EPNet: Learning to Exit with Flexible Multi-Branch Network
The DOI matches its Crossref record; 23 of the 24 references it cites were found in a registry or an index.
Abstract
Dynamic inference is an emerging technique that reduces the computational cost of deep neural network under resource-constrained scenarios, such as inference on mobile devices. One way to achieve dynamic inference is to leverage multi-branch neural networks that apply different computation on input data by following different branches. Conventional research on multi-branch neural networks mainly targeted at improving the accuracy of each branch, and use manually designed rules to decide which input follows which branch of the network. Furthermore, these networks often provide a small number of exits, limiting their ability to adapt to external changes. In this paper, we investigate the problem of designing a flexible multi-branch network and early-exiting policies that can adapt to the resource consumption to individual inference request without impacting the inference accuracy. We propose a lightweight branch structure that also provides fine-grained flexibility for early-exiting and leverage Markov decision process (MDP) to automatically learn the early-exiting policies. Our proposed model, EPNet, was effective in reducing inference cost without impacting accuracy by choosing the most suitable branch exit. We also observe that EPNet achieved 3% higher accuracy with an inference budget, compared to state-of-the-art approaches.
Overview
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
Claims other papers make about this one
Each is checked against this paper’s own text: does it say what the citing paper says it does?
In the citing paper
“Deep learning models often exhibit redundancy, where shallower layers suffice to classify 'easy' input samples accurately, while deeper layers are only necessary for more complex or ambiguous inputs [31].”
From A cascade framework for on-device uncertainty-aware event detection on microcontrollers (2026) · cites this paper as [31]
Citations
Cites
Deep Residual Learning for Image Recognition
He, K., Zhang, X., Ren, S., & Sun, J. (2016) · 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr.2016.90 · Cited by 235,474 · cited 5×
“The main branch f m can be a normal convolutional neural network, e.g. , AlexNet [14], ResNet [8] or DenseNet [11].”
BackgroundMethodInfluentialImageNet classification with deep convolutional neural networks
Krizhevsky, A., Sutskever, I., & Hinton, G. E. (2017) · Communications of the ACM · doi:10.1145/3065386 · Cited by 112,155 · cited 3×
“The main branch f m can be a normal convolutional neural network, e.g. , AlexNet [14], ResNet [8] or DenseNet [11].”
BackgroundDensely Connected Convolutional Networks
Huang, G., Liu, Z., Van Der Maaten, L., & Weinberger, K. Q. (2017) · 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr.2017.243 · Cited by 48,566
Delving Deep into Rectifiers: Surpassing Human-Level Performance on ImageNet Classification
He, K., Zhang, X., Ren, S., & Sun, J. (2015) · 2015 IEEE International Conference on Computer Vision (ICCV) · doi:10.1109/iccv.2015.123 · Cited by 19,741 · cited 2×
“The ResNet with same structure of the EP-Net’s main branch network, can only achieve 32.14% under the limited budget, which is 56.54% lower than the EPNet.”
BackgroundMethodInfluentialBatch Normalization: Accelerating Deep Network Training by Reducing Internal Covariate Shift
Ioffe, S. (2024) · arXiv (Cornell University) · doi:10.57702/o9raffed · Cited by 14,214
Distilling the Knowledge in a Neural Network
Hinton, G., Vinyals, O., & Dean, J. (2015) · arXiv · doi:10.48550/arxiv.1503.02531 · Cited by 14,204 · cited 1×
“…resource-intensive models in deployment scenarios such as mobile devices, where desired resources are constrained, prior work proposed techniques such as model pruning [5, 15], low-rank factorization [18, 19], knowledge distillation [3, 9], efficient CNN [12, 24], and hard attention model [1, 17].”
BackgroundShuffleNet: An Extremely Efficient Convolutional Neural Network for Mobile Devices
Zhang, X., Zhou, X., Lin, M., & Sun, J. (2018) · 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition · doi:10.1109/cvpr.2018.00716 · Cited by 9,982
SqueezeNet: AlexNet-level accuracy with 50x fewer parameters and <0.5MB model size
Iandola, F. N., Han, S., Moskewicz, M. W., Ashraf, K., Dally, W. J., & Keutzer, K. (2016) · arXiv · doi:10.48550/arxiv.1602.07360 · Cited by 6,201 · cited 8×
“Deep convolutional neural networks (CNNs) have achieved good accuracy on computer vision tasks such as image classification [8, Permission to make digital or hard copies of all or part of this work for personal or classroom use is granted without fee provided that copies are not made or distributed for profit or comme…”
MethodInfluentialSimple Statistical Gradient-Following Algorithms for Connectionist Reinforcement Learning
Williams, R. J. (1992) · Machine Learning · doi:10.1007/bf00992696 · Cited by 3,416
Recurrent Models of Visual Attention
Mnih, V., Heess, N., Graves, A., & Kavukcuoglu, K. (2014) · arXiv (Cornell University) · doi:10.48550/arxiv.1406.6247 · Cited by 2,568 · cited 2×
“Comparisons of EPNet on three datasets to two types of baselines demonstrate its efficacy in classification accuracy, adaptivity to sample difficulty, and resource budgets.”
Background
Research works printed but not linked: 1
- Song Han Jeff Pool John Tran and William Dally. 2015. Learning both weights and connections for efficient neural network. InAdvances in Neural Information Processing Systems 28 (NeurIPS'15). 1135--1143. Song Han Jeff Pool John Tran and William Dally. 2015. Learning both weights and connections for efficient neural network. InAdvances in Neural Information Processing Systems 28 (NeurIPS'15). 1135--1143.
Cited by
Dynamic Neural Networks: A Survey
Han, Y., Huang, G., Song, S., Yang, L., Wang, H., & Wang, Y. (2022) · IEEE Transactions on Pattern Analysis and Machine Intelligence · doi:10.1109/tpami.2021.3117837 · Cited by 848
Adaptive Inference through Early-Exit Networks: Design, Challenges and Directions
Laskaridis, S., Kouris, A., & Lane, N. D. (2021) · International Workshop on Embedded and Mobile Deep Learning (EMDL) · doi:10.1145/3469116.3470012 · Cited by 203
“EpNet [8] Vision/Classification Non-differentiable exit policy for EE-networks.”
BackgroundEarly-Exit Deep Neural Network - A Comprehensive Survey
Haseena Rahmath, P., Srivastava, V., Chaurasia, K., Pacheco, R. G., & Couto, R. S. (2024) · ACM Computing Surveys · doi:10.1145/3698767 · Cited by 162
“This section delves deeply into previous works, exploring how side branches were constructed, the strategies adopted One fc layer [ 14 , 16 , 17 , 21 , 41 , 60 , 66 , 70 , 77 , 84 , 89 , 104 , 106 , 121 , 125 , 139 , 140 , 141 , 145 , 156 ] Multiple fc layers [ 155 ] One conv layer and one fc layer…”
InfluentialDynamic Early Exit Scheduling for Deep Neural Network Inference through Contextual Bandits
Ju, W., Bao, W., Ge, L., & Yuan, D. (2021) · ACM International Conference on Information and Knowledge Management (CIKM) · doi:10.1145/3459637.3482335 · Cited by 39
“[10] is the first study that designs and trains an early exit model Full Paper Track CIKM ’21, November 1–5, 2021, Virtual Event, Australia”
BackgroundSteering Angle-Guided Multimodal Fusion Lane Detection for Autonomous Driving
Gong, Y., Zhang, X., Lu, J., Jiang, X., Wang, Z., Liu, H., Li, Z., Wang, L., Yang, Q., & Wu, X. (2024) · IEEE Transactions on Intelligent Transportation Systems · doi:10.1109/tits.2024.3507536 · Cited by 26
Neural network developments: A detailed survey from static to dynamic models
Verma, P. R., Singh, N. P., Pantola, D., & Cheng, X. (2024) · Computers & Electrical Engineering · doi:10.1016/j.compeleceng.2024.109710 · Cited by 32
Adaptive Deep Neural Network Inference Optimization with EENet
İlhan, F., Chow, K.-H., Hu, S., Huang, T., Tekin, S. F., Wei, W., Wu, Y., Lee, M., Kompella, R. R., Latapie, H. M., et al. (2024) · IEEE Winter Conference on Applications of Computer Vision · doi:10.1109/wacv57701.2024.00140 · Cited by 46
“EPNet [3] proposes using Markov decision processes however, they add an early exit classifier at each exit to increase the number of states, which is computationally unfeasible since each early exit introduces an additional computational cost during both training and inference.”
BackgroundMethodComplexity-aware Adaptive Training and Inference for Edge-Cloud Distributed AI Systems
Long, Y., Chakraborty, I., Srinivasan, G., & Roy, K. (2021) · IEEE International Conference on Distributed Computing Systems (ICDCS) · doi:10.1109/icdcs51616.2021.00061 · Cited by 30
Semi-HFL: semi-supervised federated learning for heterogeneous devices
Zhong, Z., Wang, J., Bao, W., Zhou, J., Zhu, X., & Zhang, X. (2022) · Complex & Intelligent Systems · doi:10.1007/s40747-022-00894-4 · Cited by 26
Online deep learning’s role in conquering the challenges of streaming data: a survey
Sulaiman, M. A., Farmanbar, M., Kagami, S., Belbachir, A. N., & Rong, C. (2025) · Knowledge and Information Systems · doi:10.1007/s10115-025-02351-3 · Cited by 26
SplitEE: Early Exit in Deep Neural Networks with Split Computing
Bajpai, D. J., Trivedi, V., Yadav, S. L., & Hanawal, M. K. (2023) · International Conference on AI-ML Systems (AIMLSystems) · doi:10.1145/3639856.3639873 · Cited by 27
“Pacheco [24] utilize both multi-exit DNN and DNN partitioning to offload mobile devices via multi-exit DNNs. Similarly, EPNet [4] learns when to exit considering the accuracy-overhead trade-off but in an offline fashion.”
BackgroundSun, Z., Du, X., Song, F., Wang, S., & Li, L. (2024) · ACM/IEEE International Conference on Software Engineering: Companion Proceedings (ICSE-Companion) · doi:10.1145/3597503.3639120 · Cited by 29
“Dynamic inference [10, 16, 20, 25, 31] is one of the most promising techniques to mitigate this urgent issue associated with large models.”
BackgroundMeta-GF: Training Dynamic-Depth Neural Networks Harmoniously
Sun, Y., Li, J., & Xu, X. (2022) · Lecture notes in computer science · doi:10.1007/978-3-031-20083-0_41 · Cited by 20
“Compared with the confidence-based criterion, the learned policy networks trained by reinforcement learning[25, 19, 2, 7] or variational Bayes optimization[4] have better expandability that can be transferred to other tasks such as the object tracking[19].”
MethodAdaEE: Adaptive Early-Exit DNN Inference Through Multi-Armed Bandits
Pacheco, R. G., Shifrin, M., Couto, R. S., Menasché, D. S., Hanawal, M. K., & Campista, M. E. M. (2023) · International Conference on Communications · doi:10.1109/icc45041.2023.10279243 · Cited by 18
“Similarly, EPNet [21] learns when to exit early, accounting for the tradeoff between overhead and accuracy, but the learning occurs offline.”
BackgroundLayercake: Efficient Inference Serving with Cloud and Mobile Resources
Ogden, S. S., & Guo, T. (2023) · IEEE/ACM International Symposium on Cluster, Cloud and Grid Computing (CCGrid) · doi:10.1109/ccgrid57682.2023.00027 · Cited by 12
“There are also a number of hybrid approaches that look at how to leverage resources both on-device and in-cloud by dividing the execution of a model between on-device and in-cloud resources [10, 22, 38, 41].”
BackgroundMethodSelf-citationVashist, A., Shanmugham, S. V. V., Ganguly, A., & Dinakarrao, S. M. P. (2022) · International Conference on Consumer Electronics · doi:10.1109/icce53296.2022.9730182 · Cited by 11
“EPNet [11] utilizes a dynamic approach for providing exit selection.”
MethodAdaptive Early Exiting for Collaborative Inference over Noisy Wireless Channels
Jankowski, M., Gündüz, D., & Mikolajczyk, K. (2024) · IEEE International Conference on Machine Learning for Communication and Networking (ICMLCN) · doi:10.1109/icmlcn59089.2024.10624796 · Cited by 13
“…in the context of wireless edge collaborative inference systems, and one of the first (similar NNs for accepting early exit output were proposed in [13], [14] in the context of efficient DNN inference) to study a deep learning based decision mechanism for accepting early exit inference results.”
BackgroundElastic DNN Inference with Unpredictable Exit in Edge Computing
Huang, J., Gao, Y., & Dong, W. (2023) · IEEE International Conference on Distributed Computing Systems (ICDCS) · doi:10.1109/icdcs57875.2023.00009 · Cited by 8
Zhang, Z., Liu, Z., Zhao, Y., Qiu, C., Zhang, C., & Wang, X. (2023) · IEEE Transactions on Network Science and Engineering · doi:10.1109/tnse.2023.3344850 · Cited by 9
“Many works have studied a new network architecture, the multi-exit model [26], [27], [28], which allows some test samples to exit early.”
BackgroundElastic DNN Inference With Unpredictable Exit in Edge Computing
Huang, J., Gao, Y. C., & Dong, W. (2024) · IEEE Transactions on Mobile Computing · doi:10.1109/tmc.2024.3441946 · Cited by 9
“Furthermore, the learned decision models [29], [30] determine the inference depth for different samples right from the start.”
Unsupervised Early Exit in DNNs with Multiple Exits
U, H. N. N., Hanawal, M. K., & Bhardwaj, A. (2022) · International Conference on AI-ML Systems (AIMLSystems) · doi:10.1145/3564121.3564137 · Cited by 16
“In a similar vein, EPNet [7] learns when to multi-exit accounting for the tradeoff between overhead and accuracy, but the learning occurs in an offline fashion.”
BackgroundEnsemble early exit network on human activity recognition using wearable sensors
Yu, J., Zhang, L., Cheng, D., Bu, C., Liu, L., Wu, H., & Song, A. (2025) · Computer Networks · doi:10.1016/j.comnet.2025.111409 · Cited by 6
Learning scalable dynamic filter in convolutional networks
Wu, S., Xiao, C., Ying, X., Wang, L., Yang, J., & An, W. (2023) · Pattern Recognition Letters · doi:10.1016/j.patrec.2023.10.020 · Cited by 4
Verma, P. R., Singh, N. P., & Pantola, D. (2024) · Multimedia Tools and Applications · doi:10.1007/s11042-024-19670-2 · Cited by 4
“[40].”
FPGA Acceleration of Dynamic Neural Networks: Challenges and Advancements
Dimitriou, A., Biggs, B., Hare, J. S., & Merrett, G. V. (2024) · IEEE International Conference on Omni-layer Intelligent Systems (COINS) · doi:10.1109/coins61597.2024.10622127 · Cited by 6
“In BranchyNet [26], the branch contains additional convolution and FC layers, while in EPNet [27] and DynExit [28], the convolution layers are omitted.”
BackgroundTiny Models are the Computational Saver for Large Models
Wang, Q., Cardiff, B., Frappé, A., Larras, B., & John, D. (2024) · Lecture notes in computer science · doi:10.1007/978-3-031-72992-8_10 · Cited by 7
“Alternative exiting policies include but not be limited to stopping when output entropy is low enough [57], consecutive exits yield consistent predictions [73], or a trained controller decides to exit [10].”
BackgroundRealization of Early-Exit Dynamic Neural Networks on Reconfigurable Hardware
Dimitriou, A., Xun, L., Hare, J. S., & Merrett, G. V. (2024) · IEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems · doi:10.1109/tcad.2024.3519055 · Cited by 6
“For example, in BranchyNet [9], the early-exit modules consist of convolutional and FC layers alongside some activation functions, while on EPNet [28] and dynaexit [10], the convolutional layers are skipped.”
UR2M: Uncertainty and Resource-Aware Event Detection on Microcontrollers
Jia, H., Kwon, Y. D., Ma, D., Pham, N., Qendro, L., Vu, T. N., & Mascolo, C. (2024) · IEEE International Conference on Pervasive Computing and Communications · doi:10.1109/percom59722.2024.10494467 · Cited by 11
“This indicates that these shallower representations can identify “easy” samples, thus avoiding extra computation, whereas more “difficult” samples require processing through deeper layers [29].”
BackgroundResultDyCE: Dynamically Configurable Exiting for deep learning compression and real-time scaling
Wang, Q., Cardiff, B., Frappé, A., Larras, B., & John, D. (2025) · Future Generation Computer Systems · doi:10.1016/j.future.2025.107837 · Cited by 5
Exploration of Decision Sub-Network Architectures for FPGA-based Dynamic DNNs
Dimitriou, A., Hu, M., Hare, J. S., & Merrett, G. V. (2023) · Design, Automation, and Test in Europe · doi:10.23919/date56975.2023.10137302 · Cited by 2
“N ETWORK A RCHITECTURES Decision sub-networks are small neural networks, containing convolutional, fully-connected (FC), or pooling layers [7], [9].”
BackgroundFIANCEE: Faster Inference of Adversarial Networks via Conditional Early Exits
Karpikova, P., Radionova, E., Yaschenko, A., Andrei, S., Kostyushko, L., Fabbricatore, R., & Aleksei, I. (2023) · IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings - IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings · doi:10.1109/cvpr52729.2023.01158 · Cited by 6
“In this work we refer to [17], as the state-of-the-art method for one-shot, high-resolution neural head reenactment.”
MethodEarly-Exit with Class Exclusion for Efficient Inference of Neural Networks
Wang, J., Li, B., & Zhang, G. L. (2024) · IEEE International Conference on Artificial Intelligence Circuits and Systems (AICAS) · doi:10.1109/aicas59952.2024.10595861 · Cited by 11
“EPNet [15] develops a lightweight early-exit structure and determines the exit policy using a Markov decision process.”
MethodAdyna: Accelerating Dynamic Neural Networks with Adaptive Scheduling
Li, Z., Yang, B., Li, J., Chen, T., Li, X., & Gao, M. (2025) · IEEE International Symposium on High-Performance Computer Architecture (HPCA) · doi:10.1109/hpca61900.2025.00049 · Cited by 7
“Recently, a new type of DNNs named dynamic architecture neural networks (DynNNs) have emerged in computer vision and natural language processing [12], [14], [19], [42], [59], [70].”
A cascade framework for on-device uncertainty-aware event detection on microcontrollers
Jia, H., Kwon, Y. D., Ma, D., Pham, N., Qendro, L., Vu, T., & Mascolo, C. (2026) · Pervasive and Mobile Computing · doi:10.1016/j.pmcj.2026.102208 · Cited by 2
“Deep learning models often exhibit redundancy, where shallower layers suffice to classify ’easy’ input samples accurately, while deeper layers are only necessary for more complex or ambiguous inputs [31].”
Li, B., Cao, X., Li, J., Li, J., Wei, X., Geng, J., & Zhang, R. (2025) · Electronics · doi:10.3390/electronics14132628 · Cited by 2
“EPNet [37] adopts a two-stage training approach: first, fully training the backbone network, then freezing backbone parameters and independently training each side branch classifier.”
Yin, C., Qu, H., & Li, Y. (2025) · Journal of Intelligent & Fuzzy Systems · doi:10.1177/18758967251356865 · Cited by 4
“On the KITTI dataset, our proposed algorithm is quantitatively compared with other 3D object detection algorithms, including PointRCNN, SECOND, PointPillars, VoxelNet, AVOD-FPN, F-PointNet, EPNet (Dai et al., 2020), CLOCs (Pang et al., 2020), and MVXNet (Sindagi et al., 2019).”
Early-Exit Neural Architecture Search for Energy-Harvesting Edge Computing
Sixdenier, P.-L., Deutel, M., & Teich, J. (2025) · IEEE International Symposium on Embedded Multicore/Many-core Systems-on-Chip (MCSoC) · doi:10.1109/mcsoc67473.2025.00066 · Cited by 1
Early Exiting-Enabled Siamese Tracking for Edge Intelligence Applications
Hussein, M. F., & Özbay, S. (2021) · IEEE Access · doi:10.1109/access.2021.3119604
“Later researches [22], [23] proposed learned policies for early exiting that automatically decide when to exit without using manually defined thresholds in order to maintain the accuracy.”
BackgroundExploiting Epistemic Uncertainty at Inference Time for Early-Exit Power Saving
Dymond, J., Stein, S., & Gunn, S. (2023) · Frontiers in artificial intelligence and applications · doi:10.3233/faia230323
“More recently, like in this work, the focus has shifted to optimised inference techniques for them [9, 38, 48].”
3D Object Detection Method Based on CA Sampling and Local Attention Feature Encoding
Wang, K., Chen, H., Lv, N., Zhang, J., & Chen, C. (2023) · IEEE International Conference on Smart Internet of Things (SmartIoT) · doi:10.1109/smartiot58732.2023.00040
Future Frontiers in Deep Learning: From Static to Dynamic Neural Network TechnologiesPreprint
Verma, P. R., Singh, N. P., Pantola, D., & Cheng, X. (2024) · SSRN Electronic Journal · doi:10.2139/ssrn.4832892
Adaptive Dynamic Inference Framework Using Multi-Route Neural Networks Under Constraints
Cai, H., Xiang, Z., She, H., Kuang, Y., Guo, Y., Luo, M., Wang, Y., & Luo, Y. (2025) · Communications in computer and information science · doi:10.1007/978-981-96-9911-7_9 · Cited by 1
Efficient Neural Network Compression for Fast Inference on Hardware
Wang, J., Jiang, M., & Zhang, G. L. (2025) · IEEE Computer Society Annual Symposium on VLSI · doi:10.1109/isvlsi65124.2025.11130227
MASI: Memory-Adaptive Inference Framework for Spiking Neural Networks on Edge Devices
Yu, D., Zheng, H., Lv, C., Du, X., Jiang, L., Liu, X. M., Pan, G., & Deng, S. (2026) · International World Wide Web Conference (WWW) · doi:10.1145/3774904.3792136 · Cited by 1
XAI-Exit: Interpretability-Driven Dynamic Early Exits for Efficient and Transparent DNN Inference
Haseena Rahmath, P., Abraham, A., & Chaurasia, K. (2026) · IEEE Transactions on Neural Networks and Learning Systems · doi:10.1109/tnnls.2026.3685408 · Cited by 1
“EpNet [14] formulates the task as a Markov decision process, while SkipNet [15] and BlockDrop [16] use reinforcement learning to skip layers.”
SAND: Spatially Adaptive Network Depth for Fast Sampling of Neural Implicit Surfaces
Yang, C., Hou, J., Liu, Y., Ren, S., Wei, G., Komura, T., Zhou, Y., & Wang, W. (2026) · ACM Transactions on Graphics · doi:10.1145/3811341
“EPNet [Dai et al. 2020] explores a more adaptive strategy by learning exit policies through a Markov decision process, providing finer control over the inference process.”
Liu, Z., Abou-Zeid, H., & Wu, H. (2026) · IEEE Internet of Things Journal · doi:10.1109/jiot.2026.3710690
Distribution-aware Re-representations for Multi-Scenario Recommendations
Sun, Q., Xu, Y., Wang, Z., Kang, X., Jin, K., & Gao, J. (2026) · International ACM SIGIR Conference on Research and Development in Information Retrieval (SIGIR) · doi:10.1145/3805712.3809554
Metadata
- Authors
- Dai, X., Kong, X., & Guo, T.
- Affiliations
- Worcester Polytechnic Institute (US)
- Venue
- Proceedings of the 29th ACM International Conference on Information & Knowledge Management
- Publisher
- ACM
- Published
- 2020-10-19
- Pages
- 235-244
- ISSN
- 2155-0751 (ISSN-L)
- OpenAlex
- W3094307698
- Cited by
- 48
- References
- 24
- Citations by year
- 2026: 6 · 2025: 9 · 2024: 14 · 2023: 10 · 2022: 4 · 2021: 5
- FWCI
- 1.47
- Type
- conference paper
- Language
- en
- Open access
- closed
- Locations
- ACM International Conference on Information and Knowledge Management (CIKM) (publishedVersion)
- Funders
- NSF (CNS-1815619)
- Topics
- Physical Sciences › Computer Science › Computer Vision and Pattern Recognition › Advanced Neural Network Applications
- Physical Sciences › Computer Science › Artificial Intelligence › Domain Adaptation and Few-Shot Learning
- Physical Sciences › Computer Science › Computer Networks and Communications › IoT and Edge/Fog Computing
- Keywords
- dynamic inference, multi-branch network, Markov decision process, resource-constrained inference, mobile devices, inference cost reduction
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.