In joint work with Soumik Pal, we study natural mixing processes where cards (or dominoes or mahjong tiles) are 'smushed' around on a table with two hands. How long should mixing continue. If things…

For much of its history, AI research has aimed toward building intelligent machines independently of their interactions with people. As the world of computing has evolved, and systems pervade ever…

Given n uniform points on the surface of a two-dimensional sphere, how can we partition the sphere fairly among them? "Fairly" means that each region has the same area. It turns out that if…

We prove that any random distribution satisfying conformal invariance and a form of domain Markov property and having a finite moment condition must be the Gaussian free field. We also present some…

A secure multiparty computation protocol allows a set of mutually distrusting parties to compute a joint function of their private inputs without leaking anything apart from the output of the…

The talk is divided into two sections. In the first section, I will talk about behavioral and physiological responses associated with interest, curiosity, and novelty in image viewing. We recorded…

Rapid advances in sequencing technology have led to the generation of genome-scale DNA sequencing data for more than 2 million individuals worldwide. These data represent incredibly powerful…

Optimal matching problems are random variational problems widely investigated in the mathematics and physics literature. We discuss here the optimal matching problem of an empirical measure on a…

Modern data often consists of feature vectors with a large number of features. High-dimensional geometry and Linear Algebra (Singular Value Decomposition) are two of the crucial areas which form the…

Modern data often consists of feature vectors with a large number of features. High-dimensional geometry and Linear Algebra (Singular Value Decomposition) are two of the crucial areas which form the…

Modern data often consists of feature vectors with a large number of features. High-dimensional geometry and Linear Algebra (Singular Value Decomposition) are two of the crucial areas which form the…

Modern data often consists of feature vectors with a large number of features. High-dimensional geometry and Linear Algebra (Singular Value Decomposition) are two of the crucial areas which form the…

Zachary Horvitz is a rising junior at Brown University, concentrating in computer science and anthropology. He presents at Microsoft Research background on issues around "broken window…

Rebeca is designed for modeling and formal verification of asynchronous and reactive systems in 2001, and is supported by a model-checking tool, Afra. Followed by that, an actor-based family of…

The world of fashion is rapidly adopting technology on runways around the world, bringing a spotlight to innovative new designs and experiments, also known as Haute Tech Couture. However, designers…

This talk will explore how to use social theory to understand problems of social order and its relationship to an era of Big Data. I take as a starting-point of the talk the neglected late work of…

We propose a new class of post-quantum digital signature schemes that: (a) derive their security entirely from the security of symmetric-key primitives, believed to be quantum-secure, and (b) have…

We will be presenting a series of online tutorial sessions to introduce and provide examples of the work that is possible in the Catapult Academic environment.

I will highlight some methodologies and examples of finding causal mutations in both medical genomics and comparative genomics contexts.

The increasing ubiquity of devices capable of capturing videos has led to an explosion in the amount of recorded video content. Instead of "eyeballing" the videos for potentially useful…

Longitudinal studies using perturbations enable us to study the resilience of the human microbiome. These are particularly informative in the case of antibiotic courses and colonic clean-out. The…

In this talk, I will discuss intertwining importance and connections of three principles of data science. The three principles will be demonstrated in the context of two neuroscience projects and…

Unlike passive humor appreciation, the neural correlates of real-time humor creation have been unexplored. As a case study for creativity, humor creation uniquely affords a reliable assessment of a…

Noisy data, non-convex objectives, model misspecification, and numerical instability can all cause undesired behaviors in machine learning systems. As a result, detecting actual implementation errors…

Robots manipulate with super-human speed and dexterity on factory floors. But yet they fail even under moderate amounts of clutter or uncertainty. However, human teleoperators perform remarkable acts…

We will be presenting a series of online tutorial sessions to introduce and provide examples of the work that is possible in the Catapult Academic environment.

Modern society is built on software platforms that encompass a great deal of our lives. While this is well known, software is invented by people and this comes at considerable cost. Notably,…

This talk will provide a brief overview of digital mental health, which has used common technologies such as web and mobile apps to support behaviors aimed at reducing mental health symptoms. While…

We will be presenting a series of online tutorial sessions to introduce and provide examples of the work that is possible in the Catapult Academic environment.

Christopher Bishop has a fireside chat with Harry Shum, executive vice president of Microsoft's Artificial Intelligence (AI) and Research group, discussing Microsoft Research's advancements in…

The future will be defined by autonomous computer systems that are tightly integrated with the environment, also known as Cyber-Physical systems (CPS). Resilience and security become extremely…

We will be presenting a series of online tutorial sessions to introduce and provide examples of the work that is possible in the Catapult Academic environment.

I will present the main algorithms to achieve robust, 6-DOF, state estimation for mobile robots using passive sensing. Since cameras alone are not robust enough to high-speed motion and high-dynamic…

In April of 2016, an early VR title called Audioshield was released on Steam, the largest PC gaming community in the world. According to Steam stats, gamers have played more than 693,000 hours of…

In this work, we propose to apply trust region optimization to deep reinforcement learning using a recently proposed Kronecker-factored approximation to the curvature. We extend the framework of…

Many of the major security problems are caused by small programming errors, which could be avoided by automating menial programming tasks through program synthesis. However, program synthesis is still…

Chanuki Seresinhe and Suzy Moat of The Turing Institute share their experience using Azure to measure aspects of the world that humans have been aware of, but have no data on.

An introduction to ethnography for Artificial Intelligence as well as conversational analysis and its relevance to AI.

Merve Alanyali and Suzy Moat of The Alan Turing Institute explain how they use Azure to analyze and use online data to understand human behavior. Azure helped them analyze over 25 million images in a…

This demo shows our work on a mobile robot that gives directions to visitors. Currently, this robot is navigating Microsoft Building 99, leading people, escorting and interacting with visitors and…

Microsoft Translator live enables users to hold translated conversations across two or more languages, with up to 100 participants participating at the same time using PowerPoint, iOS, Android,…

Visual search, AKA search by image, is a new way of searching for information using an image or part of an image as the query. Similar to text search, which connects keyword queries to knowledge on…

Microsoft Pix helps every photographer take better pictures. Because it incorporates AI behind the lens, it can tweak settings, select the best shots, and enhance them on the fly. It's designed to…

Building machine learning (ML) models is an involved process requiring ML experts, engineers, and labelers. The demand of models for common-sense tasks far exceeds the available "teachers"…

This demo demonstrates several applications of Microsoft's recent work in artistic style transfer for images and videos. One technology, called StyleBank, provides an explicit representation for…

When traveling to China it's best to know at least a bit of the language. The mobile app called Learn Chinese can help travelers enjoy a better journey. Learn Chinese teaches in an interactive way, by…

Project Catapult connects FPGAs together through a network to create a hyperscale, reconfigurable accelerator fabric. See how to use the Project Catapult cluster at the Texas Advanced Computing Center…

DeepFindSearching within web documents on mobile devices is difficult and unnatural: ctrl-f searches only for exact matches, and it's hard to see the search results. DeepFind takes a step toward…

From ferry schedules to dinner reservations, Cortana is the digital assistant designed to help people get things done. Cortana will eventually be everywhere people need assistance—on the phone, PC,…

Two of the most important components of speech recognition systems are the acoustic model and the language model. Those models behind Microsoft's speech recognition engine have been optimized for…

This demo shows how Custom Vision Service can be applied to many AI vision applications. For example, if a client needs to build a custom image classifier, they can submit a few images of objects, and…

As we move into the world of messaging apps, bots and botification of content, users are starting to move from keyword searches to relying on bots and assistants for their information seeking needs.…

Project InnerEye is a new AI product targeted at improving the productivity of oncologists, radiologists, and surgeons when working with radiological images. The project's main focus is in the…

Project Malmo is an open source AI experimentation platform that supports fundamental AI research. With the platform, Microsoft provides an experimentation environment in which promising approaches…

Knowing which part of a program processes which parts of an input can reveal the structure of the input as well as the structure of the program. In a URL "http://www.example.com/path/", for…

The logics required to support program verification go much beyond the class of decidable logics available today, and beyond the quantifier-free logics supported by SMT solvers today. In particular,…

This talk will give an overview of our research projects. Cloudberry is a middleware solution to support interactive exploration and visualization on large data sets ("responsive big…

JavaScript is becoming one of the most popular languages, yet it is known for its suboptimal design. To effectively use JavaScript despite its design flaws, lightweight static analyses, implemented in…

Machine learning is increasingly being adopted by various domains: governments, credit, recruiting, advertising, and many others. Fairness and equality are critical aspects, especially in light of…

When attacking the problem of Autonomous Driving, one must take into account strict computational constraints, posed by the desired low cost of sensors and processors, and by the required real-time…

Deep neural networks and other machine learning models have been transformative for building intelligent systems capable of visual recognition, speech and language understanding. While recent advances…

The number one United Nations Sustainable Development Goal is to eliminate poverty, leaving nobody behind. Researchers in the United Kingdom are harnessing the large-scale data-processing power of…

The ongoing explosion of spatiotemporal tracking data has now made it possible to analyze and model fine-grained behaviors in a wide range of domains. For instance, tracking data is now being…

Deep neural networks (DNNs) have led to significant improvements to the accuracy of machine-learning applications. For many problems, such as object classification and object detection, DNNs have led…

The characteristic durations of speech sounds vary considerably across languages, dialects, and even speakers of the same dialect. Analyses of increasingly large speech corpora reveal, however, that…

Predictive models are finding an increasing number of applications in many industries. As a result, a practical means for trading-off the cost of deploying a model versus its effectiveness is needed.…

Thus far, capturing human body motion has only been possible with precisely-positioned sensors and cameras in well-calibrated studio environments. This constraint has fundamentally restricted the…

Overview of important methods and techniques applied in user studies from the perspective of academic research, and also from the perspective of product development. The talk is ended with two…

An overview of social networks and social network analysis.

An overview of how Automatic Speech Recognition systems work and some of the challenges.

Defining pragmatics has not been easy. We discuss the major themes that have been dealt with in the field: 1) The macro perspective: Speech Act Theory 2) The micro perspective: Cooperative Principles…

An overview of some aspects of multilingualism for those that are new to the field. This talk introduces important issues and concepts to have in mind when working on your research projects.

How to do natural language processing on social media data.

Part 2 of Alan W Black's Spoken Dialogue Systems talk based on lectures he gave at Carnegie Mellon University.

Tutorial on what actual dialog systems are like and the problems we have with them, based on lectures Alan gave at Carnegie Mellon University.

Breakthroughs in Artificial Intelligence (AI) have typically shown that AI systems are good at solving specific tasks that have a well-defined goal, such as Speech Recognition, Image Captioning, Games…

Intra-tumor heterogeneity is one of the key confounding factors in deciphering tumor evolution. Malignant cells even if coming from a single tumor will have variations in their gene expression, copy…

In Bayesian analysis, the posterior follows from the data and a choice of a prior and a likelihood. These choices may be somewhat subjective and reasonably vary over some range. Thus, we wish to…

In recent years, there has been an explosion of interest in and research on the use of mental health technologies that aim to treat and manage mental health problems. From smartphone apps to bots,…

The primary, long-standing goal for Augmented Reality (AR) is bringing the real and virtual together into a common space. To maintain the illusion that these two worlds coexist, however, requires that…

In this talk we will discuss our recent work on AI techniques that improve the quality of audio signals for both machine understanding and sensory perception. Our best models utilize…

Today, experiencing virtual reality (VR) is a cumbersome experience which either requires dedicated infrastructure like infrared cameras to track the headset and hand-motion controllers (e.g. Oculus…

Speech emotion recognition promises to play an important role in various fields such as healthcare, security, HCI. This talk examines various convolutional neural network architectures for recognizing…

Many real-world problems, such as network packet routing and the coordination of autonomous vehicles, are naturally modelled as cooperative multi-agent systems. In this talk, I overview some of the…

How can a machine learn from experience? Probabilistic modelling provides a mathematical framework for understanding what learning is, and has therefore emerged as one of the principal approaches for…

In this tutorial we discuss several recent advances in deep reinforcement learning involving policy gradient methods. These methods have shown significant success in a wide range of domains, including…

This session addresses what is human about artificial intelligence. The session will inspect this question from three very different perspectives. Alex Taylor will discuss human perspectives on AI;…

This demonstration shows an iRobot Create with a UMPC (a small Tablet PC with resistive digitizer) on it. The UMPC communicates with the robot via serial port. The UMPC is used to enter the story and…

This demonstration shows a robotics playset with two robots and toy trees on a PixelSense table which displays the grass on a park with a lake. The user's cards have a word on one side and a tag on…

This demonstration shows the use of cards that have a word on one side and a tag on the other side on a PixelSense table with the nouns "A banana", "A dragon", and…

A long-term goal of artificial intelligence research is to develop artificial assistants (AI agents) that can collaborate with and empower their users. This goal raises important questions, such as…

This demonstration shows a scene with a dragon and banana in the forest and the sentence: "The dragon ate a banana in the forest." The sentence has a Fontlings idle behavior. When the…

This demonstration shows an early version of Fontlings with the sentence "Fontlings! Are fun and alive!" and the behavior "Happy" applied to the first word, and then to…

This demonstration shows a child typing "Once upon a time a dragon flew over a beach" followed by period and getting an instant reward by seeing the corresponding animation.

This demonstration shows the sentence of demo1 being animated while the child enters a new sentence "Then it fell down" follow by period. As soon as the period is typed, an animation…

This demonstration shows a child navigating through the pages of a story (one sentence per page) and seeing the corresponding animation for each page.

This demonstration shows a child typing Later Michel ate a pizza in the town followed by a period. StoryBaker automatically finds out that there is no visual for the word pizza and opens a sketchpad…

This demonstration shows a child typing "Michel kicked the pizza" followed by a period. StoryBaker animates the sentence by using the pizza drawn previously with the same background (town),…

This demonstration shows a child creating a new actor by opening the sketch tool, selecting a man body, using a picture for the face, typing a name ("Michel" in this example), and sketching…

This demonstration shows a child selecting a sentence within an existing story on the device. The selected sentence is "One day, a rocket flew over a boy in the backyard." The sentence…

This demonstration shows a child selecting a sentence within an existing story on the device. The selected sentence is "One day, a rocket flew over a boy in the backyard." The sentence…

F* is a ML-like programming language aimed at program verification, allowing its users to write programs, specifications, and prove them correct in the same tool. It is a key component of the Everest…

New York City serves over one million public school students each year, yet relatively little is understood in terms of how students progress through the school system. In this talk we use…

Humans make hundreds of routine decisions daily. More often than not, the impact of our decisions depends on the decisions of others. As AI progresses, we are offloading more and more of these…

Computer vision is arguably one of the most challenging subfields of AI. To better address the key challenges, the vision research community has long been branched off from the general AI community…

We review existing cryptographic schemes based on the hardness of computing isogenies between supersingular isogenies, and present some attacks against them. In particular, we present new techniques…

Since the time of the ancients, Astronomers have been visual thinkers. It was also a hotbed of "data science" and "open science" before those were cool. This trifecta of visual,…

How can we explain the predictions of a black-box model? In this paper, we use influence functions — a classic technique from robust statistics — to trace a model's prediction through the learning…

Moving computation and services from the desktop and local back office to large, centralized, multi-tenant data centers ("The Cloud") has been the dominant trend of the last decade in…

Within integrated development environments (IDEs), such as Visual Studio and Eclipse, program analysis tools perform sophisticated analysis to identify defects in source code. These defects are then…

3:05 PM–3:35 PM | Michael Biercuk – The Role of Quantum Control in Quantum Computing 3:40 PM–4:10 PM | Matthias Troyer – Can AIs Dream of Schrodinger's Cat? 4:15 PM–4:45 PM | Bettina Heim

9:00 AM–9:30 AM | Andras Pal Gilyen – Optimizing Quantum Optimization Algorithms 9:30 AM–10:00 AM | David Poulin – Surprising Facts About Quantum Error Correction

1:05 PM–1:35 PM | Ronald De Wolf – Quantum SDP-Solvers: Better Upper and Lower Bounds 1:40 PM–2:10 PM | Jeongwan Haah – Magic State Distillation with Low Space Overhead and Optimal Asymptotic Input…

10:50 AM–11:20 AM | Xiaodi Wu – Quantum Query Complexity of Entropy Estimation 11:25 AM–11:55 AM | Stephanie Wehner – Quantum Internet: The Certifiable Road Ahead

Each year, Microsoft Research sponsors a semester-long class at leading design schools. Students are asked to form interdisciplinary teams of two to four students to design a user experience prototype…

The primary source of water for Bangalore is the Cauvery river, which is located about 100 kilometers from the city. Given that the monsoon is not always dependable, it is critical to maximize the…

Each year, Microsoft Research sponsors a semester-long class at leading design schools. Students are asked to form interdisciplinary teams of two to four students to design a user experience prototype…

Each year, Microsoft Research sponsors a semester-long class at leading design schools. Students are asked to form interdisciplinary teams of two to four students to design a user experience prototype…

Each year, Microsoft Research sponsors a semester-long class at leading design schools. Students are asked to form interdisciplinary teams of two to four students to design a user experience prototype…

Each year, Microsoft Research sponsors a semester-long class at leading design schools. Students are asked to form interdisciplinary teams of two to four students to design a user experience prototype…

Each year, Microsoft Research sponsors a semester-long class at leading design schools. Students are asked to form interdisciplinary teams of two to four students to design a user experience prototype…

Each year, Microsoft Research sponsors a semester-long class at leading design schools. Students are asked to form interdisciplinary teams of two to four students to design a user experience prototype…

In the future, every company will be using AI, which means that every company will need a secure infrastructure that addresses AI security concerns. At the same time, the domain of computer security…

In this talk, I will describe techniques for recognizing the high-level algorithmic idea of a program and its applications in feedback generation for introductory programming education. Both…

Sampling integers with Gaussian distribution is a fundamental problem that arises in almost every application of lattice cryptography, and it can be both time consuming and challenging to implement.…

Today, thousands of scientists and engineers are applying machine learning to an extraordinarily broad range of domains, and over the last five decades, researchers have created literally thousands of…

Advances in AI promise great benefit to people and organizations. However, as we push the science of AI forward, we need to consider potential downsides, unintended consequences and costly outcomes.…

In 1982, Richard Feynman first proposed using a "quantum computer" to simulate physical systems with exponential speed over conventional computers. Quantum algorithms can solve problems in…

The new wave of excitement about AI in recent years has been based on successes in perception tasks or on domains with limited and known dynamics. Because machines have achieved human parity in…

Social signals and emotions are fundamental to human interactions and influence memory, decision-making and wellbeing. As AI systems, in particular, intelligent agents, become more advanced, there is…

Microsoft Cognitive Toolkit (CNTK) is a production-grade, open-source, deep-learning library. In the spirit of democratizing AI tools, CNTK embraces fully open development, is available on GitHub, and…

Machine learning (ML) has demonstrated success in various domains such as web search, ads, computer vision, natural language processing (NLP), and more. These success stories have led to a big focus…

Voice-driven automated agents such as personal assistants are becoming increasingly popular. However, in a multilingual and multi-cultural country like India, deploying such agents to engage with…

As the volume of data goes up, the quality of machine learning models, predictions, and services will improve. Once models are trained, predictive cloud services can be built on them, but users who…

Human society is faced with an unprecedented challenge to mitigate and adapt to changing climates, ensure resilient water supplies, sustainably feed a population of 10 billion, and stem a catastrophic…

Recent research in recurrent neural models, combined with the availability of massive amounts of dialog data, have together spurred the development of a new generation of conversational systems. Where…

Over the last decade, algorithmic developments coupled with increased computation and data resources have led to advances in well-defined verticals of AI such as vision, speech recognition, natural…

As the interaction over the web has increased, incidents of aggression and related events like trolling, cyberbullying, flaming, hate speech, etc. too have increased manifold across the globe. While…

Effective coordination of post-disaster relief operations depends critically on the availability of reliable situational information, as well as on citizen participation in the operations. The advent…

Most of the AI systems today are driven by three key components (i) data (ii) common sense knowledge and (iii) powerful learning algorithms which can harness this data and knowledge to learn task…

Tuberculosis (TB) is a completely curable disease, yet over 1 million people die from TB every year. One of the major challenges is to ensure that patients complete a full six-month course of…

This talk grounds the experience of harassment sociologically drawing attention to the way that race and gender interact to shape who experiences gender harassment and how they respond to it. Gender…

There is widespread sentiment that it is not possible to effectively utilize fast gradient methods (e.g. Nesterov's acceleration, conjugate gradient, heavy ball) for the purposes of stochastic…

Benjamini and Schramm (2001) showed that distributional limits of finite planar graphs with uniformly bounded degrees are almost surely recurrent. The major tool in their proof is a lemma which…

Scientists have long explored the different sensory inputs to better understand how humans perceive the world and control their bodies. Many of the great discoveries about the human perceptual system…

Teaching machines to read, process and comprehend natural language documents and images is a coveted goal in modern AI. We see growing interest in machine reading comprehension (MRC) due to potential…

I'll present the incipient tactics engine for F*, a programming language aimed at verification with an SMT backend. In the quest to make both F* proofs faster and more reliable and the language itself…

Advances in AI technologies have important ramifications for the development of accessible technologies. These technologies can augment the capabilities of people with sensory disabilities, enabling…

Configuring large computing installations is a difficult problem -- there are many different subsystems involved, all with their own language, and many different people with an interest in overlapping…

This talk presents recent advances in speeding up the processing of genomic data, making use of reconfigurable data flow engines based on field programmable gate arrays. Our design involves the…

We suggest a method to construct a homomorphic encryption scheme for approximate arithmetic. It supports approximate addition and multiplication of encrypted messages, together with the rescaling…

Deep learning has achieved great successes in many application scenarios, such as image classification and autonomous driving. Many of these applications are security sensitive. In this talk, we will…

Fielding AI solutions in the open world requires systems to grapple with incompleteness and uncertainty. This session will address several promising areas of research in open world AI, including…

Quantum computing is in its infancy, but Microsoft's Krysta Svore and Nathan Wiebe talk about quantum techniques as applied to AI challenges. Quantum computing can leverage quantum effects, such as…

This session talks about Project Malmo, a unique AI experimentation platform based on the game Minecraft. Microsoft Research's Katja Hofmann and Evelyne Viegas talk about the genesis of the platform,…

For AI to achieve its potential, it requires access to large amounts of often sensitive data, which can pose threats to our privacy. This session, featuring Rich Caruana and Kristin Lauter of…

AI deep learning expert and University of Montreal Professor Yoshua Bengio talks about deep learning—what it is, how it got there, where it's going, and how you can learn more about it. He discusses…

Microsoft's Lucy Vanderwende and Bill Dolan and Ohio State's Alan Ritter talk about deep learning algorithms that help machines converse in lifelike ways, drawing from social media conversations. They…

Deep learning techniques are helping Microsoft researchers develop literate machines: those that can read for understanding, then write well about what they know. Kaheer Suleman and Jianfeng Gao…

AI for Earth is a new initiative from Microsoft designed to help the planet be more sustainable. This session, with Microsoft's Lucas Joppa and Nebojsa Jojic, focuses on one early project: a land…

In this session, Ivan Tarapov demos a prototype application of assistive AI for cancer treatment: helping a radiologist create a treatment plan. AI has the capability of identifying the tumor and…

The next step for search involves InfoBots that know how to find the answer you're looking for, not just a set of webpages that may—or may not—have those answers. In this session, Microsoft's Nilesh…

Understanding the land cover types and locations within specific regions enables effective environmental conservation. With sufficiently high spatial and temporal resolution, scientists and planners…

Project InnerEye is a new AI product targeted at improving the productivity of oncologists, radiologists and surgeons when working with radiological images. The project's main focus is in the…

This demo shows our work on a mobile robot that gives directions to visitors. Currently, this robot is navigating Microsoft Building 99, leading people, escorting and interacting with visitors and…

Microsoft Pix helps every photographer take better pictures. Because it incorporates AI behind the lens, it can tweak settings, select the best shots, and enhance them on the fly. It's designed to…

Building machine learning (ML) models is an involved process requiring ML experts, engineers, and labelers. The demand of models for common-sense tasks far exceeds the available "teachers"…

Maluuba's vision is to build literate machines. The research team has built deep learning models that can process written unstructured text and answer questions against it. The demo will showcase…

When traveling to China it's best to know at least a bit of the language. The mobile app called Learn Chinese can help travelers enjoy a better journey. Learn Chinese teaches in an interactive way, by…

DeepFindSearching within web documents on mobile devices is difficult and unnatural: ctrl-f searches only for exact matches, and it's hard to see the search results. DeepFind takes a step toward…

For more than 100 years, investigators have taken a keen interest in language and cognition of deaf learners. In many ways, their research has played a central role in arriving at our current…

Zo is a sophisticated machine conversationalist with the personality of a 22-year-old with #friendgoals. She hangs out on Kik and Facebook and is always interested in a casual conversation with her…

Project Malmo is an open source AI experimentation platform that supports fundamental AI research. With the platform, Microsoft provides an experimentation environment in which promising approaches…

Microsoft Translator live enables users to hold translated conversations across two or more languages, with up to 100 participants participating at the same time using PowerPoint, iOS, Android,…

This demo shows how InstaFact brings the information and intelligence of the Satori knowledge graph into Microsoft's productivity software. InstaFact can automatically complete factual information in…

As we move into the world of messaging apps, bots and botification of content, users are starting to move from keyword searches to relying on bots and assistants for their information seeking needs.…

This demonstration uses Softbank's Pepper robot as testbed hardware to show a set of human-collaboration activities based on Microsoft Cognitive Services and other Microsoft Research technologies. As…

This demo demonstrates several applications of Microsoft's recent work in artistic style transfer for images and videos. One technology, called StyleBank, provides an explicit representation for…

Two of the most important components of speech recognition systems are the acoustic model and the language model. Those models behind Microsoft's speech recognition engine have been optimized for…

This demo shows how Custom Vision Service can be applied to many AI vision applications. For example, if a client needs to build a custom image classifier, they can submit a few images of objects, and…

Visual search, AKA search by image, is a new way of searching for information using an image or part of an image as the query. Similar to text search, which connects keyword queries to knowledge on…

Today's most popular applications are deployed as massive-scale distributed systems in the datacenter. Keeping data consistent and available despite server failures and concurrent updates is a…

As quantum computing technology becomes increasingly sophisticated, the techniques required to calibrate and certify device performance are becoming commensurately sophisticated. In this talk, I will…

Many envision that AI (artificial intelligence) will usher in the next iteration of technology revolution, where humans and machines will work side-by-side to augment, enhance, or accelerate our…

Can higher-order functional programs solve more problems than first-order programs? Answer: NO, since both program classes are Turing complete. The reason is that higher-order values can be simulated…

The Voronoi Treemap is a space-filling treemap technique that relaxes the constraints of rectangular nodes. Its organic shapes can maintain a near one-to-one aspect ratio, are flexible with their…

In January 2014, the east coast (PJM) electricity grid experienced an increase in the price of electricity from $31/MWh to $2,680/MWh in a matter of 10 minutes. This extreme price volatility has…

Mining subgraph patterns is an active area of research. Till now, the focus has primarily been on mining all subgraph patterns in the given database. However, due to the exponential subgraph search…

Network pricing games provide a framework for modeling real-world settings with two types of strategic agents: users of the network and owners (operators) of the network. Owners of the network post a…

As the volume of data goes up, the quality of machine learning models, predictions, and services will improve. Once models are trained, predictive cloud services can be built on them, but users who…

How do we make AI agents appear to be more "human"? The goal of this session is to bring together researchers in human-computer interaction, linguistics, machine learning, speech, and…

The new wave of excitement about AI in recent years has been based on successes in perception tasks or on domains with limited and known dynamics. Because machines have achieved human parity in…

In 1982, Richard Feynman first proposed using a "quantum computer" to simulate physical systems with exponential speed over conventional computers. Quantum algorithms can solve problems in…

In the future, every company will be using AI, which means that every company will need a secure infrastructure that addresses AI security concerns. At the same time, the domain of computer security…

Microsoft Cognitive Toolkit (CNTK) is a production-grade, open-source, deep-learning library. In the spirit of democratizing AI tools, CNTK embraces fully open development, is available on GitHub, and…

Human society is faced with an unprecedented challenge to mitigate and adapt to changing climates, ensure resilient water supplies, sustainably feed a population of 10 billion, and stem a catastrophic…

Machine learning (ML) has demonstrated success in various domains such as web search, ads, computer vision, natural language processing (NLP), and more. These success stories have led to a big focus…

Scientists have long explored the different sensory inputs to better understand how humans perceive the world and control their bodies. Many of the great discoveries about the human perceptual system…

Computer vision is arguably one of the most challenging subfields of AI. To better address the key challenges, the vision research community has long been branched off from the general AI community…

Teaching machines to read, process and comprehend natural language documents and images is a coveted goal in modern AI. We see growing interest in machine reading comprehension (MRC) due to potential…

Recent research in recurrent neural models, combined with the availability of massive amounts of dialog data, have together spurred the development of a new generation of conversational systems. Where…

Advances in AI technologies have important ramifications for the development of accessible technologies. These technologies can augment the capabilities of people with sensory disabilities, enabling…

Over the last decade, algorithmic developments coupled with increased computation and data resources have led to advances in well-defined verticals of AI such as vision, speech recognition, natural…

Microsoft Research India started project MEC to understand the adoption of quality educational content in India and how technology can help uplift this adoption rate. The platform saw great adoption…

In this talk, I will discuss how to relax design constraints in quantum information processing in two distinct ways. First, I discuss using principled approaches to statistical inference to make more…

Error-correcting codes play a crucial role in safeguarding data against the adverse effects of noise during communication and storage. They are also powerful tools that in the arsenal of theoretical…

Static verification of performance properties of programs is an important problem that has attracted a lot of research. However, most existing tools infer best-effort upper bounds and hope that they…

Modeling Ecosystems from Genome to Society - A symposium presenting how an international consortium that is working to build the first full-scale simulation of a complex social-ecological system. The…

Today, one can Lyft a ride andAirBnb home away from home all in minutes. However, when comes to planning for our children's enrichment activities, it takes us poor parents hours and days. Now…

Unlike traditional machine learning methods, humans often learn from natural language instruction. As users become increasingly accustomed to interacting with computer devices using speech, their…

The design of fast, scalable, and correct concurrent systems remains a notoriously difficult task. Particularly problematic is the design of fast and scalable concurrent search data structures, which…

Social multimedia refers to the multimedia content generated by social network users for social interactions. The increasing popularity of online social networks accumulates large amount of social…

The DISCO Nets is a new type of probabilistic model for estimating the conditional distribution over a complex structured output given an input. DISCO Nets allows efficient sampling from a posterior…

Deep generative models have been wildly successful at learning coherent latent representations for continuous data such as video and audio. However, generative modeling of discrete data such as…

Bayesian optimisation (BO) is an optimisation method which incrementally builds a statistical model of the objective function to refine its search. Unfortunately, due to the curse of dimensionality,…

We present an algorithm for policy search in stochastic dynamical systems using model-based reinforcement learning. The system dynamics are described with Bayesian neural networks (BNNs) that include…

Recently, a new document metric called the word mover's distance (WMD) has been proposed with unprecedented results on kNN-based document classification. The WMD elevates high-quality word embeddings…

Three years ago, James Lloyd and I presented the Automatic Statistician at the joint MSR Cambridge machine learning meeting. The aim of this project is to automate the exploratory analysis and…

We propose a novel approximate inference algorithm that approximates a target distribution by amortising the dynamics of a user-selected MCMC sampler. The idea is to initialise MCMC using samples from…

A common practice in statistics and machine learning is to assume that the statistical data types (e.g., ordinal, categorical or real-valued) of variables, and usually also the likelihood model, is…

An important problem in achieving general artificial intelligence is the data-efficient learning of representations suitable for causal reasoning, planning, and decision making. Learning such…

The task of synthesizing programs given only example input-output behaviour is experiencing a surge of interest in the machine learning community. We present two directions for applying machine…

Important decisions about people are increasingly made by algorithms: Votes are counted; voter rolls are purged; financial aid decisions are made; taxpayers are chosen for audits; air travelers are…

Taking paper-driven processes online isn't easy – but keeping it secure is a whole other matter. To help, global eSignature and Digital Transaction Management (DTM) leader DocuSign recently used…

Unsupervised learning, in particular learning general nonlinear representations, is one of the deepest problems in machine learning. Estimating latent quantities in a generative model provides a…

A single glimpse is hardly enough to triangulate the 3D shapes of a scene. However, training examples are readily available, so statistical models can be trained to map appearance to shape. The…

Arguably, most excitement about deep learning revolves around the performance of convolutional neural networks and their ability to automatically extract useful features from signals. In this talk I…

Questions in real-world scenarios are mostly factoid, such as ``any universities in Seattle?''. In order to answer factoid questions, a system needs to extract world knowledge and reason over facts.…

Automatic program transformation tools can be valuable for programmers to help them with refactoring tasks, and for Computer Science students in the form of tutoring systems that suggest repairs to…

Increasingly, educational games have become an established part of the instructional landscape, however designing a compelling and effective instructional game remains a challenge. One of the reasons…

We consider the problem of estimating the value of MAX-CUT in a graph in the streaming model of computation. We show that there exists a constant $\e_* > 0$ such that any randomized streaming…

Consider a flying drone controlled from the ground by an observer who communicates with it via wireless. We are interested in how well the drone can be controlled via a channel that accepts r…

Modern enterprises collect data from their operations and the web, and strongly depend on the collected data to make important decisions. To analyze the collected data, enterprises need to first…

Pelican is a disk-based rack-scale cold data store. It has a converged rack-scale design in which resources, such as power and cooling are provisioned just for the requirements of the workload. Thanks…

The Research University, rooted in the Land-Grant College Act at the outset of industrialization and transformed post-war with "Science the Endless Frontier" and GI Bill participation, is…

Following the March 11, 2011 earthquake and tsunamis off the northeastern coast of Japan, the Fukushima Daiichi Nuclear Power Plant personnel encountered extraordinary challenges in managing the loss…

Creative language - the sort found in novels, film, and comics - contains a wide range of linguistic phenomena, from phrasal and sentential syntactic complexity to high-level discourse structures such…

To assess information seeking/retrieval processes and outcomes, goodness of various objects involved in search such as queries and documents are evaluated. This evaluation could lead to…

High-throughput instruments such as those at national laboratories can produce terabytes of experimental data in seconds, and this rate keeps increasing. However, the available software tools to…

A scalable programming language is one in which the same concepts can describe small as well as large parts. Towards this goal, the Scala programming language unifies concepts from object and module…

Historically, the evolution of network routers was driven primarily by performance. Recently, owing to the need for better control over network operations and the constant demand for new features,…

Automatic emotion recognition from speech is a challenging task which significantly relies on the emotional relevance of specific features extracted from the speech signal. In this study, our goal is…

Personal Information Management (PIM) is an activity in which an individual stores personal information items (e.g. files, emails and Web favorites) in order to retrieve them later. Despite the fact…

Software developers rely on many different media to communicate, collaborate, and coordinate with others. Recently, social tools have dramatically changed the landscape of software engineering,…

I will present work that leverages user behavioral data to build personalized applications, which I call "behavior-powered systems". Two applications use online user interactions: 1)…

The proliferation of datacenters, smartphones, personal sensing and tracking devices, and home automation products is fundamentally changing the applications we interact with daily. Modern…

My research focuses on applying differential privacy to problems in economics, statistical hypothesis testing, and adaptive data analysis in machine learning. At a high level, a differential private…

Bugs in programs remain a pernicious problem. Research techniques in automated program improvement and repair are typically classified as either heuristic---searching over a set of syntactic changes,…

Machine learning-based natural language processing systems are amazingly effective, when plentiful labeled training data exists for the task/domain of interest. Unfortunately, for broad coverage (both…

I will report on recent results about quantum algorithms for solving computational problems in number theory. I will show how they impact the security of certain post-quantum cryptosystems. Shor's…

Audiovisual media is an attractive and rich source of information, widely used in education, entertainment and industry. But it can take painfully long hours to author, edit and navigate. My research…

Personalized medicine has been hailed as one of the main frontiers for medical research in this century. In the first half of the talk, we will give an overview on our projects that use gene…

A powerful, but limited, way to view software is as source code alone. Mathematical techniques, such as abstract interpretation and model checking, can indicate whether the program satisfies a formal…

Traditional online transaction processing (OLTP) systems severely under-utilize the micro-architectural resources; more than half of the CPU cycles go to stalls, and the number of instructions retired…

My group develops systems to help people manage information and share it with others. We consider both text (online discussion tools) and structured data (information visualization and management…

Talks and presentations from the International Workshop on Spoken Language Translation (IWSLT) 2016. IWSLT is an annual workshop on annual scientific workshop on Spoken Language Translation and…

Mirko will present the work of Translators without Borders (TWB) and the role language technology they foresee to play for their work in the future. In his talk, he will highlight on the one hand the…

In the last two years, attentional-sequence-to-sequence neural models have become the state-of-the-art in machine translation, far surpassing the accuracy phrasal translation systems of in many…