Our mission is to increase the rate of scientific discovery with Google technologies, including machine learning, iterative prediction/experimentation in large combinatorial spaces, and large scale analysis and computation. We believe these will enable more effective high throughput research in many domains.
Using Google's unique expertise, technology and scale, we collaborate with world-class institutions on challenges with large scientific and humanitarian benefit, working closely with leading scientists who have deep domain expertise and proven experimental infrastructure.
Here's a small sample of our partners.
In Cell Screening:
In Cell State:
In Drug Discovery:
In Energy:
In Materials Science:
In Quantum Chemistry:
Our projects span many fields of science, including biology, chemistry, materials science, and physics. We actively investigate new scientific problems in areas we haven't worked in yet, and maintain a flexible scope on a wide range of potential projects.
We use machine learning to scalably interpret biological images. Our approaches offer many benefits including non-invasively imputing deep cell measures on transmission microscopy images (without stains), and discovering new phenotypes and common mechanism of actions across large cell screens.
We use several machine learning approaches to represent molecules and predict their interactions. We’ve tackled a range of problems using in silico screening of molecules for affinity to specific targets, a useful step in discovering new drugs.
While there exist many computational approaches to accurately simulating molecular or material properties, often these simulations are too expensive to discover molecules or do materials designs at scale. We are using machine learning to predict the outcome of expensive simulations, allowing us to short circuit computationally difficult problems.
We use design of experiment techniques and machine learning to explore large combinatorial spaces with feedback from physical experiments. These problems range from designing a robotic protocol for differentiating stem cells, to determining the right mix of atoms to create a new functional material, to optimizing settings for a plasma fusion machine.
Achievement of Sustained Net Plasma Heating in a Fusion Experiment with the Optometrist Algorithm
E.A. Baltz, E. Trask, M. Binderbauer, M. Dikovsky, H. Gota, R. Mendoza, J.C. Platt, P.F. Riley
Scientific Reports, vol. 7 (2017), pp. 6425
Improving Phenotypic Measurements in High-Content Imaging Screens
D. Mike Ando, Cory McLean, Marc Berndl
bioRxiv (2017)
Neural Message Passing for Quantum Chemistry
Justin Gilmer, Samuel S. Schoenholz, Patrick F. Riley, Oriol Vinyals, George E. Dahl
ICML (2017)
Prediction errors of molecular machine learning models lower than hybrid DFT error
Felix Faber, Luke Hutchinson, Huang Bing, Justin Gilmer, Sam Schoenholz, George Dahl, Oriol Vinyals, Steven Kearnes, Patrick Riley, Anatole von Lilienfeld
Journal of Chemical Theory and Computation (2017)
Rapid Photovoltaic Device Characterization through Bayesian Parameter Estimation
R.E. Brandt, Rachel Kurchin, Vera Steinmann, Daniil Kitchaev, Chris Roat, Sergiu Levcenco, Gerbrand Ceder, Thomas Unold, Tonio Buonassisi
Joule, vol. 1 (2017), pp. 843-856
Varun Gulshan, Lily Peng, Marc Coram, Martin C Stumpe, Derek Wu, Arunachalam Narayanaswamy, Subhashini Venugopalan, Kasumi Widner, Tom Madams, Jorge Cuadros, Ramasamy Kim, Rajiv Raman, Philip Q Nelson, Jessica Mega, Dale Webster
JAMA (2016)
Molecular graph convolutions: moving beyond fingerprints
Steven Kearnes, Kevin McCloskey, Marc Berndl, Vijay Pande, Patrick Riley
Journal of Computer-Aided Molecular Design (2016), pp. 1-14
Massively Multitask Networks for Drug Discovery
Bharath Ramsundar, Steven Kearnes, Patrick Riley, Dale Webster, David Konerding, Vijay Pande
arXiv:1502.02072 [stat.ML] (2015)