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

Quantization is one of the core components in lossy image compression. Forneural image compression, end-to-end optimization requires differentiableapproximations of quantization, which can generally be grouped into threecategories: additive uniform noise, straight-through estimator and soft-to-hardannealing. Training with additive uniform noise approximates the quantizationerror variationally but suffers from the train-test mismatch. The other twomethods do not encounter this mismatch but, as shown in this paper, hurt therate-distortion performance since the latent representation ability isweakened. We thus propose a novel soft-then-hard quantization strategy forneural image compression that first learns an expressive latent space softly,then closes the train-test mismatch with hard quantization. In addition, beyondthe fixed integer quantization, we apply scaled additive uniform noise toadaptively control the quantization granularity by deriving a new variationalupper bound on actual rate. Experiments demonstrate that our proposed methodsare easy to adopt, stable to train, and highly effective especially on complexcompression models.

Soft then Hard: Rethinking the Quantization in Neural Image Compression