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

Learning convolutional neural networks (CNNs) with low bitwidth ischallenging because performance may drop significantly after quantization.Prior arts often discretize the network weights by carefully tuninghyper-parameters of quantization (e.g. non-uniform stepsize and layer-wisebitwidths), which are complicated and sub-optimal because the full-precisionand low-precision models have a large discrepancy. This work presents a novelquantization pipeline, Frequency-Aware Transformation (FAT), which has severalappealing benefits. (1) Rather than designing complicated quantizers likeexisting works, FAT learns to transform network weights in the frequency domainbefore quantization, making them more amenable to training in low bitwidth. (2)With FAT, CNNs can be easily trained in low precision using simple standardquantizers without tedious hyper-parameter tuning. Theoretical analysis showsthat FAT improves both uniform and non-uniform quantizers. (3) FAT can beeasily plugged into many CNN architectures. When training ResNet-18 andMobileNet-V2 in 4 bits, FAT plus a simple rounding operation already achieves70.5% and 69.2% top-1 accuracy on ImageNet without bells and whistles,outperforming recent state-of-the-art by reducing 54.9X and 45.7X computationsagainst full-precision models. We hope FAT provides a novel perspective formodel quantization. Code is available at\url{https://github.com/ChaofanTao/FAT_Quantization}.

FAT: Learning Low-Bitwidth Parametric Representation via Frequency-Aware Transformation