Litcius/Paper detail

SqueezeNet

Brett Koonce

2021Apress eBooks57 citationsDOI

Abstract

For the next few chapters, we're going to look at convolutional neural networks designed specifically for running on mobile devices, primarily phones. A lot of research has gone into building more complicated models using larger and larger clusters of computers to try and increase accuracy on the Imagenet problem. Mobile phones/edge devices are an area of machine learning that has not been explored as deeply, but in my opinion is extremely important. There is the direct goal of getting devices working on real-world devices, but to me what is interesting in particular is the idea that in finding ways of reducing the complexity of high-end approaches to something simpler, we can discover techniques that will allow us to build even larger networks.

Topics & Concepts

Computer scienceMobile deviceEnhanced Data Rates for GSM EvolutionConvolutional neural networkData scienceArtificial intelligenceHuman–computer interactionMultimediaDistributed computingWorld Wide WebAdvanced Neural Network ApplicationsBig Data and Digital EconomyGenerative Adversarial Networks and Image Synthesis
SqueezeNet | Litcius