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

Using multiple nodes and parallel computing algorithms has become a principaltool to improve training and execution times of deep neural networks as well aseffective collective intelligence in sensor networks. In this paper, weconsider the parallel implementation of an already-trained deep model onmultiple processing nodes (a.k.a. workers) where the deep model is divided intoseveral parallel sub-models, each of which is executed by a worker. Sincelatency due to synchronization and data transfer among workers negativelyimpacts the performance of the parallel implementation, it is desirable to haveminimum interdependency among parallel sub-models. To achieve this goal, wepropose to rearrange the neurons in the neural network and partition them(without changing the general topology of the neural network), such that theinterdependency among sub-models is minimized under the computations andcommunications constraints of the workers. We propose RePurpose, a layer-wisemodel restructuring and pruning technique that guarantees the performance ofthe overall parallelized model. To efficiently apply RePurpose, we propose anapproach based on $\ell_0$ optimization and the Munkres assignment algorithm.We show that, compared to the existing methods, RePurpose significantlyimproves the efficiency of the distributed inference via parallelimplementation, both in terms of communication and computational complexity.

Restructuring, Pruning, and Adjustment of Deep Models for Parallel Distributed Inference