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

Embedded systems demand on-device processing of data using Neural Networks(NNs) while conforming to the memory, power and computation constraints,leading to an efficiency and accuracy tradeoff. To bring NNs to edge devices,several optimizations such as model compression through pruning, quantization,and off-the-shelf architectures with efficient design have been extensivelyadopted. These algorithms when deployed to real world sensitive applications,requires to resist inference attacks to protect privacy of users training data.However, resistance against inference attacks is not accounted for designing NNmodels for IoT. In this work, we analyse the three-dimensionalprivacy-accuracy-efficiency tradeoff in NNs for IoT devices and propose Geckotraining methodology where we explicitly add resistance to private inferencesas a design objective. We optimize the inference-time memory, computation, andpower constraints of embedded devices as a criterion for designing NNarchitecture while also preserving privacy. We choose quantization as designchoice for highly efficient and private models. This choice is driven by theobservation that compressed models leak more information compared to baselinemodels while off-the-shelf efficient architectures indicate poor efficiency andprivacy tradeoff. We show that models trained using Gecko methodology arecomparable to prior defences against black-box membership attacks in terms ofaccuracy and privacy while providing efficiency.

GECKO: Reconciling Privacy, Accuracy and Efficiency in Embedded Deep Learning