Microsoft Research

20 minute research talks + Q&A panel 0:59 Debadeepta Dey: Efficient Forward Neural Architecture Search 15:45 Subho Mukherjee: Resource-efficient Learning for Neural Language Understanding at Scale

Neural Architecture Search (NAS) is a very promising but still young field.

This is a short presentation introducing the open source project "coax". See more at microsoft.com/en-us/research/project/coax-rl

This video shows what is happening inside the lithium niobate media when a data page hologram is written.

This video shows how holographic storage works, using green light to write data as a persistent hologram inside an optical crystal.

Listening to music is a joyful way to relax. Many people enjoy playing music in the background in their daily life, even while working.

Audio-based wildlife monitoring is an important method for studying animal habitations and for the conservation of animal species and ecosystems.

When does a machine “understand” an image? One definition is when it can generate a novel caption that summarizes the salient content within an image.

The increasing gap between the growth of datacenter traffic and electrical switch capacity is expected to worsen due to the slowdown of Moore’s law, motivating the need for a new switching

The difficulty of building large data centers in dense metro areas is pushing big cloud providers towards a different approach to scaling: multiple smaller data centers within tens of kilometers of

The amazing growth of the cloud has also led to the emergence of inflection points that may hinder future cloud scaling.

The difficulty of building large data centers in dense metro areas is pushing big cloud providers towards a different approach to scaling: multiple smaller data centers within tens of kilometers of

Episode 005 | September 08, 2020 Ensuring security and privacy of data, both personal and institutional, is of paramount importance in today’s world where data itself is a highly precious commodity.

We present a technique to model wave-based sound propagation to complement visual animation in fully dynamic scenes.

Since the naming of the field in 1956, AI has been dominated first by symbolic rule-based models, then early-generation neural (or “connectionist”) models, then probabilistic models operating over

Since the naming of the field in 1956, AI has been dominated first by symbolic rule-based models, then early-generation neural (or “connectionist”) models, then probabilistic models operating over

Learn how to export your images and animations and save them for use in other applications. More at aka.ms/expressivepixels

Learn how using multiple layers can help you create more advance animations. More at aka.ms/expressivepixels

Learn how to create and save small fragments of an image for use in other images and animations. More at aka.ms/expressivepixels

Learn about the different types of animation galleries within Expressive Pixels and how you can share, download and display animations. More at aka.ms/expressivepixels

Learn how to make your image come to life by adding frames and transitions and setting the speed of your animation. More at aka.ms/expressivepixels

Learn to create an image in Expressive Pixels. We will show you the tools to create your masterpiece and save it in your Animation Gallery. More at aka.ms/expressivepixels

Learn what you need to get started with Expressive Pixels, including downloading the app, connecting your device and previewing the menu options. More at aka.ms/expressivepixels

Hear about the concept behind Expressive Pixels and the inclusive experience where people of all abilities to express, connect and share. More at aka.ms/expressivepixels

See how to import images and animations from your files into Expressive Pixels to help jump start the creation process. More at aka.ms/expressivepixels

See how expressivity and eye-gaze music work together in harmony in this Galactic Bell Star music demo. More at microsoft.com/en-us/research/product/expressive-pixels

We present the first wave-based interactive system for practical rendering of global sound propagation effects including diffraction, in complex game and VR scenes with moving directional sources

Dan Bohus, Senior Principal Research, provides an overview of Platform for Situated Intelligence: an open-source framework that significantly accelerates development and research on multimodal

Programming critical systems with proofs, a long-standing goal of computer science, is beginning to come within reach of modern programming languages and proof assistants.

Deep neural networks have been trained to interpret images and text at increasingly high levels of accuracy.