2024 · 11 citations · 8 references
ComputingComputer EngineeringComputer ScienceEdge ComputingEmbedded SystemsMachine LearningDeep LearningEdge ArchitectureEdge Artificial IntelligenceEdge DevicesLatency OptimizationMulti-task LearningBandwidthDistributed Training At The EdgeEdge Machine LearningMulti-task Deep LearningResource Allocation ConstraintsSplit Learning
Split Computing (SC), where a Deep Neural Network (DNN) is intelligently split with a part of it deployed on an edge device and the rest on a remote server is emerging as a promising approach. It allows the power of DNNs to be leveraged for latency-sensitive applications that do not allow the entire DNN to be deployed remotely, while not having sufficient computation bandwidth available locally. In many such embedded systems scenarios, such as those in the automotive domain, computational resource constraints also necessitate Multi-Task Learning (MTL), where the same DNN is used for multiple inference tasks instead of having dedicated DNNs for each task, which would need more computing bandwidth. However, how to partition such a multi-tasking DNN to be deployed within a SC framework has not been sufficiently studied. This paper studies this problem, and MTL-Split, our novel proposed architecture, shows encouraging results on both synthetic and real-world data. The source code is available at https://github.com/intelligolabs/MTL-Split.
8
Knowledge Distillation: A Survey
Jianping Gou, Baosheng Yu, Stephen J. Maybank et al. · International Journal of Computer Vision · 2021 · 3.5K citations · Full text
Computing, Chemical Engineering, Knowledge Representation +2
Pruning and quantization for deep neural network acceleration: A survey
Tailin Liang, John Glossner, Lei Wang et al. · Neurocomputing · 2021 · 867 citations
Yiping Kang, Johann Hauswald, Cao Gao et al. · ACM SIGARCH Computer Architecture News · 2017 · 748 citations
Distilled Split Deep Neural Networks for Edge-Assisted Real-Time Systems
Yoshitomo Matsubara, Sabur Baidya, Davide Callegaro et al. · 2019 · 94 citations · Full text