English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

Deep neural networks (DNNs) have substantial computational and memoryrequirements, and the compilation of its computational graphs has a greatimpact on the performance of resource-constrained (e.g., computation, I/O, andmemory-bound) edge computing systems. While efficient execution of theircomputational graph requires an effective scheduling algorithm, generating theoptimal scheduling solution is a challenging NP-hard problem. Furthermore, thecomplexity of scheduling DNN computational graphs will further increase onpipelined multi-core systems considering memory communication cost, as well asthe increasing size of DNNs. Using the synthetic graph for the trainingdataset, this work presents a reinforcement learning (RL) based schedulingframework RESPECT, which learns the behaviors of optimal optimizationalgorithms and generates near-optimal scheduling results with short solvingruntime overhead. Our framework has demonstrated up to $\sim2.5\times$real-world on-chip inference runtime speedups over the commercial compiler withten popular ImageNet models deployed on the physical Coral Edge TPUs system.Moreover, compared to the exact optimization methods, the proposed RLscheduling improves the scheduling optimization runtime by up to 683$\times$speedups compared to the commercial compiler and matches the exact optimalsolutions with up to 930$\times$ speedups. Finally, we perform a comprehensivegeneralizability test, which demonstrates RESPECT successfully imitates optimalsolving behaviors from small synthetic graphs to large real-world DNNscomputational graphs.

RESPECT: Reinforcement Learning based Edge Scheduling on Pipelined Coral Edge TPUs