BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Memento EPFL//
BEGIN:VEVENT
SUMMARY:Towards optimal chemical space search with generative virtual scre
 ening
DTSTART:20260303T151500
DTEND:20260303T161500
DTSTAMP:20260528T093152Z
UID:f286212fa9e9d786e89850e1578d6e266b354a78d262f86c9a70d990
CATEGORIES:Conferences - Seminars
DESCRIPTION:Morgan Thomas is a Postdoctoral Fellow at Khalifa University i
 n Abu Dhabi\, where he leads machine learning in medicine initiatives with
  Prof. Andreas Bender. Following his experience on the AstraZeneca Gradua
 te Programme\, he received his PhD from the University of Cambridge spons
 ored by Nxera Pharmaceuticals. He conducted his first postdoctoral fello
 wship with Prof. Gianni de Fabritiis at Universitat Pompeu Fabra supporte
 d by a Johnson & Johnson Innovative Medicine Grant. His research focuses 
 on machine learning methods to automate and improve drug discovery\, parti
 cularly through generative AI\, structure-based principles\, and benchmar
 king frameworks. He consistently works at the interface of academia and i
 ndustry\, grounding his research in practical application.\nEfficiently se
 arching virtual chemical space is central to discovering new chemical matt
 er. Traditionally\, virtual screening of enumerated libraries and related 
 approximations has been used to navigate accessible subsets of chemical sp
 ace. More recently\, generative AI has emerged as a strategy for implicitl
 y exploring vastly larger chemical spaces. In this talk\, I compare these
  paradigms for chemical space search\, focusing on computational efficienc
 y\, practical progressability\, and evaluability. I will share recent work
  on improving the sample efficiency of reinforcement learning–based gene
 rative methods and introduce new benchmarks that stress-test current appro
 aches under extreme search requirements\, clarifying how close we are to t
 ruly optimal chemical space exploration.
LOCATION:https://epfl.zoom.us/j/68447908297?pwd=OU5JUGJUSUhZc0ZNYjQ2WENvYI
 NRdz09
STATUS:CONFIRMED
END:VEVENT
END:VCALENDAR