Microsoft Research Webinars

Traditional deep learning frameworks such as TensorFlow and PyTorch support training on a single deep neural network (DNN) model, which involves computing the weights iteratively for the DNN model. Designing a DNN model for a task remains an experimental science and is typically a practice of deep learning model exploration. Retrofitting such exploratory-training into the training process of a single DNN model, as supported by current deep learning frameworks, is unintuitive, cumbersome, and inefficient. In…

Microsoft Soundscape uses 3D spatial audio to help people understand their surroundings. From its original goal of empowering people with sight loss to be more independent, Soundscape has grown to provide richer and more immersive experiences for anyone. Soundscape turns a headset into an audio display--like a map augmented with 3D sound. It can also work in the background, providing heads-up, hands-free ambient awareness.

In this webinar, join Marina Polishchuk, in exploring how RESTler—the first stateful REST API fuzzer—can help efficiently find security and reliability bugs in cloud services. RESTler analyzes a Swagger/OpenAPI specification and produces a fuzzing grammar that contains information about requests and their dependencies. RESTler only fuzzes a request if all its dependent resources have been successfully created—this enables RESTler to achieve deeper coverage out of the box. RESTler also offers a pluggable model for checking…

In this webinar, Senior Researcher Xu Tan will address the challenges with advancing neural TTS, specifically the high computational cost and slow inference speed in online serving; word skipping and repeating issues, poor voice quality, and lack of voice controllability; the large amounts of training data needed for improved voice synthesis; and the practical challenges in TTS voice adaptation. He’ll introduce his team’s work addressing these challenges—including fast TTS, end-to-end TTS, low-resource TTS, and adaptive…

Join Microsoft researchers Su Lin Blodgett and Alexandra Olteanu, from the FATE Group at Microsoft Research Montréal, to examine pitfalls in some state-of-the-art approaches to measuring computational harms in language technologies. For such measurements of harms to be effective, it is important to clearly articulate both: 1) the construct to be measured and 2) how the measurements operationalize that construct. The webinar will also overview possible approaches practitioners could take to proactively identify issues that…

The webinar will present examples of how these learnings are shaping our research on developing principles and tools for bringing the AI principle of reliability and safety to reality. In particular, it will showcase an ecosystem of open-source tools that are intended to accelerate the machine learning (ML) development life cycle by identifying and mitigating failures in a faster, systematic, and rigorous way.

AI has transformed modern life via previously unthinkable feats, from machines that can master the ancient board game Go and self-driving cars to developments we experience more routinely, such as virtual agents and personalized product recommendations. Simultaneously, these new opportunities have raised new challenges—most notably, challenges that have highlighted the potential for AI systems to cause fairness-related harms. Indeed, the fairness of AI systems is one of the key concerns facing society as AI continues…

In this webinar, join Junwon Park and Zygmunt Lenyk, to learn how the team built Microsoft Vision Model using multi-task learning and web-supervised datasets. You’ll examine how the model lowers cost in a production setting while achieving state-of-the-art performance by using multi-task learning and optimizing separately for multiple classification tasks, including ImageNet-22k, COCO, and two web-supervised datasets of 40 million image-label pairs collected from image search engines.

In this webinar, learn how the latest AI techniques can be applied to multiplayer games in the challenging and diverse 3D environment of Minecraft. Join the researchers as they demonstrate how Project Malmo—a platform for AI experimentation built on Minecraft—provides an ideal environment for designing different and rich training tasks and how reinforcement learning agents can be trained in these scenarios. They’ll provide examples of tasks, agent implementations, and the latest research done in this…

This webinar will focus on some of the recent vision-language pretraining (VLP) approaches for image captioning. We will cover our latest approaches, including object-semantics aligned pretraining (OSCAR) and visual-vocabulary pretraining (VIVO). We will also discuss their key principles and how we address the core challenges in image caption generation. Join us to learn how our discovery leads to a new image captioning framework that achieves state-of-the-art performance on the nocaps benchmark (developed to evaluate NOC…

In this webinar, Dr. Ozan Oktay and Dr. Anton Schwaighofer will analyze these challenges within the context of image-guided radiotherapy procedures and will present the latest research outputs of Project InnerEye in tackling these challenges. The first part of the webinar will focus on a research study that evaluates the potential clinical impact of DL models within the context of radiotherapy planning procedures. The second part of the talk will introduce the open-source InnerEye Deep…

In this webinar, join Jeff Delmerico and Helen Oleynikova, Senior Scientists at the Mixed Reality and AI Zurich Lab, in exploring how mixed reality (MR)—which blends the digital world with the physical one—can be used to provide an easier and safer way to interact with robots.