In the news | Microsoft Innovation
Building a computer that solves practical problems at the speed of light
There’s an old saying: When the only tool you have is a hammer, every problem looks like a nail. Sometimes referred to as “the law of the instrument,” that hammer-and-nail idea is a common pitfall in research; when you’re not…
In the news | BBC Newsnight
AI cuts treatment time for cancer radiotherapy
A new type of artificial-intelligence technology that cuts the time cancer patients must wait before starting radiotherapy is to be offered at cost price to all NHS trusts in England. It helps doctors calculate where to direct the therapeutic radiation…
Research Focus: Week of June 19, 2023
In this issue: Our new Responsible AI Maturity Model; FoundWright helps “re-find” web pages; Trace-guided Inductive Synthesis of Recursive Functional Programs; a wait-free algorithm for weak reference counting; and new research on concurrency testing.
In the news | This American Life
Greetings, People Of Earth: First Contact
This American Life’s David Kestenbaum (NPR) interviewed Peter Lee, Sébastien Bubeck, Ece Kamar, and Ronen Eldan about their experiments with GPT-4 and the emerging capabilities of large language models, using examples from the “Sparks” paper.
DeepSpeed ZeRO++: A leap in speed for LLM and chat model training with 4X less communication
| DeepSpeed Team and Andrey Proskurin
Large AI models are transforming the digital world. Generative language models like Turing-NLG, ChatGPT, and GPT-4, powered by large language models (LLMs), are incredibly versatile, capable of performing tasks like summarization, coding, and translation. Similarly, large multimodal generative models like…
Collaborators: Renewable energy storage with Bichlien Nguyen and David Kwabi
| Gretchen Huizinga, Bichlien Nguyen, and David Kwabi
Researcher Bichlien Nguyen is an organic electrochemist turned technologist. Professor David Kwabi is a mechanical engineer. Their work uses ML to help discover organic compounds for renewable energy storage. Learn about their collaboration.
CVPR 2023 | 掩码图像建模MIM的理解、局限与扩展
编者按：掩码图像建模（Masked Image Modeling, MIM）的提出，为计算机视觉模型训练引入无监督学习做出了重要贡献。得益于 MIM 的预训练算法，计算机视觉领域在近年来持续输出着优质的研究成果。然而整个业界对 MIM 机制的研究仍存在不足。 秉持着不断扩展前沿技术边界的探索精神，微软亚洲研究院的研究员们在理解 MIM 作用机制，以及基于这些机制提升现有 MIM 算法的领域，取得了...
Microsoft at CVPR 2023: Pushing the boundaries of computer vision
| Baining Guo and Steve Lin
In the vast realm of artificial intelligence, few fields have captivated our imagination and pushed the boundaries of possibility quite like computer vision. At the core of this domain of research and innovation lies the ambition to empower technologies for…
In the news | Microsoft Research Podcast
Collaborators: Renewable energy storage with Bichlien Nguyen and David Kwabi