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

We reduce training time in convolutional networks (CNNs) with a method that,for some of the mini-batches: a) scales down the resolution of input images viadownsampling, and b) reduces the forward pass operations via pooling on theconvolution filters. Training is performed in an interleaved fashion; somebatches undergo the regular forward and backpropagation passes with originalnetwork parameters, whereas others undergo a forward pass with pooled filtersand downsampled inputs. Since pooling is differentiable, the gradients of thepooled filters propagate to the original network parameters for a standardparameter update. The latter phase requires fewer floating point operations andless storage due to the reduced spatial dimensions in feature maps and filters.The key idea is that this phase leads to smaller and approximate updates andthus slower learning, but at significantly reduced cost, followed by passesthat use the original network parameters as a refinement stage. Deciding howoften and for which batches the downsmapling occurs can be done eitherstochastically or deterministically, and can be defined as a traininghyperparameter itself. Experiments on residual architectures show that we canachieve up to 23% reduction in training time with minimal loss in validationaccuracy.

Training CNNs faster with Dynamic Input and Kernel Downsampling