N. Gruver, M. Finzi, S. Qiu, and A. G. Wilson, “Large language models are zero-shot time series forecasters,” Advances in Neural Information Processing Systems, vol. 36, 2024

2024
[8]

Compressing large-scale transformer-based models: A case study on bert,

P. Ganesh, Y . Chen, X. Lou, M. A. Khan, Y . Yang, H. Sajjad, P. Nakov, D. Chen, and M. Winslett, “Compressing large-scale transformer-based models: A case study on bert,” Transactions of the Association for Computational Linguistics, vol. 9, pp. 1061–1080, 2021

work page 2021

Show all 50 references

[9]

A com- prehensive survey on model compression and acceleration,

T. Choudhary, V . Mishra, A. Goswami, and J. Sarangapani, “A com- prehensive survey on model compression and acceleration,” Artificial Intelligence Review, vol. 53, pp. 5113–5155, 2020

2020
[10]

Task partitioning and offloading in dnn-task enabled mobile edge computing networks,

M. Gao, R. Shen, L. Shi, W. Qi, J. Li, and Y . Li, “Task partitioning and offloading in dnn-task enabled mobile edge computing networks,” IEEE Transactions on Mobile Computing, vol. 22, no. 4, pp. 2435–2445, 2023

2023
[11]

Joint dnn partition and resource allocation for task offloading in edge-cloud-assisted iot environ- ments,

W. Fan, L. Gao, Y . Su, F. Wu, and Y . Liu, “Joint dnn partition and resource allocation for task offloading in edge-cloud-assisted iot environ- ments,” IEEE Internet of Things Journal, 2023

2023
[12]

Hastening stream offloading of inference via multi-exit dnns in mobile edge computing,

Z. Liu, J. Song, C. Qiu, X. Wang, X. Chen, Q. He, and H. Sheng, “Hastening stream offloading of inference via multi-exit dnns in mobile edge computing,” IEEE Transactions on Mobile Computing, 2022

2022
[13]

Unleashing the low-precision computation potential of tensor cores on gpus,

G. Li, J. Xue, L. Liu, X. Wang, X. Ma, X. Dong, J. Li, and X. Feng, “Unleashing the low-precision computation potential of tensor cores on gpus,” in 2021 IEEE/ACM International Symposium on Code Generation and Optimization (CGO). IEEE, 2021, pp. 90–102

2021
[14]

Network pruning via transformable architecture search,

X. Dong and Y . Yang, “Network pruning via transformable architecture search,” Advances in Neural Information Processing Systems , vol. 32, 2019

2019
[15]

Distributed dnn inference with fine-grained model partitioning in mobile edge computing networks,

H. Li, X. Li, Q. Fan, Q. He, X. Wang, and V . C. Leung, “Distributed dnn inference with fine-grained model partitioning in mobile edge computing networks,” IEEE Transactions on Mobile Computing, 2024

2024
[16]

Joint multiuser dnn partitioning and computational resource allocation for collaborative edge intelligence,

X. Tang, X. Chen, L. Zeng, S. Yu, and L. Chen, “Joint multiuser dnn partitioning and computational resource allocation for collaborative edge intelligence,” IEEE Internet of Things Journal , vol. 8, no. 12, pp. 9511– 9522, 2020

2020
[17]

Pyramidfl: A fine-grained client selection framework for efficient federated learning,

C. Li, X. Zeng, M. Zhang, and Z. Cao, “Pyramidfl: A fine-grained client selection framework for efficient federated learning,” in Proceedings of the 28th Annual International Conference on Mobile Computing And Networking, 2022, pp. 158–171

2022
[18]

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

2020
[19]

Mergesfl: Split federated learning with feature merging and batch size regulation,

Y . Liao, Y . Xu, H. Xu, L. Wang, Z. Yao, and C. Qiao, “Mergesfl: Split federated learning with feature merging and batch size regulation,” in 2024 IEEE 40th International Conference on Data Engineering (ICDE) . IEEE, 2024, pp. 2054–2067

2024
[20]

Accelerating federated learning with data and model parallelism in edge computing,

Y . Liao, Y . Xu, H. Xu, Z. Yao, L. Wang, and C. Qiao, “Accelerating federated learning with data and model parallelism in edge computing,” IEEE/ACM Transactions on Networking, 2023

2023
[21]

Decentralized federated learning with intermediate results in mobile edge computing,

S. Chen, Y . Xu, H. Xu, Z. Jiang, and C. Qiao, “Decentralized federated learning with intermediate results in mobile edge computing,” IEEE Transactions on Mobile Computing, 2022

2022
[22]

Oort: Efficient federated learning via guided participant selection,

F. Lai, X. Zhu, H. V . Madhyastha, and M. Chowdhury, “Oort: Efficient federated learning via guided participant selection,” in 15th {USENIX} Symposium on Operating Systems Design and Implementation ( {OSDI} 21), 2021, pp. 19–35

2021
[23]

Asynchronous decentralized federated learning for heterogeneous devices,

Y . Liao, Y . Xu, H. Xu, M. Chen, L. Wang, and C. Qiao, “Asynchronous decentralized federated learning for heterogeneous devices,” IEEE/ACM Transactions on Networking, 2024

2024
[24]

A smart collaborative framework for dynamic multi-task offloading in iiot-mec networks,

Z. Ai, W. Zhang, M. Li, P. Li, and L. Shi, “A smart collaborative framework for dynamic multi-task offloading in iiot-mec networks,” Peer-to-Peer Networking and Applications, vol. 16, no. 2, pp. 749–764, 2023

2023
[25]

Adaptive configuration for heterogeneous participants in decentralized federated learning,

Y . Liao, Y . Xu, H. Xu, L. Wang, and C. Qian, “Adaptive configuration for heterogeneous participants in decentralized federated learning,” in IEEE INFOCOM 2023-IEEE Conference on Computer Communications. IEEE, 2023, pp. 1–10

2023
[26]

Parallelsfl: A novel split federated learning framework tackling heterogeneity issues,

Y . Liao, Y . Xu, H. Xu, Z. Yao, L. Huang, and C. Qiao, “Parallelsfl: A novel split federated learning framework tackling heterogeneity issues,” arXiv preprint arXiv:2410.01256, 2024

2024 arXiv
[27]

Decentralized federated learning with adaptive configuration for heterogeneous partici- pants,

Y . Liao, Y . Xu, H. Xu, L. Wang, C. Qian, and C. Qiao, “Decentralized federated learning with adaptive configuration for heterogeneous partici- pants,” IEEE Transactions on Mobile Computing, 2023

2023
[28]

Task allocation methods and optimization techniques in edge computing: A systematic review of the literature,

V . Patsias, P. Amanatidis, D. Karampatzakis, T. Lagkas, K. Micha- lakopoulou, and A. Nikitas, “Task allocation methods and optimization techniques in edge computing: A systematic review of the literature,” Future Internet, vol. 15, no. 8, p. 254, 2023

2023
[29]

Response time optimization for cloudlets in mobile edge computing,

M. Tiwary, D. Puthal, K. S. Sahoo, B. Sahoo, and L. T. Yang, “Response time optimization for cloudlets in mobile edge computing,” Journal of Parallel and Distributed Computing, vol. 119, pp. 81–91, 2018

2018
[30]

Dis- tributed inference acceleration with adaptive dnn partitioning and of- floading,

T. Mohammed, C. Joe-Wong, R. Babbar, and M. D. Francesco, “Dis- tributed inference acceleration with adaptive dnn partitioning and of- floading,” in IEEE INFOCOM 2020 - IEEE Conference on Computer Communications, 2020, pp. 854–863

2020
[31]

Adaptive inference through early-exit networks: Design, challenges and directions,

S. Laskaridis, A. Kouris, and N. D. Lane, “Adaptive inference through early-exit networks: Design, challenges and directions,” in Proceedings of the 5th International Workshop on Embedded and Mobile Deep Learning, 2021, pp. 1–6

2021
[32]

Energy-efficient offloading for mobile edge computing in 5g heterogeneous networks,

K. Zhang, Y . Mao, S. Leng, Q. Zhao, L. Li, X. Peng, L. Pan, S. Maharjan, and Y . Zhang, “Energy-efficient offloading for mobile edge computing in 5g heterogeneous networks,” IEEE access, vol. 4, pp. 5896–5907, 2016

2016
[33]

Efficient multi-user computation offloading for mobile-edge cloud computing,

X. Chen, L. Jiao, W. Li, and X. Fu, “Efficient multi-user computation offloading for mobile-edge cloud computing,” IEEE/ACM transactions on networking, vol. 24, no. 5, pp. 2795–2808, 2015

2015
[34]

Load balancing in processor sharing systems,

E. Altman, U. Ayesta, and B. J. Prabhu, “Load balancing in processor sharing systems,” Telecommunication Systems, vol. 47, pp. 35–48, 2011

2011
[35]

A course in queueing theory,

M. Haviv, “A course in queueing theory,”Springer, New York, NY, vol. 10, pp. 978–1, 2013

2013
[36]

Fully distributed task offloading in vehicular edge computing,

Q. Ma, H. Xu, H. Wang, Y . Xu, Q. Jia, and C. Qiao, “Fully distributed task offloading in vehicular edge computing,” IEEE Transactions on Vehicular Technology, 2023

2023
[37]

Game theoretical task offloading for profit maximization in mobile edge computing,

H. Teng, Z. Li, K. Cao, S. Long, S. Guo, and A. Liu, “Game theoretical task offloading for profit maximization in mobile edge computing,”IEEE Transactions on Mobile Computing, vol. 22, no. 9, pp. 5313–5329, 2022

2022
[38]

Imagenet large scale visual recognition challenge,

O. Russakovsky, J. Deng, H. Su, J. Krause, S. Satheesh, S. Ma, Z. Huang, A. Karpathy, A. Khosla, M. Bernstein et al., “Imagenet large scale visual recognition challenge,” International journal of computer vision , vol. 115, pp. 211–252, 2015

2015
[39]

Deep residual learning for image recognition,

K. He, X. Zhang, S. Ren, and J. Sun, “Deep residual learning for image recognition,” in Proceedings of the IEEE conference on computer vision and pattern recognition, 2016, pp. 770–778

2016
[40]

Clue: A chinese language understanding evaluation benchmark,

L. Xu, H. Hu, X. Zhang, L. Li, C. Cao, Y . Li, Y . Xu, K. Sun, D. Yu, C. Yu et al., “Clue: A chinese language understanding evaluation benchmark,” arXiv preprint arXiv:2004.05986, 2020

2004 arXiv
[41]

Bert: Pre-training of deep bidirectional transformers for language understanding,

J. Devlin, M.-W. Chang, K. Lee, and K. Toutanova, “Bert: Pre-training of deep bidirectional transformers for language understanding,” arXiv preprint arXiv:1810.04805, 2018

2018 arXiv
[42]

Collaborative satellite computing through adaptive dnn task splitting and offloading,

S. Peng, X. Hou, Z. Shen, Q. Zheng, J. Jin, A. Tagami, and J. Yuan, “Collaborative satellite computing through adaptive dnn task splitting and offloading,” arXiv preprint arXiv:2405.03181, 2024

2024 arXiv
[43]

Task offloading for mobile edge computing in software defined ultra-dense network,

M. Chen and Y . Hao, “Task offloading for mobile edge computing in software defined ultra-dense network,” IEEE Journal on Selected Areas in Communications, vol. 36, no. 3, pp. 587–597, 2018

2018
[44]

Deep reinforcement learning for task offloading in mobile edge computing systems,

M. Tang and V . W. Wong, “Deep reinforcement learning for task offloading in mobile edge computing systems,” IEEE Transactions on Mobile Computing, vol. 21, no. 6, pp. 1985–1997, 2022

1985
[45]

Distributed optimization for energy-efficient fog computing in the tactile internet,

Y . Xiao and M. Krunz, “Distributed optimization for energy-efficient fog computing in the tactile internet,” IEEE Journal on Selected Areas in Communications, vol. 36, no. 11, pp. 2390–2400, 2018

2018
[46]

Edge intelligence: A computational task offloading scheme for dependent iot application,

H. Xiao, C. Xu, Y . Ma, S. Yang, L. Zhong, and G.-M. Muntean, “Edge intelligence: A computational task offloading scheme for dependent iot application,” IEEE Transactions on Wireless Communications , vol. 21, no. 9, pp. 7222–7237, 2022

2022
[47]

Cnn partitioning and offloading for vehicular edge networks in web3,

X. Xu, S. Tang, L. Qi, X. Zhou, F. Dai, and W. Dou, “Cnn partitioning and offloading for vehicular edge networks in web3,” IEEE Communications Magazine, vol. 61, no. 8, pp. 36–42, 2023. 11

2023
[48]

Edge ai: On-demand accelerating deep neural network inference via edge computing,

E. Li, L. Zeng, Z. Zhou, and X. Chen, “Edge ai: On-demand accelerating deep neural network inference via edge computing,” IEEE Transactions on Wireless Communications, vol. 19, no. 1, pp. 447–457, 2020

2020
[49]

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
[50]

edeepsave: Saving dnn inference using early exit during handovers in mobile edge environment,

W. Ju, D. Yuan, W. Bao, L. Ge, and B. B. Zhou, “edeepsave: Saving dnn inference using early exit during handovers in mobile edge environment,” ACM Transactions on Sensor Networks (TOSN), vol. 17, no. 3, pp. 1–28, 2021

2021

Pith tools

Reviewed August 11, 2026 · model on record in the stance chip above.

[1] [1]

Global smart street lighting market report,

“Global smart street lighting market report,” https://www.businesswire.com/news/home/20231219038618/, 2023

work page arXiv 2023

[2] [2]

An implementation of efficient smart street lights with crime and accident monitoring: A review,

C. T. Yong, C. V . Hao, B. Ruskhan, S. K. Y . Lim, T. G. Boon, T. S. Wei, and S. B. I. A. Shah, “An implementation of efficient smart street lights with crime and accident monitoring: A review,” Journal of Survey in Fisheries Sciences, pp. 287–305, 2023

work page 2023

[3] [3]

State-of-the-art in smart streetlight systems: a review,

M. Mahoor, Z. S. Hosseini, A. Khodaei, A. Paaso, and D. Kushner, “State-of-the-art in smart streetlight systems: a review,”IET Smart Cities, vol. 2, no. 1, pp. 24–33, 2020

work page 2020

[4] [4]

Edge computing: Vision and challenges,

W. Shi, J. Cao, Q. Zhang, Y . Li, and L. Xu, “Edge computing: Vision and challenges,” IEEE internet of things journal , vol. 3, no. 5, pp. 637–646, 2016

2016

[5] [5]

Improving language understanding by generative pre-training,

A. Radford, K. Narasimhan, T. Salimans, I. Sutskever et al., “Improving language understanding by generative pre-training,” 2018

2018

[6] [6]

Llama: Open and efficient foundation language models,

H. Touvron, T. Lavril, G. Izacard, X. Martinet, M.-A. Lachaux, T. Lacroix, B. Rozi `ere, N. Goyal, E. Hambro, F. Azhar et al. , “Llama: Open and efficient foundation language models,” arXiv preprint arXiv:2302.13971, 2023

arXiv 2023

[7] [7]

Large language models are zero-shot time series forecasters,

N. Gruver, M. Finzi, S. Qiu, and A. G. Wilson, “Large language models are zero-shot time series forecasters,” Advances in Neural Information Processing Systems, vol. 36, 2024

2024

[8] [8]

Compressing large-scale transformer-based models: A case study on bert,

P. Ganesh, Y . Chen, X. Lou, M. A. Khan, Y . Yang, H. Sajjad, P. Nakov, D. Chen, and M. Winslett, “Compressing large-scale transformer-based models: A case study on bert,” Transactions of the Association for Computational Linguistics, vol. 9, pp. 1061–1080, 2021

work page 2021