Microsoft Research

Batch normalization (BatchNorm) is a widely adopted technique that enables faster and more stable training of deep neural networks.

Modern applications must collect and store massive amounts of data. Cloud storage offers these applications simplicity: the abstraction of a failure-free, perfectly scalable black-box.

We study high-dimensional estimation in a setting where an adversary is allowed to arbitrarily corrupt an $\varepsilon$-fraction of the samples.

Episode 75 | May 8, 2019 Dr.

Introducing TORC—TOuch Rigid Controller—a new rigid haptic controller prototype that renders virtual object characteristics and behaviors such as texture and compliance.

Episode 74 |

Making sense of large and complex data requires methods that integrate human judgment and domain expertise with modern data processing systems.

Project Florence is a speculative glimpse into our future, where both our natural and digital worlds could co-exist in harmony through enhanced communication.

Episode 73 | April 24, 2019 In the world of relational databases, structured query language, or SQL, has long been King of the Queries, primarily because of its ubiquity and unparalleled performance.

Our greatest opportunity for problem-solving comes not from humans alone or from Artificial Intelligence (AI) alone, but by combining them in distributed networks.

Online programming communities, like Stack Overflow, have norms that are not obvious nor inclusive to the 50 million programmers visiting monthly.

Many of the recent cryptocurrency bugs and exploits are due to flaws or weaknesses of the underlying blockchain programming languages or virtual machines.

Robots out in the world today work for people but not with people.

Many status-quo interfaces for tablets with pen + touch input capabilities force users to reach for device-centric UI widgets at fixed locations, rather than sensing and adapting to the user-centric

Episode 72 |

Private or confidential information is used in several applications, including not just cryptographic implementations but also machine-learning algorithms, databases, and parsers.

Users often fall for phishing emails, reuse simple passwords, and fail to effectively utilize "provably" secure systems.

Learning in Data Scarce Visual and Multimodal Applications Using Vectorized and Composable Representations The vision and learning group at CVT (Center for Vision Technologies) at SRI has developed

Episode 71 | April 10, 2019 Dr.

Why do we see so little reuse of web UI across organizations?

Data lakes (e.g., enterprise data catalogs and Open Data portals) are data dumps if users cannot find and utilize the data in them.

High-Fidelity Simulations in AirSim enables custom solutions for automation.

High-Fidelity Simulations in AirSim allows detailed modeling of various characteristics of wind-turbines.

Episode 70 |

AI researchers are striving to create intelligent machines that complement human reasoning and enrich human experiences and capabilities.

AI researchers are striving to create intelligent machines that complement human reasoning and enrich human experiences and capabilities.

AI researchers are striving to create intelligent machines that complement human reasoning and enrich human experiences and capabilities.

AI researchers are striving to create intelligent machines that complement human reasoning and enrich human experiences and capabilities.

AI researchers are striving to create intelligent machines that complement human reasoning and enrich human experiences and capabilities.

AI researchers are striving to create intelligent machines that complement human reasoning and enrich human experiences and capabilities.