eMatrix

微软亚洲研究院通讯

eMatrix

MSR Asia PR Newsletter

2017年06月

2017 June

中文  |  EN

焦点纵横

  • 人工智能助力资管行业腾飞 华夏-微软宣布在人工智能投资领域战略合作

    6月13日,中国领先资产管理公司华夏基金同国际科技巨头微软公司在亚太地区设立的微软亚洲研究院举办战略合作发布会,宣布双方将就人工智能在金融服务领域的应用展开战略合作研究。此次跨界合作研究将结合微软在人工智能领域的深厚积淀以及华夏基金强大的投研实力,旨在研究智能投资的疆界,推动资产管理行业智能化转型。

  • 微软亚洲研究院周明:未来5-10年,NLP将走向成熟

    在微软亚洲研究院举办的自然语言处理(NLP)前沿技术分享会上,副院长周明从自然语言处理是什么、微软亚洲研究院在自然语言处理方向的最新成果以及自然语言处理技术的未来发展方向三部分,对自然语言处理技术进行了全方位解读。在周明看来,未来5-10年,自然语言处理技术将走向成熟,并将迎来过去60 年发展最迅速的时期。

Spotlight

  • China Asset Management and Microsoft Announce Strategic Cooperation in AI-Aided Investment

    China Asset Management (China AMC), a leading asset management company in China, and Microsoft Research Asia, Microsoft’s research arm in the Asia Pacific region, jointly hosted a strategic cooperation conference to announce their research collaboration on the application of artificial intelligence (AI) in financial services. This cross-industry cooperation will combine Microsoft's rich accumulation of artificial intelligence research and China AMC’s powerful strength in investment research to push for digital transformation in the financial industry.

  • Ming Zhou: NLP technologies will grow towards maturity in the next 5 to 10 years

    On June 1st, Microsoft Research Asia conducted a media workshop on advanced Natural Language Processing technologies. Ming Zhou, Assistant Managing Director of MSR Asia introduced the history and development of researches on NLP and summarized the great achievements of MSR Asia in NLP. He also presented his views on the future trend of NLP technologies, as well as potential research directions. He anticipates NLP technologies to head towards maturity in the next 5 to 10 years.

媒体视角

近日,腾讯科技对微软全球资深副总裁、微软亚太研发集团主席兼微软亚洲研究院院长洪小文进行了专访,洪小文就微软亚洲研究院的职能、目前微软人工智能在2B和2C业务上的应用、人工智能未来的发展方向以及人类与人工智能的相处方式等方面进行了深度分享。洪小文博士表示,人工智能绝对不是无所不能,要想进步需关注基础研究;人工智能不会取代人,AI(人工智能)+HI(人类智能)才是真正的未来。

微软亚洲研究院副院长张益肇在未来论坛举办的“脑科学与人工智能”主题讲座上详细介绍了人工智能在脑科学领域的应用,并归纳了人工智能的三个关键词——数据、辅助、突破。张益肇博士表示,人工智能已经开始普及,数据是中国开展相关领域科学研究的优势。同时,他还在讲话中重点谈到了人工智能对人类疾病的辅助作用。而关于未来,张益肇则认为人工智能想要追上人类的“学习能力”,需要从“学习能力”和“自省能力”入手。

近日,人民日报采访了微软亚洲研究院副院长刘铁岩,并在近期发表的人工智能主题文章中引用了刘铁岩博士的诸多观点。刘铁岩表示,人工智能看上去这两年才火,但事实上多年来一直有人在背后做研究,如今它早已渗透到了人们生活的方方面面。而目前人工智能的第三次浪潮更加稳健,它像燃料一样,与各个产业深度结合,有着比以往更多的、实际落地的应用场景。

2017微软亚洲研究院学术日(Academic Day)活动近日在台湾举办,此次活动主题聚焦于人工智能与机器人研究,来自全球产学研各界的专家齐聚一堂。活动上,微软全球资深副总裁、微软亚太研发集团主席兼微软亚洲研究院院长洪小文表示,微软亚洲研究院相当重视台湾科技人才的创新能力,在近几年推广了研究生实习,并展开了与台湾各大学的合作研究计划,提供自由、开放的研究环境,解决未来问题。

世界顶级的计算机视觉学术会议 CVPR 2017(计算机视觉与模式识别会议)将于7月在夏威夷举行。而在此之前,微软亚洲研究院在北京举办了“微软亚洲研究院创研论坛——CVPR 2017 论文分享会”,来自微软亚洲研究院、中科院、英特尔中国研究院、北京大学、清华大学等国内外计算机视觉领域学术界、工业界的优秀代表们携各自在CVPR 2017发表的最新研究结果和相关技术观点,在此次论坛上进行了分享与探讨。

微软专为iPhone和iPad打造的基于人工智能技术的相机应用Microsoft Pix最近推出了新功能。新功能之一“Pix风格”能够将照片转换成类似知名艺术作品的风格,第二个新功能“Pix绘画”则可以制造出以所选艺术风格一笔一笔绘制作品的延时拍摄效果。新功能由微软亚洲研究院、Skype团队和Pix团队共同开发,通过微软研究人员开发的智能算法套件,该应用能够消除拍摄中的不确定因素,让每一次快门都能拍出精美的照片。

Media Coverage

Tencent conducted an exclusive interview with Hsiao-Wuen Hon, Corporate Vice President of Microsoft and Managing Director of MSR Asia. Hsiao-Wuen Hon shared comprehensive information on the function of Microsoft Research Asia, the application of Microsoft’s AI in commercial and consumer business, the future trend of AI, and, from his perspective, the manner in which humans can coexist with AI. He explained that humans will not be replaced by AI, and that AI + HI is what the future entails.

ChinaNews reported the keynote speech on the subject of “brain science and artificial intelligence” delivered by Eric Chang, Senior Director of Technology Strategy at MSR Asia, during the Future Forum workshop. Eric pointed out that the three key words for AI are data, subsidiarity and breakthrough. He said that AI is becoming universal and that data is the core advantage of China’s scientific research in this area. He also highlighted the supporting role that AI plays in treating human diseases and argued that for AI to catch up with humans, focus should be placed on the learning ability and introspective ability of AI.

People’s Daily conducted an interview with Tie-Yan Liu, Principal Researcher at MSR Asia, on the topic of artificial intelligence and quoted him amply in the article. Tie-Yan Liu explained that AI may appear to have become popular only in the past two years, but in fact, people have been researching it for years, and AI has already penetrated into all aspects of people’s lives. He believes that the third wave of AI is more robust than ever, as AI becomes integrated with various industries and gains more application scenarios.

Microsoft Research Asia Academy Day 2017 was recently held in Taiwan. The theme this year was artificial intelligence and robotics research. Academic and industry experts from around the world gathered together. Hsiao-Wuen Hon, Corporate Vice President of Microsoft and Managing Director of MSR Asia, expressed appreciation of Taiwan’s talents. MSR Asia is committed to cooperating on research programs with Taiwanese universities, and has provided open research environments to solve future problems.

Microsoft Research Asia Innovation and Research Forum - CVPR 2017 Paper Sharing Workshop was successfully held in Beijing recently. Leading academic figures, including professors, researchers, and students from top universities and institutions, and industry figures from major startup corporations participated in the workshop onsite. These representative delegates in the field of computer vision from around the world shared their CVPR papers and held a heated discussion during the forum on their latest research results and technical viewpoints.

Microsoft Pix, an AI-powered camera app for the iPhone and iPad, has recently released new features – Pix Styles and Pix Painting, which can automatically lend snapshots a stylistic flair reminiscent of masterpiece paintings and artsy photos hanging in famous galleries. MSR Asia, the Skype team and the Pix team joined efforts to develop these new features. Researchers from MSR Asia developed an intelligent algorithm suite that can take the guesswork out of getting great photos with every tap of the shutter button.

“博”闻创新

  • 以数据的名义——浅谈三维几何的处理与分析

    分析与处理三维形体是计算机图形学中的一个基本任务与研究方向。结合近期微软亚洲研究院网络图形组的相关工作,主管研究员刘洋撰写了系列文章以分享数据在三维几何处理与分析的作用与效力。上篇主要介绍了三维几何处理中的去噪问题,下篇则分享了微软亚洲研究院网络图形组在三维几何处理领域的研究工作——如何利用卷积神经网络高效处理三维数据。

  • 从洪荒到智能——数据驱动的材质属性建模发展历程

    机器学习早在10年前就已在计算机图形学中被应用,那时,它被称之为“数据驱动”方法。网络图形组主管研究员董悦基于数据驱动方法讲述了早年间图形学中的机器学习技术,并以微软亚洲研究院网络图形组在数据驱动图形学方向上的材质属性建模工作为例,介绍了数据驱动图形学的发展历史。

  • 可变形卷积网络:计算机新“视”界

    如何适应图像中复杂的几何形变问题是物体识别的主要难点,也是计算机视觉领域多年来的核心问题之一。近日,微软亚洲研究院视觉计算组公布了一篇题为“Deformable Convolutional Networks”(可变形卷积网络)的论文,首次在卷积神经网络中引入了学习空间几何形变的能力。经验证,可变形卷积网络在物体检测,语义分割等计算机视觉任务上都获得了重大的性能提升。

  • 计算机视觉,路在何方

    在近期举办的“微软亚洲研究院创研论坛——CVPR 2017论文分享会“的圆桌讨论环节中,五位计算机视觉领域的资深专家:微软亚洲研究院首席研究员曾文军与中科院的陈熙霖教授,北京航空航天大学的王蕴红教授,北大的林宙辰教授以及美国中佛罗里达大学的齐国君教授,就目前计算机视觉领域研究中的热门问题给出了他们各自的见解。

  • 微软自然语言理解平台LUIS:从零开始,帮你开发智能音箱

    在微软开发者大会Build 2017上,智能音箱Invoke受到了广泛关注,高效地开发语义理解模块是智能音箱开发的关键环节。微软亚洲研究院大数据挖掘组研发了LUIS的新一代算法,让非NLP专业的开发者能够轻松地创建和维护高质量的自然语言理解模型,并无缝对接到相关的智能应用当中。

  • 手把手带你入门微软Graph Engine

    出自微软亚洲研究院的Graph Engine是一个基于内存的分布式大规模图数据处理引擎,能够帮助用户高效地处理大规模图数据,且更方便地构建实时查询应用和高吞吐量的离线分析平台。自发布以来受到了学术界和工业界的广泛关注。本文中,Graph Engine的主要设计与开发者之一,机器学习组副研究员李亚韬详细介绍了这一平台,并演示了一些快速上手的实例。

AI Headlines

  • In the name of data - Brief introduction to the treatment and analysis of 3D geometry

    Analysis and processing of 3D geometry is a fundamental research direction for computer graphics. Yang Liu, Lead Researcher at Microsoft Research Asia, composed a series of articles that integrate Internet Graphics Group’s recent work in this area to share the role and effectiveness of data in 3D geometry processing and analysis. The first part mainly introduces the denoising problem in 3D geometry processing, and the second part talks about the group’s research in processing 3D data with convolutional neural networks.

  • The development history of the data driven modeling of material property modeling

    Ten years ago, machine learning was already being utilized in computer graphics and was known as the data driven approach. In this article, Yue Dong, Lead Researcher at Internet Graphics Group, illustrated machine learning techniques based on data-driven methods in early graphics, and introduced the development history of data driven graphics with several examples of Internet Graphics Group’s material property modeling work.

  • Deformable convolutional networks: New visual field for computers

    Ten years ago, machine learning was already being utilized in computer graphics and was known as the data driven approach. In this article, Yue Dong, Lead Researcher at Internet Graphics Group, illustrated machine learning techniques based on data-driven methods in early graphics, and introduced the development history of data driven graphics with several examples of Internet Graphics Group’s material property modeling work.

  • Where is the way for computer vision?

    Recently, Microsoft Research Asia Innovation and Research Forum - CVPR 2017 Paper Sharing Workshop was successfully held. Five senior experts in the field of computer vision, including Wenjun Zeng, Principal Researcher at MSR Asia, Professor Xilin Chen from the Chinese Academy of Sciences, Professor Yunhong Wang from BUAA, Professor Zhouchen Lin from PKU, and Professor Guojun Qi from the University of Central Florida, held a discussion and presented their views on popular issues in the research of computer vision during the panel session.

  • Microsoft’s natural language understanding platform LUIS: Helping you develop intelligent speakers starting from scratch

    Intelligent speaker Invoke has received widespread attention from the public since it was released at Build 2017. Because the efficient development of a semantic understanding module played an important role, Big Data Mining Group researched and developed a new LUIS algorithm that allows non-professional NLP developers to easily create and maintain high-quality natural language understanding models, and to realize seamless connection to the relevant intelligent applications.

  • Acquainting you with Microsoft Graph Engine

    Graph Engine is a distributed in-memory large-scale data processing engine that can help users efficiently deal with large-scale data and more conveniently build real-time query applications and high-throughput, offline-analysis platforms. It has received widespread attention from the world of academia and the industry since it was released. Yatao Li, one of the main designers and developers of Graph Engine and an associate researcher at Machine Learning Group, gave a comprehensive introduction to the platform and demonstrated a few quick start examples in the video.

“微”言梦想

“吃豆人”的历史最高分是怎么获得的

微软研究人员在近期创建了一套基于人工智能的系统,以学习如何在80年代风靡一时的电子游戏“吃豆小姐”中获得最高分。该系统运用了“分而治之”的方法,可广泛用于对AI代理进行训练,使之掌握能够增强人类能力的复杂任务。

从视觉智能到智能家居系统

上海交通大学的林巍峣副教授与微软亚洲研究院合作,基于深度学习技术,设计了一个针对视觉目标的实时检测、跟踪系统。该系统可以可靠的检测异常事件,并根据用户感兴趣的内容自适应地创建“在线”摘要视频。

西安交通大学与微软签署战略合作备忘录

6月7日,西安交通大学与微软(中国)有限公司签署战略合作备忘录,双方将在校企合作、协同育人等方面展开广泛实质性合作。一直以来,微软亚洲研究院与国内高校在科研合作、人才培养等方面都保持着密切的合作与交流。

虚拟与混合现实的未来

想象过一副普通的眼镜就能让你身临其境体验全息场景吗?微软在SIGGRAPH 2017的最新论文带我们向这个美好的愿景迈出了一大步。研究员们开发出了一款真正的近眼全息屏幕,它可以解决视野与设备外形间平衡、用户视觉矫正、高精度的聚焦控制等问题。

微软发布升级版认知工具包,加速深度学习研发

微软在开源平台正式发布了2.0版本的微软认知工具包(Microsoft Cognitive Toolkit),用于生产级和企业级的深度学习工作负载。更新版包括了测试版发布以来累积的几百项新功能,简化了深度学习过程,能够与更广泛的AI生态系统实现无缝整合。

量子计算与密码安全

基于云的电子商务系统和私人数据安全性很大程度上依赖于公钥加密系统的可靠性。未来强大的量子计算机、量子算法可能攻破它们背后的数学难题。因此,微软研究院的研究员提出了Supersingular Isogeny Graph中的寻路问题,并将其应用到密钥交换与数字签名中。

Weibo Highlights

How Microsoft researchers used AI to master Ms. Pac-Man

Microsoft researchers created an artificial intelligence-based system that learned how to obtain the maximum score on the addictive 1980s video game Ms. Pac-Man using a divide-and-conquer method. This system could have broad implications for teaching AI agents to perform complex tasks that augment human capabilities.

From visual intelligence to intelligent home systems

In collaboration with MSR Asia, Lin Weiyao, Associate Professor at Shanghai Jiao Tong University, designed a real-time inspection and detection system for visual targets based on deep learning. The system can detect abnormal events and create “online” video summaries according to users’ interests.

Xi'an Jiao Tong University and Microsoft China sign MOU

On June 7th, Xi'an Jiao Tong University signed an MOU with Microsoft China to carry out strategic cooperation and collaborative talent fostering.MSR Asia maintains close cooperation and communication with domestic universities in scientific research and talent fostering.

The future of virtual and mixed reality

Have you ever imagined that a pair of glasses could allow you to experience holographic scenes? Microsoft published a paper at SIGGRAPH 2017, which took a big step toward realizing this wonderful vision. Researchers developed a near-eye holographic screen that can balance FOV and appearances, and that also solves optical correction and focusing control problems.

Microsoft released upgraded Cognitive Toolkit to accelerate R&D of deep learning

Microsoft officially released Microsoft Cognitive Toolkit Version 2.0 on open source platforms and included hundreds of new functions. It can be applied to commercial-grade deep learning workloads, simplifying the deep learning process and achieving seamless integration with the AI ecosystem.

Quantum computing and cryptographic security

The security of cloud-based e-commerce systems and private data largely depend on the reliability of public key cryptography. Future quantum computers and quantum algorithms will solve the mathematical problems behind the reliability issue. Researchers from Microsoft Research proposed routing problems in Supersingular Isogeny Graph and will apply them in key exchange and digital signature.

欢迎订阅我们

输入邮箱地址,即可每月收到期刊更新哦!

订 阅