BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//CERN//INDICO//EN
BEGIN:VEVENT
SUMMARY:Generative deep learning with high-resolution SDO EUV images of th
 e Sun
DTSTART;VALUE=DATE-TIME:20230512T100000Z
DTEND;VALUE=DATE-TIME:20230512T101500Z
DTSTAMP;VALUE=DATE-TIME:20240704T115655Z
UID:indico-contribution-351@meetings.aip.de
DESCRIPTION:Speakers: Frederic Effenberger (Ruhr-University Bochum)\nWe pr
 esent results of generative deep learning as applied to a large database o
 f solar images and discuss challenges in training and validation\, in part
 icular with distributed training on GPU supercomputers. Our dataset is bas
 ed on SDO EUV data with high resolution and with excellent coverage since 
 2010. This dataset is thus well suited to study the application of advance
 d machine learning techniques that require large amounts of data for train
 ing\, such as deep learning approaches. We focus on diffusion type models 
 and generative adversarial deep learning (GANs). We address the potential 
 of data augmentation techniques for improved learning and image quality an
 d the opportunities for latent space structure exploration and control. Th
 e challenges in reaching the highest resolutions with good image fidelity 
 are analyzed. Potential application downstream that can make use of such g
 enerated images are briefly discussed and the need for a community-driven\
 , physics-based basis to establish evaluation criteria for generative mode
 ls will be emphasized.\n\nhttps://meetings.aip.de/event/24/contributions/3
 51/
LOCATION:Haus H\, Telegrafenberg
URL:https://meetings.aip.de/event/24/contributions/351/
END:VEVENT
END:VCALENDAR