BEGIN:VCALENDAR
VERSION:2.0
PRODID:Linklings LLC
BEGIN:VTIMEZONE
TZID:Europe/Stockholm
X-LIC-LOCATION:Europe/Stockholm
BEGIN:DAYLIGHT
TZOFFSETFROM:+0100
TZOFFSETTO:+0200
TZNAME:CEST
DTSTART:19700308T020000
RRULE:FREQ=YEARLY;BYMONTH=3;BYDAY=-1SU
END:DAYLIGHT
BEGIN:STANDARD
TZOFFSETFROM:+0200
TZOFFSETTO:+0100
TZNAME:CET
DTSTART:19701101T020000
RRULE:FREQ=YEARLY;BYMONTH=10;BYDAY=-1SU
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
DTSTAMP:20190719T085745Z
LOCATION:HG E 1.2
DTSTART;TZID=Europe/Stockholm:20190614T113000
DTEND;TZID=Europe/Stockholm:20190614T120000
UID:submissions.pasc-conference.org_PASC19_sess127_msa157@linklings.com
SUMMARY:IBM RXN for Chemistry: Predicting Chemical Reactions using the Mol
 ecular Transformer
DESCRIPTION:Minisymposium\nComputer Science and Applied Mathematics, Chemi
 stry and Materials\n\nIBM RXN for Chemistry: Predicting Chemical Reactions
  using the Molecular Transformer\n\nSchwaller, Laino, Gaudin, Bolgar, Beka
 s...\n\nOrganic synthesis – making complex molecules out of simpler 
 building blocks – is a crucial part in the production and therefore,
  also in the discovery of novel molecules and materials. One necessary yet
  unsolved step in synthesis planning is solving the forward problem: given
  reactants and reagents, predict the products. We treat chemical reaction 
 prediction as a sequence-2-sequence problem, similar to neural machine tra
 nslation, and learn to transform the SMILES strings of the reactants-reage
 nts into the ones of the products. We show that a multi-head attention Mol
 ecular Transformer model outperforms all algorithms in the literature, ach
 ieving a top-1 accuracy above 90% on a common benchmark dataset. Our algor
 ithm requires no handcrafted rules, and accurately predicts subtle chemica
 l transformations. Crucially, our model can accurately estimate its own un
 certainty. Furthermore, we show that model is able to handle inputs withou
 t reactant-reagent split and including stereochemistry, which makes our me
 thod universally applicable across existing datasets. A fully-trained mode
 l is freely available on the IBM RXN for Chemistry platform (https://rxn.r
 es.ibm.com). Using this platform, chemists can simply draw reactants and r
 eagents and predict the most likely outcomes of the reaction.
END:VEVENT
END:VCALENDAR