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 Tools annual

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 Tools monthly

Global: Zendy Plus monthly

Presents the access of premium content as premium feature

Premium Content

Global: Zendy Plus annual

Global: Zendy Free Plan

Quantization has emerged as a promising direction for model compression.Recently, data-free quantization has been widely studied as a promising methodto avoid privacy concerns, which synthesizes images as an alternative to realtraining data. Existing methods use classification loss to ensure thereliability of the synthesized images. Unfortunately, even if these images arewell-classified by the pre-trained model, they still suffer from low semanticsand homogenization issues. Intuitively, these low-semantic images are sensitiveto perturbations, and the pre-trained model tends to have inconsistent outputwhen the generator synthesizes an image with poor semantics. To this end, wepropose Robustness-Guided Image Synthesis (RIS), a simple but effective methodto enrich the semantics of synthetic images and improve image diversity,further boosting the performance of downstream data-free compression tasks.Concretely, we first introduce perturbations on input and model weight, thendefine the inconsistency metrics at feature and prediction levels before andafter perturbations. On the basis of inconsistency on two levels, we design arobustness optimization objective to enhance the semantics of synthetic images.Moreover, we also make our approach diversity-aware by forcing the generator tosynthesize images with small correlations in the label space. With RIS, weachieve state-of-the-art performance for various settings on data-freequantization and can be extended to other data-free compression tasks.

Robustness-Guided Image Synthesis for Data-Free Quantization