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 and its applications have gained tremendous interest recentlyin both academia and industry. Restricted Boltzmann Machines (RBMs) offer a keymethodology to implement deep learning paradigms. This paper presents a novelapproach for realizing hybrid CMOS-OxRAM based deep generative models (DGM). Inour proposed hybrid DGM architectures, HfOx based (filamentary-type switching)OxRAM devices are extensively used for realizing multiple computational andnon-computational functions such as: (i) Synapses (weights), (ii) internalneuron-state storage, (iii) stochastic neuron activation and (iv) programmablesignal normalization. To validate the proposed scheme we have simulated twodifferent architectures: (i) Deep Belief Network (DBN) and (ii) StackedDenoising Autoencoder for classification and reconstruction of hand-writtendigits from a reduced MNIST dataset of 6000 images. Contrastive-divergence (CD)specially optimized for OxRAM devices was used to drive the synaptic weightupdate mechanism of each layer in the network. Overall learning rule was basedon greedy-layer wise learning with no back propagation which allows the networkto be trained to a good pre-training stage. Performance of the simulated hybridCMOS-RRAM DGM model matches closely with software based model for a 2-layersdeep network. Top-3 test accuracy achieved by the DBN was 95.5%. MSE of the SDAnetwork was 0.003, lower than software based approach. Endurance analysis ofthe simulated architectures show that for 200 epochs of training (single RBMlayer), maximum switching events/per OxRAM device was ~ 7000 cycles.

Design Exploration of Hybrid CMOS-OxRAM Deep Generative Architectures