A Collection of White Papers from the BDEC2 Workshop in San Diego, California October 15–17, 2019 Call for Demonstrator Proposals for the BDEC2 Workshop in San Diego, CA iv SmartFlows: “AI for CI” on top of the Computing Continuum for AI-Integrated Applications Ilkay Altintas, Kyle Marcus, Volkan Vural, Shweta Purawat, and Daniel Crawl 1 ZettaFlow: Towards High-Performance ML-based Analytics across the Digital Continuum Gabriel Antoniu, Alexandru Costan, and Ovidiu Marcu 3 Self-Improving Extreme-Scale Systems with Hierarchical Reinforcement Learning Prasanna Balaprakash 7 AI-Aided Scientific Computing Applications on Large-Scale Computing Platforms Rongqiang Cao and Yangang Wang 10 Lossy compression and AI for scientific data: how do they interplay? Franck Cappello, Robert Underwood, Sheng Di, Justin M. Wozniak, and Jon C. Calhoun 12 Feeding the Beast: High Performance Data Pipeline for Large-Scale Deep Learning Cong Xu, Antonio Lain, Paolo Faraboschi, Nic Dube, and Dejan Milojicic 14 Towards Intelligent Management of Heterogeneous Memories with Deep Reinforcement Learning Balazs Gerofi 16 AI and ML for High Energy Physics Maria Girone and Viktor Khristenko 18 Scientific Machine Learning Benchmarks and Datasets Tony Hey 21 Edge-to-Cloud Processing with InLocus and INDIANA Erza Kissel 23 Large-Scale Optimization Strategies for AI&HPC Workloads Yu Liu 25 A Community Machine Learning Commons for Advancing Machine Learning in Earth System Science Richard Loft 27 LUMI, the Pan-European Pre-Exascale Supercomputer – Designed for the Converge of AI and HPC Pekka Manninen and Sebastian von Alfthan 29 Big Data Assimilation Incorporating Deep Learning with Phased Array Radar Data and Numerical Weather Prediction Takemasa Miyoshi 32 Learning at Scale and Learning for Batch Scheduling Bruno Raffin, Olivier Richard, and Denis Trystram 34 Cyberinfrastructure for AI: Rapid Response and Recovery after Hurricane with AI and UAS Maryam Rahnemoonfar and Robin Murphy 37 Spatial Sensor Data, Effective Training and the Computing Continuum Joel Saltz 38 Inline Processing with FPGAs for Edge-to-HPC AI Workflows Kentaro Sano 40 Data Compression with Deep Predictive Neural Network Rupak Roy, Kento Sato, Jian Guo, Jen s Domke, Weikuan Yu, Takaki Hatsui, and Yasumasa Joti 42 AI Feedback Loop for Redshift Surveys in Astronomy Alex Szalay 44