Google Brain Team

Research Freedom

Google Brain team members set their own research agenda, with the team as a whole maintaining a portfolio of projects across different time horizons and levels of risk.

Google Scale

As part of Google and Alphabet, the team has resources and access to projects impossible to find elsewhere. Our broad and fundamental research goals allow us to actively collaborate with, and contribute uniquely to, many other teams across Alphabet who deploy our cutting edge technology into products.

Open Culture

We believe that openly disseminating research is critical to a healthy exchange of ideas, leading to rapid progress in the field. As such, we publish our research regularly at top academic conferences and release our tools, such as TensorFlow, as open source projects.

Working on the Brain Team

• Looking Back on 2017
• Meet a few of our machine learning makers
• Colin Raffel's Year at Brain
• Google Brain Residency Program — 7 months in and looking ahead
• Google Brain Residency Program — One Year Later
• Google Brain Team's Approach to Research
• Looking Back on 2016
• Reddit AMA 2016
• Reddit AMA 2017
• The Great A.I. Awakening

Papers Accepted to NIPS, 2017

• A Meta-Learning Perspective on Cold-Start Recommendations for Items
• AdaGAN: Boosting Generative Models
• Affine-Invariant Online Optimization
• Approximation and Convergence Properties of Generative Adversarial Learning
• Attention is All You Need
• Avoiding discrimination through causal reasoning
• Bridging the Gap Between Value and Policy Based RL
• Dynamic Routing between Capsules
• Filtering Variational Objectives
• Interpolated Policy Gradient: Merging On-Policy and Off-Policy Gradient Estimation for Deep Reinforcement Learning
• Investigating the learning dynamics of deep neural networks using random matrix theory
• Learning Hierarchical Information Flow with Recurrent Neural Modules
• Modulating early visual processing by language
• Multi-Armed Bandits with Metric Movement Costs
• Nonlinear random matrix theory for deep learning
• On Blackbox Backpropagation and Jacobian Sensing
• PASS-GLM: polynomial approximate sufficient statistics for scalable Bayesian GLM inference
• REBAR: Low-variance, unbiased gradient estimates for discrete latent variable models
• Reducing Reparameterization Gradient Variance
• SGD learns the conjugate class of the network
• SVCCA: Singular Vector Canonical Correlation Analysis for Deep Understanding and Improvement
• The Unreasonable Effectiveness of Random Orthogonal Embeddings
• Value Prediction Network

Papers Accepted to CoRL, 2017

• End-to-End Learning of Semantic Grasping
• Learning Robotic Manipulation of Granular Media
• Manifold Regularization for Kernelized LSTD
• Trust-PCL: An Off-Policy Trust Region Method for Continuous Control