对话 2021 微软学者：青年从不缺少描摹未来的潜力
编者按：“微软学者”奖学金是微软亚洲研究院自 1999 年启动的一项面向亚太地区计算机科学以及相关专业和交叉学科的优秀博士生的项目，旨在发掘优秀、有潜力的低年级博士生。 距离 2022 年“微软学者”奖学金申请截止仅有 16 天，我们带来了与往届“微软学者”的对话，期待他们的科研经历与成长感悟，能给予同样在科研之路上探索的你许多鼓励与勇气。 “微软学者”们如何选择与探索自己的研究领域？怎样的科研习...
Improve edge-device AI efficiency
Machine learning models are increasingly running on edge hardware, such as mobile phones or Internet of Things (IoT) devices. Motivations include protection of private data and avoidance of networking latency, for example with applications that recognize speech. Ensuring efficient inference…
Emit less carbon from AI
Multiple activities are involved in developing and using machine learning models, including selection of model architectures and algorithms, hyperparameter tuning, training on existing datasets, and making predictions on new data (aka inference). Optimizing results across these activities involves many complex…
Empower AI developers
Progress in machine learning is measured in part through the constant improvement of performance metrics such as accuracy or latency. Carbon footprint metrics, while being an equally important target, have not received the same degree of attention. With contributions from…
In the news | Microsoft News
Two NHS surgeons are using Azure AI to spot patients facing increased risks during surgery
Partnering people with large language models to find and fix bugs in NLP systems
| Scott Lundberg and Marco Tulio Ribeiro
Advances in platform models—large-scale models that can serve as foundations across applications—have significantly improved the ability of computers to process natural language. But natural language processing (NLP) models are still far from perfect, sometimes failing in embarrassing ways, like translating…
(De)ToxiGen: Leveraging large language models to build more robust hate speech detection tools
| Hamid Palangi and Dipankar Ray
It’s a well-known challenge that large language models (LLMs)—growing in popularity thanks to their adaptability across a variety of applications—carry risks. Because they’re trained on large amounts of data from across the internet, they’re capable of generating inappropriate and harmful…
In the news | Microsoft Security Response Center Blog
New Research Paper: Pre-hijacking Attacks on Web User Accounts
In 2020, MSRC awarded two Identity Project Research Grants to support external researchers working to further strengthen the security of identity protocols and systems. Today we are pleased to release the results of the first of these projects. This research, led by…
ACL 2022 | NLP领域最新热门研究，你一定不能错过！
编者按：作为自然语言处理领域的国际顶级学术会议，ACL 每年都吸引了大量学者投稿和参会，今年的 ACL 大会将于5月22日至5月27日举办。值得注意的是，这也是 ACL 大会采用 ACL Rolling Review 机制后的首次尝试。在此次会议中，微软亚洲研究院有多篇论文入选，本文精选了其中的6篇进行简要介绍，论文主题涵盖了：编码器解码器框架、自然语言生成、知识神经元、抽取式文本摘要、预训练语言...