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 Learning (DL) algorithms are the central focus of modern machinelearning systems. As data volumes keep growing, it has become customary totrain large neural networks with hundreds of millions of parameters to maintainenough capacity to memorize these volumes and obtain state-of-the-art accuracy.To get around the costly computations associated with large models and data,the community is increasingly investing in specialized hardware for modeltraining. However, specialized hardware is expensive and hard to generalize toa multitude of tasks. The progress on the algorithmic front has failed todemonstrate a direct advantage over powerful hardware such as NVIDIA-V100 GPUs.This paper provides an exception. We propose SLIDE (Sub-LInear Deep learningEngine) that uniquely blends smart randomized algorithms, with multi-coreparallelism and workload optimization. Using just a CPU, SLIDE drasticallyreduces the computations during both training and inference outperforming anoptimized implementation of Tensorflow (TF) on the best available GPU. Ourevaluations on industry-scale recommendation datasets, with large fullyconnected architectures, show that training with SLIDE on a 44 core CPU is morethan 3.5 times (1 hour vs. 3.5 hours) faster than the same network trainedusing TF on Tesla V100 at any given accuracy level. On the same CPU hardware,SLIDE is over 10x faster than TF. We provide codes and scripts forreproducibility.

SLIDE : In Defense of Smart Algorithms over Hardware Acceleration for Large-Scale Deep Learning Systems