Hi all,
Tomorrow Rafa will give group meeting. The title and abstract are below.
Next week there will be no group meeting because of convocation. When we
come back on June 2nd, we'll start off our "Great / Crazy Ideas" series
with Borja and Steven. The full first round of talks is in the schedule
<https://docs.google.com/spreadsheets/d/1LioqdopHr5ZAmn0GaCQwLUTYoU4ZJuvfrMYE6zCVhqI/edit#gid=0>
now,
and the code I used to generate it is here after Rafa's abstract. If anyone
has a conflict with their assigned time we'll skip that week and push the
whole thing back.
Finally, starting on June 9th group meeting will start at 3:30 rather than
2:30.
Best,
Ian
-----------------
Speaker: Rafa Gómez-Bombarelli
Title: Deep learning chemical space
Abstract: Virtual screening is increasingly proven as a tool to test new
molecules for a given application. Through simulation we can gauge whether
a molecule will be a promising candidate in an automatic and robust way. A
large remaining challenge, however, is how to perform optimizations over a
discrete space of size at least 10^60.
Despite the size of chemical space, or perhaps precisely because of
it, coming up with novel, stable, makeable molecules is not trivial.
First-principles approaches to generating new molecules fail to capture
the intuition embedded in the approx 100 million existing molecules.
Tomorrow I will report our progress towards developing an autoencoder that
allows us to project molecular space into a continuous, differentiable
representation.
David Duvenaud (now at the Adams group at SEAS and very soon at University
of Toronto) has contributed loads to this work and will be joining us
tomorrow for the talk. It'll be a great opportunity to bring up everything
you always wanted to know about deep learning but were afraid to ask.
-----------------
import random
people = ["Doran", "David G", "Rafa", "James",
"Christoph", "Steven",
"Salvatore", "Borja",
"Jorge", "Dmitrij", "Semion", "Dmitry",
"Sam", "Joey", "Adrian", "Ian",
"Thomas", "Jonathan",
"Ben", "Jacob", "Nico", "Teresa",
"Stephanie", "Jennifer"]
random.seed(42)
random.shuffle(people)
for i in xrange(len(people) / 2):
print "%s, %s" % (people[2 * i], people[2 * i + 1])
Show replies by date