Benjamin Sanchez-Lengeling
Research Scientist, Google Deepmind
Abstract:Properties of the physical world come to life through our human senses, digitizing these senses allows us to catalog, search and design percepts. We have made remarkable progress in the domain of vision, hearing, but what about the rest? Olfaction is a chemical sense, and in this talk, I will present our recently published work on how we built a digital representation of olfaction for single compounds, which we call the Primary Odor Map and can be used in a variety of olfactory tasks. To validate our representation, we selected a set of 400 novel and diverse molecules with no known recorded scent and had them rated by a panel of trained humans. We compared our predicted olfactory profiles with the panel consensus response and found that our machine learning model outperforms any single human in the panel. The story will touch on deep learning, molecular representations, compound discovery pipelines, training humans to reliably rate odor percepts, as well as diverse applications of a principal odor map.
Speaker Bio: Benjamin Sanchez-Lengeling is a researcher at Google DeepMind, solving chemical problems leveraging data-driven techniques. Ben designs, builds, and evaluates computational tools that enable molecular discoveries, covering small molecules, polymers, chemical mixtures, and proteins. Striving to bring computational predictions into the lab by designing experimental validation with collaborators, prioritizing the interpretation of our discoveries, and making research clear and approachable. Ben graduated with a Ph.D. in Chemistry and Chemical Biology and a secondary field in Computational Science & Engineering from Harvard University under the supervision of Alán Aspuru Guzik. Besides research, he is also passionate about science education and divulgation. He is one of thefounders and organizers of a STEM-education NGO Clubes de Ciencia Mexico and a LatinX-centered AI conference RIIAA.
Microsoft Teams meeting
Join on your computer, mobile app or room device
Click here to join the meeting
Meeting ID: 214 767 653 586
Passcode: jAZajt
Download Teams | Join on the web
Or call in (audio only)
+1 647-794-1609,,970944516# Canada, Toronto
Phone Conference ID: 970 944 516#