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

Explicit deep generative models (DGMs), e.g., VAEs and Normalizing Flows,have shown to offer an effective data modelling alternative for losslesscompression. However, DGMs themselves normally require large storage space andthus contaminate the advantage brought by accurate data density estimation. Toeliminate the requirement of saving separate models for different targetdatasets, we propose a novel setting that starts from a pretrained deepgenerative model and compresses the data batches while adapting the model witha dynamical system for only one epoch. We formalise this setting as that ofOne-Shot Online Adaptation (OSOA) of DGMs for lossless compression and proposea vanilla algorithm under this setting. Experimental results show that vanillaOSOA can save significant time versus training bespoke models and space versususing one model for all targets. With the same adaptation step number oradaptation time, it is shown vanilla OSOA can exhibit better space efficiency,e.g., $47\%$ less space, than fine-tuning the pretrained model and saving thefine-tuned model. Moreover, we showcase the potential of OSOA and motivate moresophisticated OSOA algorithms by showing further space or time efficiency withmultiple updates per batch and early stopping.

OSOA: One-Shot Online Adaptation of Deep Generative Models for Lossless Compression