Pith. sign in

REVIEW 4 major objections 7 minor 68 references

AdaScale: Dynamic Context-aware DNN Scaling via Automated Adaptation Loop on Mobile Devices

T0 review · 4 major / 7 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 AdaScale claims an automated on-device loop that scales deep models to fluctuating resources, with 1.51-6.2x faster inference, 4.69x lower energy, and 66.89% less training overhead.

desk verdict Interesting system, but headline claims are contradicted by the paper's own tables. read the letter →

arxiv 2412.00724 v1 pith:HHISU5O5 submitted 2024-12-01 cs.AI

Yuzhan Wang , Sicong Liu , Bin Guo , Boqi Zhang , Ke Ma , Yasan Ding , Hao Luo , Yao Li
show 1 more author
Zhiwen Yu
This is my paper · ORCID
classification cs.AI
keywords deepmodelscalingelasticinferenceresourceefficiencyautomatedadaptationloopmulti-branchearlyexitruntimeperformanceprofilingmobilelearningdynamiccontext
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

AdaScale sets out to make deep neural network deployment on mobile devices self-managing. Instead of re-designing or re-compressing a model when a phone's CPU, GPU, or memory load changes, AdaScale trains one elastic network once and lets a runtime loop switch among its pre-trained branches to match current resources. The paper argues this loop is feasible because a compact combination of lightweight compression operators shrinks the search space, multi-stage shared-parameter training keeps training costs down, and device-sensing plus latency and energy predictors make the choice fast. If true, the payoff is concrete: reported gains of 5.09% higher accuracy, 66.89% lower training overhead, 1.51-6.2x faster inference, 4.69x lower energy, and under 4% accuracy loss in resource-deficient contexts. The result matters because mobile inference must keep working as device state fluctuates, and manually adapting models for every context does not scale.

What carries the argument

The central mechanism is the multi-branch self-evolutionary network, an elastic network with early-exit branches inserted after layers that show strong performance during training, where each branch is built from an ensemble of lightweight compression operators. Branches share parameters through a multi-stage training algorithm: the first segment is trained to an accuracy threshold, the next segment reuses its weights, and updates either freeze earlier weights or refine them only when accuracy improves. Around this network, AdaScale closes an adaptation loop with three components: a resource availability block that computes a load index from CPU, GPU, and memory utilization; a model performance profiler that estimates energy from memory operations, power, clock frequencies, and an approximated cache hit rate, and estimates latency from computational load and device state; and a performance-guided search that minimizes a weighted sum of latency and energy subject to constraints, using B+-trees over intrinsic and predictive performance tables. The loop lets the network select a variant in about 0.043 milliseconds and switch branches on the device as context shifts.

What would settle it

Run a four-branch AdaScale network on an embedded board while varying CPU, GPU, and memory load, and compare each branch's predicted energy and latency against watt-meter and wall-clock measurements; if the predicted ranking of branches diverges from the measured ranking under realistic loads, the adaptation loop can select a worse variant than the one the paper claims.

Watch

Extended reading notes

Core claim

AdaScale's central claim is that dynamic, context-aware scaling of a deployed deep model can be automated on the device. The framework builds a self-evolutionary multi-branch network whose branches combine lightweight compression operators such as depthwise separable convolutions, dense connections, decomposed convolutions, and SVD-based weight reduction, and trains these branches with a multi-stage parameter-sharing procedure so each exit classifier is accurate on its own. At runtime, a resource availability block monitors CPU, GPU, and memory to form a device load index, a model performance profiler estimates latency and energy for candidate variants, and a performance-guided search over a B+-tree-indexed table selects the variant that meets accuracy, latency, and energy constraints. On image classification benchmarks on mobile and embedded platforms, the paper reports that this loop beats prior adaptive and lightweight networks: accuracy up by 5.09%, training overhead down by 66.89%, inference 1.51-6.2x faster, energy 4.69x lower, and accuracy loss kept below 4% even under resource-deficient and dynamically changing conditions. The intended contribution is to move model compression from a costly offline design step into a real-time, closed-loop runtime service.

Load-bearing premise

The load-bearing premise is that the runtime energy and latency predictors reflect what actually happens on the device; if those predictions are miscalibrated for a given context, the search may choose the wrong variant and the reported speedups and savings do not follow.

Editorial extensions

If this is right

  • A single elastic network can serve a range of mobile and embedded devices without per-device recompression, since variant selection happens at runtime rather than during development.
  • The runtime search overhead of roughly 0.043 ms is small relative to typical inference latencies, so adaptation decisions do not themselves stall an app.
  • Multi-stage shared-parameter training keeps the cost of producing a scalable family of models near 4.6 GPU-hours with a search space of 0.48×10^3 configurations, far below the spaces reported for baselines like AdaptiveNet and LegoDNN.
  • Adding early-exit branches costs only about 1% storage overhead as the network grows, while cutting measured inference latency by roughly 60% in simulated user and random load scenarios.

Reading between the lines

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

  • Because the adaptive loop only needs a set of pre-trained branches and a fast table lookup, a natural next step is to extend the same idea to other model families, such as on-device transformers, by rebuilding the operator ensemble and performance tables for those operators.
  • The energy model is calibrated with coefficients borrowed from prior work; an extension the paper does not implement is online recalibration of those coefficients from direct watt-meter readings, which would make the loop self-tuning on unseen hardware.
  • The paper's branch-placement rule, insert branches after layers with strong accuracy gains, suggests a testable design principle for other elastic networks: a cheap accuracy-gain scan during pretraining can decide where to put exits, avoiding exhaustive search over all possible branch locations.
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

4 major / 7 minor

Summary. AdaScale proposes an elastic inference framework for mobile devices that combines a multi-branch self-evolutionary network, an ensemble of lightweight compression operators, a runtime resource-awareness module, and a performance-guided search over a B+ tree. The paper claims that AdaScale improves accuracy by 5.09%, reduces training overhead by 66.89%, speeds up inference by 1.51-6.2x, and lowers energy cost by 4.69x while keeping accuracy loss below 4%, and it reports experiments on CIFAR-10, CIFAR-100, and two embedded platforms (Jetson Nx and Raspberry Pi 4B) against adaptive baselines (LegoDNN, AdaptiveNet, AdaDeep, AdaSpring, NAS) and lightweight single networks (MobileNet, ShuffleNet, SqueezeNet, ResNet18, Transformer).

Significance. If the claimed dynamic adaptation loop worked as described, the paper would make a useful contribution to mobile DNN deployment: the idea of combining a lightweight multi-branch backbone with low-overhead runtime search is timely, and the B+ tree search (0.043 ms overhead) plus multi-stage training are sensible engineering choices. The paper is also honest in aiming at an end-to-end system rather than a single benchmark number. However, the evaluation as written does not support the central 'AdaScale outperforms existing methods' assertion: the headline numbers are not traceable to the tables, and one headline claim (4.69x lower energy) is directly contradicted by the reported measurements. No code or machine-checked artifacts are provided, and key parameters of the adaptation loop are left unspecified, so the reproducibility of the central mechanism is not established.

major comments (4)
  1. [Abstract; §V-B, Tables II-III] The headline performance metrics in the abstract are not mapped to any row of the reported tables. The claimed 66.89% training-overhead reduction appears nowhere: Table II reports 4.575 GPU-hours for AdaScale versus 8.6 for LegoDNN, which is a 46.8% reduction, and versus 13 for AdaptiveNet, which is a 64.8% reduction. The claimed 5.09% accuracy improvement is likewise unidentifiable: on CIFAR-10 the best AdaScale stage reaches 91.17% versus 90.01% for Transformer, a 1.16 percentage-point difference, and on CIFAR-100 AdaScale (72.12%) is below ResNet18 (75.31%). These discrepancies are load-bearing because the central 'outperforms' assertion rests on these exact numbers.
  2. [§IV-B, Eq. (3)-(5); Tables III-IV] The energy-saving claim is contradicted by the paper's own tables. Tables III and IV list AdaScale per-inference energy of 1285.17-2758.59 mJ on Jetson Nx, while MobileNetV3-small is listed at 120.61 mJ and MobileNetV3-large at 412.08 mJ. AdaScale is therefore 10-20x more energy-intensive than those baselines, not 4.69x lower energy as claimed in the abstract, and no baseline in either table yields a 4.69x ratio. In addition, the energy predictor in Eq. (3) depends on coefficients Pcpu, Pgpu, Pmem and on the cache estimate of Eq. (5), which the text states are taken from the authors' AdaEnlight paper; no calibration or validation against directly measured energy is reported, so the runtime selection objective in Eq. (6) cannot be audited.
  3. [§V-D; Abstract] The latency speedup range in the abstract (1.51-6.2x) does not match the ranges reported in §V-D: 1.14-1.85x on Jetson Nx and 1.71-6.29x on Raspberry Pi 4B, and no single baseline is named for either range. The largest value, 6.29x, is attributed to the Raspberry Pi comparison with ShuffleNet, but no per-method speedup table or plot is provided. From Table III alone, AdaScale Stage1 at 9.10 ms versus ResNet18 at 16.60 ms gives only a 1.82x speedup on Jetson Nx, so the claimed 'faster inference' headline is not reproducible from the data presented.
  4. [§IV-A, §IV-C, Algorithm 1] Core parameters of the adaptation loop are unspecified. The load-index weights Wcpu, Wgpu, and WM in Eq. (2), the search-loss weights alpha and beta in Eq. (6), and the per-branch accuracy thresholds Acc_i in Algorithm 1 are never assigned values, and no sensitivity or ablation study is reported. Without these values or a measurement protocol, another group cannot reproduce the automated adaptation loop, and it is unclear whether the reported latency/accuracy behavior is robust or tuned to the particular devices and workloads used.
minor comments (7)
  1. [§II-A] The bullet list contains a duplicated 'Dynamic adaptivity' entry, and the same paragraph appears twice in the section; one copy should be removed.
  2. [§I and §II-A] The text says related work is reviewed in §IV, but the related-work section is actually §VI; the internal section references should be corrected throughout.
  3. [Table II] The row for AdaDeep cites reference [10], which is the AdaSpring reference; it should presumably cite [40].
  4. [§V-F] The summary paragraph refers to the proposed framework as 'AdaBranch' rather than 'AdaScale'.
  5. [Tables III-IV] The energy values for MobileNetV3-large (412.08 mJ) and MobileNetV3-small (120.61 mJ) differ from the other baselines by an order of magnitude, and ShuffleNet/SqueezeNet report identical energy on CIFAR-10 and CIFAR-100; the measurement procedure and units should be rechecked and documented.
  6. [Table I, §V] Tiny ImageNet is listed as dataset D3 in Table I, but no experimental result for Tiny ImageNet appears anywhere in §V; if it was used, the results should be reported, otherwise it should be removed from the dataset list.
  7. [§III-A2] The 'branch diversity establishment' discussion states that deeper layers perform better and some layers are less critical, but it reports no quantitative measurements supporting these observations; since branch placement is a design decision, the evidence base should be shown.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: the reported accuracy, latency, training, and energy results are benchmark comparisons against external baselines, not outputs of the paper's own predictor equations; the sole same-author citation supplies profiler coefficients rather than the claimed results.

full rationale

The paper's central claims are supported by cross-device comparisons against external lightweight and adaptive baselines (Tables II–IV, Figures 5–9), not by the runtime predictor equations. The energy and latency profiler in Section IV-B is used only to guide the runtime search in Eq. (6)–(7); the evaluation tables report measured energy in millijoules and latency in milliseconds on Jetson Nx, Raspberry Pi, and Turtlebot, so no headline number is generated by the profiler itself. The only same-group citation that feeds the design, AdaEnlight [45], supplies empirical coefficients for the energy model in Eq. (3); this is a design dependency, not a circular reduction, because those coefficients were not fitted to the target results of this paper and the reported benchmark numbers do not reduce to Eq. (3). No equation defines X in terms of Y and then presents Y as an independent derivation: Eq. (1) is a standard constrained optimization, Eq. (6)–(7) select among pre-stored variant tables, and Algorithm 1 is a sequential multi-stage training procedure. Inconsistencies such as the 4.69x energy claim not being reproducible from Table III are correctness/verifiability concerns, not circularity. The derivation chain is therefore self-contained with respect to circularity, even though some reported aggregate numbers may not follow from the paper's own tables.

Assumptions & free parameters 4 free parameters · 5 assumptions · 0 invented entities

The central claims rest mainly on measured performance, but the adaptation loop depends on unvalidated predictors and unspecified weights: load index weights, search-loss weights, per-branch thresholds, and energy coefficients are not reported. No new physical or mathematical entity is introduced.

free parameters (4)
  • load index weights Wcpu, Wgpu, WM = not reported
    Eq (2) defines device load from CPU, GPU, and memory utilizations with these weights; no values or fitting procedure are given, yet the load index triggers runtime adaptation.
  • search loss weights alpha, beta = not reported
    Eq (6) defines the objective J = alpha*Latency + beta*Energy; the weights balance latency and energy but are never specified, so the claimed optimal model selection is not reproducible.
  • per-branch accuracy thresholds Acc_i = not reported
    Algorithm 1 trains each segment until Acc_i is met; thresholds are user-defined in the pseudocode but no values are given, controlling where training stops and how branches are formed.
  • energy coefficients Pcpu, Pgpu, Pmem and cache parameters = not reported
    Eq (3) requires power consumption rates and clock frequencies; the text says these draw on empirical data from AdaEnlight [45], but values are not provided, so the energy predictor cannot be independently evaluated.
assumptions (5)
  • domain assumption Device load can be summarized by a weighted combination of CPU, GPU, and memory utilization (Eq (2)).
    Used in Section IV-A to trigger adaptation; assumes these three utilizations capture resource availability and that the weights are stable across devices.
  • domain assumption Energy consumption of a DNN is linearly separable into CPU, GPU, and memory operation costs scaled by clock frequency and cache hit rate (Eq (3)).
    Used in Sections IV-B and IV-C to score model variants; no calibration data are shown within the paper.
  • ad hoc to paper Cache hit rate can be estimated from access times by delta_cache = 1 - (t0 - tavg)/t0 (Eq (5)).
    This estimator is introduced for this paper; no experimental validation is provided that access-time ratios track actual cache hit rates.
  • ad hoc to paper Branch placement observations, such as deeper layers performing better and some layers being less critical, generalize across tasks and architectures.
    Section III-A2 uses these observations to decide where to place branches and compression operators; they are not proven or evaluated separately.
  • domain assumption Networks of similar scale can reach comparable accuracy, motivating staged training with weight reuse.
    Cites Wei et al. [44] in Section III-B; the paper relies on this to justify training branches separately and sharing weights.

how reviews work

0 comments
Cite this review

Pith. "Pith review of AdaScale: Dynamic Context-aware DNN Scaling via Automated Adaptation Loop on Mobile Devices." pith.science (2026). https://pith.science/paper/HHISU5O5

@misc{pith2026241200724,
  author       = {Pith},
  title        = {Pith review of: AdaScale: Dynamic Context-aware DNN Scaling via Automated Adaptation Loop on Mobile Devices},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/HHISU5O5}},
  note         = {Machine review of arXiv:2412.00724}
}
read the original abstract

Deep learning is reshaping mobile applications, with a growing trend of deploying deep neural networks (DNNs) directly to mobile and embedded devices to address real-time performance and privacy. To accommodate local resource limitations, techniques like weight compression, convolution decomposition, and specialized layer architectures have been developed. However, the \textit{dynamic} and \textit{diverse} deployment contexts of mobile devices pose significant challenges. Adapting deep models to meet varied device-specific requirements for latency, accuracy, memory, and energy is labor-intensive. Additionally, changing processor states, fluctuating memory availability, and competing processes frequently necessitate model re-compression to preserve user experience. To address these issues, we introduce AdaScale, an elastic inference framework that automates the adaptation of deep models to dynamic contexts. AdaScale leverages a self-evolutionary model to streamline network creation, employs diverse compression operator combinations to reduce the search space and improve outcomes, and integrates a resource availability awareness block and performance profilers to establish an automated adaptation loop. Our experiments demonstrate that AdaScale significantly enhances accuracy by 5.09%, reduces training overhead by 66.89%, speeds up inference latency by 1.51 to 6.2 times, and lowers energy costs by 4.69 times.

Figures

Figures reproduced from arXiv: 2412.00724 by the authors.

Figure 1
Figure 1. Comparison of different model adaptive deployment approaches. Left: pre-deployment on-server model generation. [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. The framework of AdaScale includes two main components: the pretraining of a multi-branch self-evolutionary network [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. Illustration of multi-branch self-evolutionary network. [PITH_FULL_IMAGE:figures/full_fig_p006_3.png] view at source ↗
Figures from the paper (6 more)
Figure 4
Figure 4. Figure 4: Left: a workflow of calculating memory access and cache rate. Right: pseudo codes in Python. [PITH_FULL_IMAGE:figures/full_fig_p008_4.png]
Figure 5
Figure 5. Figure 5: AdaScale improves DNN efficiency by optimizing search space, training time, and search overhead. [PITH_FULL_IMAGE:figures/full_fig_p010_5.png]
Figure 6
Figure 6. Figure 6: Inference latency of different DNNs under dynamic context on mobile devices [PITH_FULL_IMAGE:figures/full_fig_p012_6.png]
Figure 7
Figure 7. Figure 7: (a),(b) Average inference latency under dynamic context on mobile devices. (c) Storage overhead with different branches. [PITH_FULL_IMAGE:figures/full_fig_p012_7.png]
Figure 8
Figure 8. Figure 8: AdaScale evaluates device resources and inference latency under different load scenarios. (a) and (b) show resource [PITH_FULL_IMAGE:figures/full_fig_p013_8.png]
Figure 9
Figure 9. Figure 9: Inference latency of different DNNs on mobile devices [PITH_FULL_IMAGE:figures/full_fig_p013_9.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

68 extracted references · 39 canonical work pages

  1. [1]

    Autonomous driving assistance with dynamic objects using traffic surveillance cameras,

    K. Cho and D. Cho, “Autonomous driving assistance with dynamic objects using traffic surveillance cameras,” Applied Sciences , vol. 12, no. 12, p. 6247, 2022

    work page 2022

  2. [2]

    Roadside infrastructure support for urban automated driving,

    M. Pechinger, G. Schr ¨oer, K. Bogenberger, and C. Markgraf, “Roadside infrastructure support for urban automated driving,” IEEE Transactions on Intelligent Transportation Systems , 2023

    work page 2023

  3. [3]

    Enabling resource-efficient aiot system with cross-level optimization: A survey,

    S. Liu, B. Guo, C. Fang, Z. Wang, S. Luo, Z. Zhou, and Z. Yu, “Enabling resource-efficient aiot system with cross-level optimization: A survey,” IEEE Communications Surveys & Tutorials , 2023

    work page 2023

  4. [4]

    Surgical and medical applications of drones: A comprehensive review,

    J. C. Rosser Jr, V . Vignesh, B. A. Terwilliger, and B. C. Parker, “Surgical and medical applications of drones: A comprehensive review,” JSLS: Journal of the Society of Laparoendoscopic Surgeons , vol. 22, no. 3, 2018

    work page 2018

  5. [5]

    A survey on behav- ioral biometric authentication on smartphones,

    A. Mahfouz, T. M. Mahmoud, and A. S. Eldin, “A survey on behav- ioral biometric authentication on smartphones,” Journal of information security and applications , vol. 37, pp. 28–37, 2017

    work page 2017

  6. [6]

    Moodexplorer: Towards com- pound emotion detection via smartphone sensing,

    X. Zhang, W. Li, X. Chen, and S. Lu, “Moodexplorer: Towards com- pound emotion detection via smartphone sensing,” Proceedings of the ACM on Interactive, Mobile, Wearable and Ubiquitous Technologies , vol. 1, no. 4, pp. 1–30, 2018

    work page 2018

  7. [7]

    Attrleaks on the edge: Exploiting information leakage from privacy-preserving co- inference,

    Z. Wang, K. Liu, J. Hu, J. Ren, H. Guo, and W. Yuan, “Attrleaks on the edge: Exploiting information leakage from privacy-preserving co- inference,” Chinese Journal of Electronics , vol. 32, no. 1, pp. 1–12, 2023

    work page 2023

  8. [8]

    nnperf: Demystifying dnn runtime inference latency on mobile platforms,

    H. Chu, X. Zheng, L. Liu, and H. Ma, “nnperf: Demystifying dnn runtime inference latency on mobile platforms,” in Proceedings of the 21st ACM Conference on Embedded Networked Sensor Systems , 2023, pp. 125–137

    work page 2023

Show all 68 references
  1. [9]

    Deep learning on mobile and embedded devices: State-of-the-art, challenges, and future directions,

    Y . Chen, B. Zheng, Z. Zhang, Q. Wang, C. Shen, and Q. Zhang, “Deep learning on mobile and embedded devices: State-of-the-art, challenges, and future directions,” ACM Computing Surveys (CSUR), vol. 53, no. 4, pp. 1–37, 2020

    2020

  2. [10]

    Adaspring: Context- adaptive and runtime-evolutionary deep model compression for mobile applications,

    S. Liu, B. Guo, K. Ma, Z. Yu, and J. Du, “Adaspring: Context- adaptive and runtime-evolutionary deep model compression for mobile applications,” Proceedings of the ACM on Interactive, Mobile, Wearable and Ubiquitous Technologies, vol. 5, no. 1, pp. 1–22, 2021

    2021

  3. [11]

    Adaptive weight compression for memory-efficient neural networks,

    J. H. Ko, D. Kim, T. Na, J. Kung, and S. Mukhopadhyay, “Adaptive weight compression for memory-efficient neural networks,” in Design, Automation & Test in Europe Conference & Exhibition (DATE), 2017 . IEEE, 2017, pp. 199–204

    2017

  4. [12]

    8-bit optimizers via block-wise quantization,

    T. Dettmers, M. Lewis, S. Shleifer, and L. Zettlemoyer, “8-bit optimizers via block-wise quantization,” arXiv preprint arXiv:2110.02861 , 2021

    2021 arXiv

  5. [13]

    Mixconv: Mixed depthwise convolutional kernels,

    M. Tan and Q. V . Le, “Mixconv: Mixed depthwise convolutional kernels,” arXiv preprint arXiv:1907.09595 , 2019

    1907 arXiv

  6. [14]

    Adashadow: Responsive test-time model adaptation in non-stationary mobile environments,

    C. Fang, S. Liu, Z. Zhou, B. Guo, J. Tang, K. Ma, and Z. Yu, “Adashadow: Responsive test-time model adaptation in non-stationary mobile environments,” arXiv preprint arXiv:2410.08256 , 2024

    2024 arXiv

  7. [15]

    Inceptionnext: When inception meets convnext,

    W. Yu, P. Zhou, S. Yan, and X. Wang, “Inceptionnext: When inception meets convnext,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , 2024, pp. 5672–5683

    2024

  8. [16]

    Dynamic convolution: Attention over convolution kernels,

    Y . Chen, X. Dai, M. Liu, D. Chen, L. Yuan, and Z. Liu, “Dynamic convolution: Attention over convolution kernels,” in Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, 2020, pp. 11 030–11 039

    2020

  9. [17]

    Deep guided attention network for joint denoising and demosaicing in real image,

    T. Zhang, Y . Fu, J. Zhang, and C. Yan, “Deep guided attention network for joint denoising and demosaicing in real image,” Chinese Journal of Electronics, vol. 33, no. 1, pp. 303–312, 2024

    2024

  10. [18]

    Adaknife: Flexible dnn offloading for inference acceleration on heterogeneous mobile devices,

    S. Liu, H. Luo, X. Li, Y . Li, B. Guo, Z. Yu, Y . Wang, K. Ma, Y . Ding, and Y . Yao, “Adaknife: Flexible dnn offloading for inference acceleration on heterogeneous mobile devices,” IEEE Transactions on Mobile Computing, 2024

    2024

  11. [19]

    Context-aware adaptive surgery: A fast and effective framework for adaptative model partition,

    H. Wang, B. Guo, J. Liu, S. Liu, Y . Wu, and Z. Yu, “Context-aware adaptive surgery: A fast and effective framework for adaptative model partition,” Proceedings of the ACM on Interactive, Mobile, Wearable and Ubiquitous Technologies, vol. 5, no. 3, pp. 1–22, 2021

    2021

  12. [20]

    Once-for-all: Train one network and specialize it for efficient deployment,

    H. Cai, C. Gan, T. Wang, Z. Zhang, and S. Han, “Once-for-all: Train one network and specialize it for efficient deployment,” arXiv preprint arXiv:1908.09791, 2019

    1908 arXiv

  13. [21]

    Proxylessnas: Direct neural architecture search on target task and hardware,

    H. Cai, L. Zhu, and S. Han, “Proxylessnas: Direct neural architecture search on target task and hardware,” arXiv preprint arXiv:1812.00332 , 2018

    2018 arXiv

  14. [22]

    Progressive neural architecture search,

    C. Liu, B. Zoph, M. Neumann, J. Shlens, W. Hua, L.-J. Li, L. Fei-Fei, A. Yuille, J. Huang, and K. Murphy, “Progressive neural architecture search,” in Proceedings of the European conference on computer vision (ECCV), 2018, pp. 19–34

    2018

  15. [23]

    Darts: Differentiable architecture search,

    H. Liu, K. Simonyan, and Y . Yang, “Darts: Differentiable architecture search,” arXiv preprint arXiv:1806.09055 , 2018

    2018 arXiv

  16. [24]

    Adaptivenet: Post-deployment neural architecture adaptation for diverse edge environments,

    H. Wen, Y . Li, Z. Zhang, S. Jiang, X. Ye, Y . Ouyang, Y . Zhang, and Y . Liu, “Adaptivenet: Post-deployment neural architecture adaptation for diverse edge environments,” in Proceedings of the 29th Annual International Conference on Mobile Computing and Networking , 2023, pp. 1–17

    2023

  17. [25]

    Legodnn: block-grained scaling of deep neural networks for mobile vision,

    R. Han, Q. Zhang, C. H. Liu, G. Wang, J. Tang, and L. Y . Chen, “Legodnn: block-grained scaling of deep neural networks for mobile vision,” in Proceedings of the 27th Annual International Conference on Mobile Computing and Networking , 2021, pp. 406–419

    2021

  18. [26]

    Neulens: spatial-based dynamic accel- eration of convolutional neural networks on edge,

    X. Hou, Y . Guan, and T. Han, “Neulens: spatial-based dynamic accel- eration of convolutional neural networks on edge,” in Proceedings of the 28th Annual International Conference on Mobile Computing And Networking, 2022, pp. 186–199

    2022

  19. [27]

    Hardware- accelerated platforms and infrastructures for network functions: A survey of enabling technologies and research studies,

    P. Shantharama, A. S. Thyagaturu, and M. Reisslein, “Hardware- accelerated platforms and infrastructures for network functions: A survey of enabling technologies and research studies,” IEEE Access, vol. 8, pp. 132 021–132 085, 2020

    2020

  20. [28]

    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

  21. [29]

    Very deep convolutional networks for large-scale image recognition,

    K. Simonyan and A. Zisserman, “Very deep convolutional networks for large-scale image recognition,” arXiv preprint arXiv:1409.1556 , 2014

    2014 arXiv

  22. [30]

    Mobilenets: Efficient convo- lutional neural networks for mobile vision applications,

    A. G. Howard, M. Zhu, B. Chen, D. Kalenichenko, W. Wang, T. Weyand, M. Andreetto, and H. Adam, “Mobilenets: Efficient convo- lutional neural networks for mobile vision applications,” arXiv preprint arXiv:1704.04861, 2017

    2017 arXiv

  23. [31]

    Squeezenet: Alexnet-level accuracy with 50x fewer parameters and¡ 0.5 mb model size,

    F. N. Iandola, S. Han, M. W. Moskewicz, K. Ashraf, W. J. Dally, and K. Keutzer, “Squeezenet: Alexnet-level accuracy with 50x fewer parameters and¡ 0.5 mb model size,” arXiv preprint arXiv:1602.07360 , 2016

    2016 arXiv

  24. [32]

    Shufflenet: An extremely effi- cient convolutional neural network for mobile devices,

    X. Zhang, X. Zhou, M. Lin, and J. Sun, “Shufflenet: An extremely effi- cient convolutional neural network for mobile devices,” in Proceedings of the IEEE conference on computer vision and pattern recognition , 2018, pp. 6848–6856

    2018

  25. [33]

    Going deeper with convolutions,

    C. Szegedy, W. Liu, Y . Jia, P. Sermanet, S. Reed, D. Anguelov, D. Erhan, V . Vanhoucke, and A. Rabinovich, “Going deeper with convolutions,” in Proceedings of the IEEE conference on computer vision and pattern recognition, 2015, pp. 1–9

    2015

  26. [34]

    Rethinking the inception architecture for computer vision,

    C. Szegedy, V . Vanhoucke, S. Ioffe, J. Shlens, and Z. Wojna, “Rethinking the inception architecture for computer vision,” in Proceedings of the IEEE conference on computer vision and pattern recognition , 2016, pp. 2818–2826

    2016

  27. [35]

    Ghostnet: More features from cheap operations,

    K. Han, Y . Wang, Q. Tian, J. Guo, C. Xu, and C. Xu, “Ghostnet: More features from cheap operations,” in Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, 2020, pp. 1580– 1589

    2020

  28. [36]

    Efficientnet: Rethinking model scaling for con- volutional neural networks,

    M. Tan and Q. Le, “Efficientnet: Rethinking model scaling for con- volutional neural networks,” in International conference on machine learning. PMLR, 2019, pp. 6105–6114

    2019

  29. [37]

    Mnasnet: Platform-aware neural architecture search for mobile,

    M. Tan, B. Chen, R. Pang, V . Vasudevan, M. Sandler, A. Howard, and Q. V . Le, “Mnasnet: Platform-aware neural architecture search for mobile,” in Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , 2019, pp. 2820–2828

    2019

  30. [38]

    Densenet: Implementing efficient convnet descriptor pyra- mids,

    F. Iandola, M. Moskewicz, S. Karayev, R. Girshick, T. Darrell, and K. Keutzer, “Densenet: Implementing efficient convnet descriptor pyra- mids,” arXiv preprint arXiv:1404.1869 , 2014

    2014 arXiv

  31. [39]

    Deep k-means: Re-training and parameter sharing with harder cluster assign- ments for compressing deep convolutions,

    J. Wu, Y . Wang, Z. Wu, Z. Wang, A. Veeraraghavan, and Y . Lin, “Deep k-means: Re-training and parameter sharing with harder cluster assign- ments for compressing deep convolutions,” in International Conference on Machine Learning . PMLR, 2018, pp. 5363–5372

    2018

  32. [40]

    Adadeep: A usage-driven, automated deep model compression framework for enabling ubiquitous intelligent mobiles,

    S. Liu, J. Du, K. Nan, Z. Zhou, H. Liu, Z. Wang, and Y . Lin, “Adadeep: A usage-driven, automated deep model compression framework for enabling ubiquitous intelligent mobiles,” IEEE Transactions on Mobile Computing, vol. 20, no. 12, pp. 3282–3297, 2020

    2020

  33. [41]

    Enabling latency-sensitive dnn inference via joint optimization of model surgery and resource allocation in heterogeneous edge,

    Z. Huang, F. Dong, D. Shen, H. Wang, X. Guo, and S. Fu, “Enabling latency-sensitive dnn inference via joint optimization of model surgery and resource allocation in heterogeneous edge,” in Proceedings of the 51st International Conference on Parallel Processing , 2022, pp. 1–11

    2022

  34. [42]

    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

  35. [43]

    Learning both weights and con- nections for efficient neural network,

    S. Han, J. Pool, J. Tran, and W. Dally, “Learning both weights and con- nections for efficient neural network,” Advances in neural information processing systems, vol. 28, 2015

    2015

  36. [44]

    Nn-stretch: Automatic neural network branching for parallel inference on heterogeneous multi-processors,

    J. Wei, T. Cao, S. Cao, S. Jiang, S. Fu, M. Yang, Y . Zhang, and Y . Liu, “Nn-stretch: Automatic neural network branching for parallel inference on heterogeneous multi-processors,” in Proceedings of the 21st Annual International Conference on Mobile Systems, Applications and S...

    2023

  37. [45]

    Adaenlight: Energy-aware low-light video stream enhancement on mo- bile devices,

    S. Liu, X. Li, Z. Zhou, B. Guo, M. Zhang, H. Shen, and Z. Yu, “Adaenlight: Energy-aware low-light video stream enhancement on mo- bile devices,” Proceedings of the ACM on Interactive, Mobile, Wearable and Ubiquitous Technologies, vol. 6, no. 4, pp. 1–26, 2023

    2023

  38. [46]

    Learning multiple layers of features from tiny images,

    A. Krizhevsky and G. Hinton, “Learning multiple layers of features from tiny images,” https://www.cs.toronto.edu/ ∼kriz/ learning-features-2009-TR.pdf, 2009, cIFAR-10 dataset

    2009

  39. [47]

    Learning multiple layers of features from tiny images,

    ——, “Learning multiple layers of features from tiny images,” https://www.cs.toronto.edu/∼kriz/learning-features-2009-TR.pdf, 2009, cIFAR-100 dataset

    2009

  40. [48]

    Tiny imagenet visual recognition challenge,

    Y . Le and X. Yang, “Tiny imagenet visual recognition challenge,” CS 231N, vol. 7, no. 7, p. 3, 2015

    2015

  41. [49]

    Neural architecture search with reinforcement learning,

    B. Zoph and Q. V . Le, “Neural architecture search with reinforcement learning,” arXiv preprint arXiv:1611.01578 , 2016

    2016 arXiv

  42. [50]

    Attention is all you need,

    A. Vaswani, N. Shazeer, N. Parmar, J. Uszkoreit, L. Jones, A. N. Gomez, Ł. Kaiser, and I. Polosukhin, “Attention is all you need,” Advances in neural information processing systems , vol. 30, 2017

    2017

  43. [51]

    Tiny models are the computational saver for large models,

    Q. Wang, B. Cardiff, A. Frapp ´e, B. Larras, and D. John, “Tiny models are the computational saver for large models,” arXiv preprint arXiv:2403.17726, 2024

    2024 arXiv

  44. [52]

    Pmc: A privacy-preserving deep learning model customization framework for edge computing,

    B. Liu, Y . Li, Y . Liu, Y . Guo, and X. Chen, “Pmc: A privacy-preserving deep learning model customization framework for edge computing,” Proceedings of the ACM on Interactive, Mobile, Wearable and Ubiq- uitous Technologies, vol. 4, no. 4, pp. 1–25, 2020

    2020

  45. [53]

    Distributed deep convolutional neural networks for the internet-of-things,

    S. Disabato, M. Roveri, and C. Alippi, “Distributed deep convolutional neural networks for the internet-of-things,” IEEE Transactions on Com- puters, vol. 70, no. 8, pp. 1239–1252, 2021

    2021

  46. [54]

    Learning in the wild: When, how, and what to learn for on-device dataset adaptation,

    S. Lee and S. Nirjon, “Learning in the wild: When, how, and what to learn for on-device dataset adaptation,” in Proceedings of the 2nd International Workshop on Challenges in Artificial Intelligence and Machine Learning for Internet of Things , 2020, pp. 34–40

    2020

  47. [55]

    A survey of quantization methods for efficient neural network infer- ence,

    A. Gholami, S. Kim, Z. Dong, Z. Yao, M. W. Mahoney, and K. Keutzer, “A survey of quantization methods for efficient neural network infer- ence,” in Low-Power Computer Vision. Chapman and Hall/CRC, 2022, pp. 291–326

    2022

  48. [56]

    Deep compression: Compressing deep neural networks with pruning, trained quantization and huffman coding,

    S. Han, H. Mao, and W. J. Dally, “Deep compression: Compressing deep neural networks with pruning, trained quantization and huffman coding,” arXiv preprint arXiv:1510.00149 , 2015

    2015 arXiv

  49. [57]

    Distilling the knowledge in a neural network,

    G. Hinton, O. Vinyals, and J. Dean, “Distilling the knowledge in a neural network,” arXiv preprint arXiv:1503.02531 , 2015

    2015 arXiv

  50. [58]

    Con- densenet: An efficient densenet using learned group convolutions,

    G. Huang, S. Liu, L. Van der Maaten, and K. Q. Weinberger, “Con- densenet: An efficient densenet using learned group convolutions,” in Proceedings of the IEEE conference on computer vision and pattern recognition, 2018, pp. 2752–2761

    2018

  51. [59]

    Squeezenext: Hardware-aware neural network design,

    A. Gholami, K. Kwon, B. Wu, Z. Tai, X. Yue, P. Jin, S. Zhao, and K. Keutzer, “Squeezenext: Hardware-aware neural network design,” in Proceedings of the IEEE conference on computer vision and pattern recognition workshops, 2018, pp. 1638–1647

    2018

  52. [60]

    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

  53. [61]

    A survey on evolutionary neural architecture search,

    Y . Liu, Y . Sun, B. Xue, M. Zhang, G. G. Yen, and K. C. Tan, “A survey on evolutionary neural architecture search,” IEEE transactions on neural networks and learning systems , vol. 34, no. 2, pp. 550–570, 2021

    2021

  54. [62]

    Edgecompress: Coupling multi-dimensional model compression and dynamic inference for edgeai,

    H. Kong, D. Liu, S. Huai, X. Luo, R. Subramaniam, C. Makaya, Q. Lin, and W. Liu, “Edgecompress: Coupling multi-dimensional model compression and dynamic inference for edgeai,” IEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems , 2023

    2023

  55. [63]

    Latency-aware differentiable neural architecture search,

    Y . Xu, L. Xie, X. Zhang, X. Chen, B. Shi, Q. Tian, and H. Xiong, “Latency-aware differentiable neural architecture search,” arXiv preprint arXiv:2001.06392, 2020

    2001 arXiv

  56. [64]

    Neuralpower: Predict and deploy energy-efficient convolutional neural networks,

    E. Cai, D.-C. Juan, D. Stamoulis, and D. Marculescu, “Neuralpower: Predict and deploy energy-efficient convolutional neural networks,” in Asian Conference on Machine Learning . PMLR, 2017, pp. 622–637

    2017

  57. [65]

    Chamnet: Towards efficient network design through platform-aware model adaptation,

    X. Dai, P. Zhang, B. Wu, H. Yin, F. Sun, Y . Wang, M. Dukhan, Y . Hu, Y . Wu, Y . Jiaet al., “Chamnet: Towards efficient network design through platform-aware model adaptation,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , 2019, pp. 11...

    2019

  58. [66]

    Nn-meter: Towards accurate latency prediction of deep-learning model inference on diverse edge devices,

    L. L. Zhang, S. Han, J. Wei, N. Zheng, T. Cao, Y . Yang, and Y . Liu, “Nn-meter: Towards accurate latency prediction of deep-learning model inference on diverse edge devices,” in Proceedings of the 19th Annual International Conference on Mobile Systems, Applications, and Servi...

    2021

  59. [67]

    Searching the deployable convolution neural networks for gpus,

    L. Wang, C. Yu, S. Salian, S. Kierat, S. Migacz, and A. F. Florea, “Searching the deployable convolution neural networks for gpus,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2022, pp. 12 227–12 236

    2022

  60. [68]

    On-demand deep model compression for mobile devices: A usage-driven model selection framework,

    S. Liu, Y . Lin, Z. Zhou, K. Nan, H. Liu, and J. Du, “On-demand deep model compression for mobile devices: A usage-driven model selection framework,” in Proceedings of the 16th annual international conference on mobile systems, applications, and services , 2018, pp. 389–400

    2018

Pith tools

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