1st Workshop on Hot Topics in Distributed Machine Learning

Aim

Modern society is experiencing the third spring of AI, with Distributed Machine Learning (DiML) systems used daily in healthcare, mobile, computer vision, commercial activities, finance, and natural language processing. Designing DiML systems is challenging, requiring reliability, efficiency, and trustworthiness. Academia and industry are researching solutions, highlighting the need for more venues focused on DiML. Current DiML workshops often lack interest in lessons from unsuccessful experiments or new perspectives on established topics. To address this, we envision HotDiML as a platform for critical thinking, sharing both successes and failures, and fostering constructive discussions. HotDiML encourages novel directions, accepting position papers that offer fresh perspectives or critical analyses, even if not fully developed. These papers can provide valuable insights to advance the field. This workshop invites researchers and practitioners to: (i) provide insights on DiML design challenges, (ii) present preliminary results, (iii) critique other works constructively, and (iv) share lessons learned from testing hypotheses.

Topics of Interest

The workshop invites authors to submit papers on the following (but not limited to) topics of interest:

Security of FL — Privacy and anonymity in FL — Network-based solutions for DiML — Multi-party computation for ML — Homomorphic encryption for DiML — Differential privacy in FL — Exciting and unorthodox applications of DiML — DiML scalability — Online learning — Hardware solutions for enabling DiML — Split learning — Cloud continuum for DiML — Theoretical findings concerning DiML and FL — Neuro-symbolic distributed learning — FL-based protocols for security and privacy