项目
Dataflow-based Dialogue
We’ve developed a new dialogue modeling framework for dialogue that can handle complex scenarios. Our approach is based on representing dialogue states as dataflow graphs.
Human language technologies
Advancing how machines understand and generate human language across speech, text, and translation.
视频
Building multimodal, integrative AI systems with Platform for Situated Intelligence
In the last decade, we’ve seen fast-paced progress in many individual AI areas, such as computer vision, speech, and machine translation. However, anyone who’s tried bringing multiple AI technologies together in end-to-end systems designed for…
微软研究院博客
DeepSpeed: Extreme-scale model training for everyone
In February, we announced DeepSpeed, an open-source deep learning training optimization library, and ZeRO (Zero Redundancy Optimizer), a novel memory optimization technology in the library, which vastly advances large model training by improving scale, speed,…
网络研讨会
Building multimodal, integrative AI systems with Platform for Situated Intelligence webinar
In this webinar, Dan Bohus, a Senior Principal Researcher in the Perception and Interaction Group at Microsoft Research, will introduce Platform for Situated Intelligence, an open-source framework that aims to address these challenges and accelerate and…
