BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Memento EPFL//
BEGIN:VEVENT
SUMMARY:"Machine learning in chemistry and beyond" (ChE-651) seminar by Es
 ther Heid "Graph-convolutional neural network: From molecules to chemical 
 reactions"
DTSTART:20220322T151500
DTEND:20220322T161500
DTSTAMP:20260603T162340Z
UID:aaea0baf4991f706711ac8b96477310d343646dd8352928a64e48571
CATEGORIES:Conferences - Seminars
DESCRIPTION:Esther Heid        \nGraph-convolutional neural networ
 ks (GCNNs) are very successful in predicting various properties of molecul
 es\, outperforming descriptor or fingerprint based neural networks if plen
 ty of data is available. GCNNs allow for a learned extraction of important
  characteristics of a molecule and enable end-to-end learning\, instead of
  relying on expert\, system-dependent knowledge. However\, the properties 
 of chemical reactions\, i.e. the combination of reactant and product molec
 ules\, are not readily accessible with current GCNNs which are designed to
  take molecular graphs as input. We therefore developed GCNNs based on the
  condensed graphs of reaction (CGR)\, where reactants and products are mer
 ged into a single pseudo-molecular graph\, i.e. an artificial graph transi
 tion state. In this talk\, the anatomy of molecular GCNNs will be discusse
 d in detail\, as well as the changes necessary to encode reactions instead
  of molecules. Compared to previous approaches such as neural networks on 
 reaction fingerprints\, transformer models on reaction SMILES strings or s
 eparate GCNNs for reactant and product encodings\, GCNNs on CGRs offer a c
 omparable or better performance with a lower number of parameters. We show
 case the performance on different regression and classification tasks\, su
 ch as the prediction of activation energies\, rate constants or the classi
 fication into name reactions. Furthermore\, possible pitfalls for molecule
  and reaction GCNNs will be discussed.\n\nEsther Heid received her Bachel
 or’s (2014) and Master’s degree (2016) in Chemistry from the Universit
 y of Vienna\, Austria\, as well as a doctoral degree in Chemistry\, with f
 ocus on Theoretical Chemistry\, in 2019. Within her PhD program she visite
 d the London Imperial College\, as well as the University of Maryland Balt
 imore for short term research trips.\nIn her thesis\, she investigated how
  computer simulation can be utilized to study solvation dynamics and other
  solvent effects in bulk phase and close to biomolecules such as saccharid
 es or proteins. Her expertise comprises molecular dynamics simulations\, q
 uantum mechanical calculations\, machine learning\, method development and
  the automation of work flows.\nIn 2020 she joined the Green Research grou
 p\, holding an Erwin-Schroedinger Postdoctoral Fellowship from the Austria
 n Science Fund\, which enables her to conduct research on the development 
 of a computer-aided tool for finding novel multi-enzyme networks which yie
 ld a specified target molecule. The project utilizes recent developments i
 n biocatalysis\, retrosynthesis and cheminformatics\, and aims toward a mo
 re efficient\, selective and environmentally favorable synthesis of compou
 nds through the inclusion of biocatalytic transformations. She furthermore
  works on developing new machine learning methods for molecular and reacti
 on property predictions.\nVisit LinkedIn\, Google Scholar or Github t
 o connect or learn more.
LOCATION:https://epfl.zoom.us/j/64473017589?pwd=Vmpnd1pleGhEb1hFb3kxUlNIUW
 JyQT09
STATUS:CONFIRMED
END:VEVENT
END:VCALENDAR