Papers of the day   All papers

Machine Learning for Scent: Learning Generalizable Perceptual Representations of Small Molecules

Comments

Google AI: Our research team is using graph neural networks to predict the olfactory properties of molecules, expanding our understanding of smell & odor, with potential applications ranging from odorant synthesis to scent digitization. Learn more at https://goo.gle/2BvfUM6 https://t.co/3VOV75hrv2

21 replies, 1504 likes


hardmaru 😷: Machine Learning for Scent: Learning Generalizable Perceptual Representations of Small Molecules https://arxiv.org/abs/1910.10685 👃🏻

1 replies, 78 likes


DataScienceNigeria: What is SMELLING @GoogleAI? Artificial Intelligence can use graph neural networks to identify molecules & predict smells by understanding the relationship between a molecule’s structure & its odour using structure-odor relationship (QSOR) modeling. Read https://arxiv.org/pdf/1910.10685.pdf https://t.co/eta9mSNSDJ

1 replies, 78 likes


Alan Aspuru-Guzik: Happy to be part of this research on smell prediction with Alex Wiltschko @beangoben and Jennifer Wei. The GIF for the GNN are epic in this blog post. @UofT @VectorInst @CIFAR_News @chemuoft @UofTCompSci Paper here https://arxiv.org/abs/1910.10685

2 replies, 67 likes


Alan Aspuru-Guzik: Want to know about QSOR? Quantitative Structure -Odor relationships? Check out our new paper with @GoogleAI @beangoben Jennifer Wei Brian Lee and Alex Wiltschko ! A Google-led paper that explores the boundaries of what is possible ! Honored to be involved https://arxiv.org/pdf/1910.10685.pdf

3 replies, 57 likes


WIRED Science: The science of smell lags behind many other fields. But researchers at Google Brain have a new breakthrough: they trained machine-learning algorithms to predict molecules’ smell based on their structures. This could help answer long-standing questions. https://trib.al/OLMdHTh

2 replies, 56 likes


bayo adekanmbi: From facial to olfactory identity! Aligned with the sensory perception by IBM & Symrise fragrance. Awesome to see “smell” treated like a multi-label classification & novel concept of “odor embedding,” similar to RGB’s treatment as an embedding for vision. I can SMELL the future!

0 replies, 54 likes


Eric Topol: —OK, but how it will it smell? —Let me ask the machine ;-) https://arxiv.org/pdf/1910.10685.pdf @googleAI #AI @UofT @UofTCompSci @VectorInst https://ai.googleblog.com/2019/10/learning-to-smell-using-deep-learning.html by Alexander Wiltschko and colleagues https://t.co/bvhBTfRmX7

1 replies, 19 likes


Daisuke Okanohara: Graph NN can predict odor from a molecule's structure better than existing methods, and provide a meaningful odor space where similar scents are clustered. Similar to visual and auditory senses, NN can (at least) simulate smell sense. https://arxiv.org/abs/1910.10685

0 replies, 10 likes


Joel Mainland: 2/3 A new preprint uses graph neural networks to turn molecular structure into vectors. The representation they learned from GoodScents and Leffingwell databases performed well on a very different task--the DREAM Olfaction Challenge. https://arxiv.org/abs/1910.10685

1 replies, 8 likes


Alex Zhavoronkov: This was my dream for a very long time but unfortunately, we need to focus on drugs. Glad to see Ben and @A_Aspuru_Guzik explore this space first. Kudos to the luchadores, great paper! https://arxiv.org/abs/1910.10685

0 replies, 4 likes


Machine Learning: Science and Technology: Great new work from the group of @MLSTjournal Editorial Board member @A_Aspuru_Guzik @GoogleAI

0 replies, 3 likes


Brundage Bot: Machine Learning for Scent: Learning Generalizable Perceptual Representations of Small Molecules. Benjamin Sanchez-Lengeling, Jennifer N. Wei, Brian K. Lee, Richard C. Gerkin, Alán Aspuru-Guzik, and Alexander B. Wiltschko http://arxiv.org/abs/1910.10685

1 replies, 1 likes


Content

Found on Oct 24 2019 at https://arxiv.org/pdf/1910.10685.pdf

PDF content of a computer science paper: Machine Learning for Scent: Learning Generalizable Perceptual Representations of Small Molecules