Cortana for business–now the worker’s digital assistant

Sina WeiboBaiduLinkedInQQGoogle+RedditEvernote分享


Microsoft revealed a little more about Cortana Analytics and how it is going to use machine learning to revolutionise business intelligence and productivity via its Azure cloud.

Joseph Sirosh is the corporate vice president of the Information Management and Machine Learning (IMML) team in the Cloud and Enterprise group at Microsoft. He is also a maths nut and perhaps a geek. Almost all of what he said was several levels above my pay grade so I will do my best not to disappoint TWire’s erudite readers (and keep it light).

“I did my PHD in 1995 in neural networks. That led to my working on fraud detection [machine learning] software for a major credit card gateway and that led to joining Microsoft where I can pursue my passion of machine learning,” he said.

“Maths applies to data. Software is eating the world (apologies to Marc Andreessen for that saying) but data is drowning the world,” he said.

“30 years ago data was almost all analogue, now it iP addresss attached almost all analogue, now its all digital and it has an pursue my passionm s all digital and much of it has an IP address attached. The intelligence you can derive is amazing,” he said.

The problem with big data, lots of data, Petabytes, Exabytes or the stuff is that it presents an enormous challenge to store and analyse, especially in real time. Microsoft and its Azure Machine Learning and the Azure Gallery is a good place to find out more.

Joseph spoke about the  ‘connected dairy cow’ , a project from Fujitsu in Japan. Pedometers were attached to each cow and the bovine walking characteristics analysed via Azure to predict estrus. Not only did the data reveal the best insemination time and improved the pregnancy rate by 67% it helped predict the resultant gender of the offspring. The results were so good that Fujitsu expects this to be rolled out to 500,000 cows soon.

Joseph said that with Azure cloud developers could build immensely complex systems in just a few days and introduced two plucky Adelaide undergraduates who hope to move the highly manual and inaccurate vineyard yield prediction models to the cloud using Azure.

Harry Lucas, 20, Liam Ellul, 23 are part of the Seer Insights team. They described an intelligent system that assists vineyard staff, growers and wineries in improving the accuracy of their yield estimates. The key is to use Azure to find patterns and insights from both the current data and from historical data and shifting that from an analysis of what’s happened in the past to an accurate prediction of what might occur in the future.

Joseph commented that what Seer had done is harness computing power, machine learning and analytics that were not available even a few years ago.

Then we heard from Max Valente from Thought studio a software development company he began in Sydney in 2011 as a think tank developing technology to solve business problems and create business opportunities.

Max developed the first virtual wardrobe where a person stands in front of a screen and tries on virtual clothes. The latest venture (it has two sites working as proof of concept) uses sensors and devices like Kinect to map out a shoppers experience in a store.

“Kinect gives us the gender, approximate age and time someone enter and exits the store. We can marry that with things like the weather, promotional activity and more to build a profile. We then use sensors connected to Raspberry Pi to show where they go in the store, where they linger, what they try on – we collect a lot of data,” Max said. One advantage is immediately apparent – identifying shoplifters by facial and other ‘secret’ Kinect recognition.

“The goal is to enhance a shopper experience, understand their journey, to lay out stores more logically, to know what’s hot and what’s not, and of course to increase sales,” said Max.

“The entire system is only possible due to Microsoft’s Azure cloud, analytics, storage and tools,” Max added.

So finally to Cortana

Cortana is a digital assistant that can be personal (as in running on a PC or smartphone) or used for business. The former relies on getting to know you well and the latter on getting to know the company well – and keeping all secrets.

Cortana analytics is too broad a subject to cover here but Joseph demonstrated a medical support practice that uses Cortana as the natural language interface between nurses/doctors and patients in a call centre environment. The result has been much better care, better analysis, easier follow up and less need to hospitalise.

Joseph said that Cortana could develop perceptual intelligence to benefit any business – helping them to get closer to customers, knowing more about what they do, and what customers want etc.

He finished with a haiku:

The cloud turns hardware into software

The cloud turns software into services

The cloud turns data into intelligence

The cloud turns intelligence into action

I then walked outside into Sydney’s beautiful sunshine – not a [visible] cloud in sight! The cloud turns into rain, rain turns into …


The Dawn of Artificial Intelligence




Why a deep-learning genius left Google & joined Chinese tech shop Baidu (interview)


SUNNYVALE, California — Chinese tech company Baidu has yet to make its popular search engine and other web services available in English. But consider yourself warned: Baidu could someday wind up becoming a favorite among consumers.

The strength of Baidu lies not in youth-friendly marketing or an enterprise-focused sales team. It lives instead in Baidu’s data centers, where servers run complex algorithms on huge volumes of data and gradually make its applications smarter, including not just Web search but also Baidu’s tools for music, news, pictures, video, and speech recognition.

Despite lacking the visibility (in the U.S., at least) of Google and Microsoft, in recent years Baidu has done a lot of work on deep learning, one of the most promising areas of artificial intelligence (AI) research in recent years. This work involves training systems called artificial neural networks on lots of information derived from audio, images, and other inputs, and then presenting the systems with new information and receiving inferences about it in response.

Two months ago, Baidu hired Andrew Ng away from Google, where he started and led the so-called Google Brain project. Ng, whose move to Baidu follows Hugo Barra’s jump from Google to Chinese company Xiaomi last year, is one of the world’s handful of deep-learning rock stars.

Ng has taught classes on machine learning, robotics, and other topics at Stanford University. He also co-founded massively open online course startup Coursera.

He makes a strong argument for why a person like him would leave Google and join a company with a lower public profile. His argument can leave you feeling like you really ought to keep an eye on Baidu in the next few years.

“I thought the best place to advance the AI mission is at Baidu,” Ng said in an interview with VentureBeat.

Baidu’s search engine only runs in a few countries, including China, Brazil, Egypt, and Thailand. The Brazil service was announced just last week. Google’s search engine is far more popular than Baidu’s around the globe, although Baidu has already beaten out Yahoo and Microsoft’s Bing in global popularity, according to comScore figures.

And Baidu co-founder and chief executive Robin Li, a frequent speaker on Stanford’s campus, has said he wants Baidu to become a brand name in more than half of all the world’s countries. Presumably, then, Baidu will one day become something Americans can use.

Now that Ng leads Baidu’s research arm as the company’s chief scientist out of the company’s U.S. R&D Center here, it’s not hard to imagine that Baidu’s tools in English, if and when they become available, will be quite brainy — perhaps even eclipsing similar services from Apple and other tech giants. (Just think of how many people are less than happy with Siri.)

A stable full of AI talent

But this isn’t a story about the difference a single person will make. Baidu has a history in deep learning.

A couple years ago, Baidu hired Kai Yu, a engineer skilled in artificial intelligence. Based in Beijing, he has kept busy.

“I think Kai ships deep learning to an incredible number of products across Baidu,” Ng said. Yu also developed a system for providing infrastructure that enables deep learning for different kinds of applications.

“That way, Kai personally didn’t have to work on every single application,” Ng said.

In a sense, then, Ng joined a company that had already built momentum in deep learning. He wasn’t starting from scratch.

Only a few companies could have appealed to Ng, given his desire to push artificial intelligence forward. It’s capital-intensive, as it requires lots of data and computation. Baidu, he said, can provide those things.

Baidu is nimble, too. Unlike Silicon Valley’s tech giants, which measure activity in terms of monthly active users, Chinese Internet companies prefer to track usage by the day, Ng said.

“It’s a symptom of cadence,” he said. “What are you doing today?” And product cycles in China are short; iteration happens very fast, Ng said.

Plus, Baidu is willing to get infrastructure ready to use on the spot.

“Frankly, Kai just made decisions, and it just happened without a lot of committee meetings,” Ng said. “The ability of individuals in the company to make decisions like that and move infrastructure quickly is something I really appreciate about this company.”

That might sound like a kind deference to Ng’s new employer, but he was alluding to a clear advantage Baidu has over Google.

“He ordered 1,000 GPUs [graphics processing units] and got them within 24 hours,” Adam Gibson, co-founder of deep-learning startup Skymind, told VentureBeat. “At Google, it would have taken him weeks or months to get that.”

Not that Baidu is buying this type of hardware for the first time. Baidu was the first company to build a GPU cluster for deep learning, Ng said — a few other companies, like Netflix, have found GPUs useful for deep learning — and Baidu also maintains a fleet of servers packing ARM-based chips.

Now the Silicon Valley researchers are using the GPU cluster and also looking to add to it and thereby create still bigger artificial neural networks.

But the efforts have long since begun to weigh on Baidu’s books and impact products. “We deepened our investment in advanced technologies like deep learning, which is already yielding near term enhancements in user experience and customer ROI and is expected to drive transformational change over the longer term,” Li said in a statement on the company’s earnings the second quarter of 2014.

Next step: Improving accuracy

What will Ng do at Baidu? The answer will not be limited to any one of the company’s services. Baidu’s neural networks can work behind the scenes for a wide variety of applications, including those that handle text, spoken words, images, and videos. Surely core functions of Baidu like Web search and advertising will benefit, too.

“All of these are domains Baidu is looking at using deep learning, actually,” Ng said.

Ng’s focus now might best be summed up by one word: accuracy.

That makes sense from a corporate perspective. Google has the brain trust on image analysis, and Microsoft has the brain trust on speech, said Naveen Rao, co-founder and chief executive of deep-learning startup Nervana. Accuracy could potentially be the area where Ng and his colleagues will make the most substantive progress at Baidu, Rao said.

Matthew Zeiler, founder and chief executive of another deep learning startup, Clarifai, was more certain. “I think you’re going to see a huge boost in accuracy,” said Zeiler, who has worked with Hinton and LeCun and spent two summers on the Google Brain project.

One thing is for sure: Accuracy is on Ng’s mind.

“Here’s the thing. Sometimes changes in accuracy of a system will cause changes in the way you interact with the device,” Ng said. For instance, more accurate speech recognition could translate into people relying on it much more frequently. Think “Her”-level reliance, where you just talk to your computer as a matter of course rather than using speech recognition in special cases.

“Speech recognition today doesn’t really work in noisy environments,” Ng said. But that could change if Baidu’s neural networks become more accurate under Ng.

Ng picked up his smartphone, opened the Baidu Translate app, and told it that he needed a taxi. A female voice said that in Mandarin and displayed Chinese characters on screen. But it wasn’t a difficult test, in some ways: This was no crowded street in Beijing. This was a quiet conference room in a quiet office.

“There’s still work to do,” Ng said.

‘The future heroes of deep learning’

Meanwhile, researchers at companies and universities have been hard at work on deep learning for decades.

Google has built up a hefty reputation for applying deep learning to images from YouTube videos, data center energy use, and other areas, partly thanks to Ng’s contributions. And recently Microsoft made headlines for deep-learning advancements with its Project Adam work, although Li Deng of Microsoft Research has been working with neural networks for more than 20 years.

In academia, deep learning research groups all over North America and Europe. Key figures in the past few years include Yoshua Bengio at the University of Montreal, Geoff Hinton of the University of Toronto (Google grabbed him last year through its DNNresearch acquisition), Yann LeCun from New York University (Facebook pulled him aboard late last year), and Ng.

But Ng’s strong points differ from those of his contemporaries. Whereas Bengio made strides in training neural networks, LeCun developed convolutional neural networks, and Hinton popularized restricted Boltzmann machines, Ng takes the best, implements it, and makes improvements.

“Andrew is neutral in that he’s just going to use what works,” Gibson said. “He’s very practical, and he’s neutral about the stamp on it.”

Not that Ng intends to go it alone. To create larger and more accurate neural networks, Ng needs to look around and find like-minded engineers.

“He’s going to be able to bring a lot of talent over,” Dave Sullivan, co-founder and chief executive of deep-learning startup Ersatz Labs, told VentureBeat. “This guy is not sitting down and writing mountains of code every day.”

And truth be told, Ng has had no trouble building his team.

“Hiring for Baidu has been easier than I’d expected,” he said.

“A lot of engineers have always wanted to work on AI. … My job is providing the team with the best possible environment for them to do AI, for them to be the future heroes of deep learning.”


Facebook invents an intelligence test for machines


John is in the playground. Bob is in the office. Where is John? If you know the answer, you’re either a human, or software taking its first steps towards full artificial intelligence. Researchers at Facebook’s AI lab in New York say an exam of simple questions like this could help in designing machines that think like people.

Computing pioneer Alan Turing famously set his own test for AI, in which a human tries to sort other humans from machines by conversing with both. However, this approach has a downside.

“The Turing test requires us to teach the machine skills that are not actually useful for us,” says Matthew Richardson, an AI researcher at Microsoft. For example, to pass the test an AI must learn to lie about its true nature and pretend not to know facts a human wouldn’t.

These skills are no use to Facebook, which is looking for more sophisticated ways to filter your news feed. “People have a limited amount of time to spend on Facebook, so we have to curate that somehow,” says Yann LeCun, Facebook’s director of AI research. “For that you need to understand content and you need to understand people.”

In the longer term, Facebook also wants to create a digital assistant that can handle a real dialogue with humans, unlike the scripted conversations possible with the likes of Apple’s Siri.

Similar goals are driving AI researchers everywhere to develop more comprehensive exams to challenge their machines. Facebook itself has created 20 tasks, which get progressively harder – the example at the top of this article is of the easiest type. The team says any potential AI must pass all of them if it is ever to develop true intelligence.

Each task involves short descriptions followed by some questions, a bit like a reading comprehension quiz. Harder examples include figuring out whether one object could fit inside another, or why a person might act a certain way. “We wanted tasks that any human who can read can answer,” says Facebook’s Jason Weston, who led the research.

Having a range of questions challenges the AI in different ways, meaning systems that have a single strength fall short.

The Facebook team used its exam to test a number of learning algorithms, and found that none managed full marks. The best performance was by a variant of a neural network with access to an external memory, an approach that Google’s AI subsidiary DeepMind is also investigating. But even this fell down on tasks like counting objects in a question or spatial reasoning.

Richardson has also developed a test of AI reading comprehension, called MCTest. But the questions in MCTest are written by hand, whereas Facebook’s are automatically generated.

The details for Facebook’s tasks are plucked from a simulation of a simple world, a little like an old-school text adventure, where characters move around and pick up objects. Weston says this is key to keeping questions fresh for repeated testing and learning.

But such testing has its problems, says Peter Clark of the Allen Institute for Artificial Intelligence in Seattle, because the AI doesn’t need to understand what real-world objects the words relate to. “You can substitute a dummy word like ‘foobar’ for ‘cake’ and still be able to answer the question,” he says. His own approach, Aristo, attempts to quiz AI with questions taken from school science exams.

Whatever the best approach, it’s clear that tech companies like Facebook and Microsoft are betting big on human-level AI. Should we be worried? Recently the likes of Stephen Hawking, Elon Musk and even Bill Gates have warned that AI researchers must tread carefully.

LeCun acknowledges people’s fears, but says that the research is still at an early stage, and is conducted in the open. “All machines are still very dumb and we are still very much in control,” he says. “It’s not like some company is going to come out with the solution to AI all of a sudden and we’re going to have super-intelligent machines running around the internet.”




【刘挺】哈尔滨工业大学教授,社会计算与信息检索研究中心主任,2010-2014年任哈工大计算机学院副院长。中国计算机学会理事、曾任CCF YOCSEF总部副主席;中国中文信息学会常务理事、社会媒体处理专业委员会主任。曾任“十一五”国家863 计划“中文为核心的多语言处理技术”重点项目总体组专家, 2012 年入选教育部新世纪优秀人才计划。主要研究方向为社会计算、信息检索和自然语言处理,已完成或正在承担的国家973课题、国家自然科学基金重点项目、国家863计划项目等各类国家级科研项目20余项,在国内外重要期刊和会议上发表论文80余篇,获2010年钱伟长中文信息处理科学技术一等奖,2012 年黑龙江省技术发明二等奖。


【刘挺】大家好,我是哈工大刘挺。感谢杨静群主提供的在线分享的机会。2014年11月1-2日,第三届全国社会媒体处理大会(Social Media Processing, SMP 2014)在北京召开,12个特邀报告,800多名听众,大会充分介绍了社会媒体处理领域的研究进展,与会者参与热情很高,2015年11月将在广州华南理工大学大学举办第四届全国社会媒体处理大会(SMP 2015),欢迎大家关注。








【刘挺】我们实验室是哈尔滨工业大学社会计算与信息检索研究中心,我们的技术理想是“理解语言,认知社会”。换句话说,我们的研究方向主要包括自然语言处理(Natural Languge Processing, NLP)和基于自然语言处理的社会计算,此次分享我重点谈自然语言处理。





【刘挺】相应的,2011年苹果公司在iPhone 4s系统里面推出Siri语音对话系统,引起业内震动。百度、讯飞、搜狗先后推出类似的语音助手。但近来,语音助手的用户活跃度一般,并没有达到预期的成为移动端主流搜索形式的目标。




【胡颖之】@刘挺 这个问题我们调研过,不知国外情况如何,大部分人觉得,在外对着手机自言自语有点尴尬,且往往还需要调整识别不准的词。如果是一问一答,就直接电话好了,不需要语音。



















【刘 挺】为什么要启动沃森、高考机器人这类的项目呢?要从搜索引擎的不足说起。海量数据搜索引擎的成功掩盖了语义方面的问题,在海量的信息检索中,有时候,数 据量的增加自然导致准确率的上升,比如问“《飘》的作者是谁”,如果被检索的文本中说“米切尔写了《飘》”,则用关键词匹配的方法是很难给出答案的,但由 于Web的数据是海量的,是冗余的,我们不断地追加文本,就可能在某个文本中找到“《飘》的作者是美国作家米切尔”这样的话,于是利用简单地字面匹配就可以找出问题和答案的关联,从而解决问题。因此,是海量的信息暂时掩盖了以往我们没有解决的一些深层问题,比如语义问题。








【刘 挺】问答系统有两大类:一类是以知识获取或事务处理为目的的,尽快完成任务结束问答过程是系统成功的标志;另一类是以聊天及情感沟通为目的的,让对话能够 持续进行下去,让用户感到他对面的机器具有人的常识与情感,给人以情感慰藉。我们认为微软“小娜”属于前者,“小冰”属于后者。
































【周志华】常识问题,图灵奖得主John MaCarthy后半辈子都在研究这个问题。悲观点说,在我退休以前看不到有希望。路过看到谈常识,随口说一句。



































【朱进】@杨静lillian 恕我直言,机器的智能筛选免不了是弱智的决定。只要编程这种形式存在,真正意义上的创造就很难想象。




语 音识别、图像识别属于感知层面,语言理解、图像视频的理解、知识推理与数据挖掘属于认知计算,在人类认知世界并认清规律的基础上,需要对社会发展的未来进 行预测并辅助决策,更高的计算则是创造计算,比如我们正在研制的机器自动作文属于创造计算。情感与认知密切相关,应该属于认知计算层面。









【朱 进】假如是个制作质量极差的电影,但是谁都没看过,制作方按常规方式宣传,机器能预测出来票房会极差吗?最简单的道理,完全同样的内容,换个相近的名字作 为新电影再放,机器会对票房给出跟第一次结果一样的预测吗?如果第三次换个名字再放哪?题目很牛,所有的宣传都很牛。问题是,预测的机器难道不需要先看一 遍电影再猜吗?另外,这机器真的能看懂电影吗?














【朱进】@白硕 按 我的理解,所谓的预测是在首映之前就做出来的。第一天的票房可以跟机器的预测一致。不过看电影的人又不是傻子,第一场一过,马上电影很臭不值得看的舆论就 传播出去了。后面的人还会按照之前的预测那样挤到电影院里吗?按我的理解,票房的关键还是片子的质量。可是片子的质量再没看到之前其实是不知道的。

【刘挺】@朱进 ,短期预测易,长期预测难,因为在事件演进过程中会有多种因素干扰。预测有两种,一种是事前预测,一种是在事件推进中根据已经获悉的事态对下一步事态进行预测。

【朱进】@刘挺 我咋觉得长期更容易猜准啊,因为时间对于涨落是有平滑的。






自然语言处理技术趋势:1. 从句法分析向深度语义分析发展;2. 从单句分析向篇章(更大范围语境)发展;3. 引入人的因素,包括众包等手段对知识进行获取;4. 从客观事实分析到主观情感计算;5. 以深度学习为代表的机器学习技术在NLP中的应用





百度语音识别新突破–Deep Speech系统


相关论文:DeepSpeech: Scaling up end-to-end speech recognition

Chinese search engine giant Baidu says it has developed a speech recognition system, called Deep Speech, the likes of which has never been seen, especially in noisy environments. In restaurant settings and other loud places where other commercial speech recognition systems fail, the deep learning model proved accurate nearly 81 percent of the time.

That might not sound too great, but consider the alternative: commercial speech-recognition APIs against which Deep Speech was tested, including those for Microsoft Bing, Google and Wit.AI, topped out at nearly 65 percent accuracy in noisy environments. Those results probably underestimate the difference in accuracy, said Baidu Chief Scientist Andrew Ng, who worked on Deep Speech along with colleagues at the company’s artificial intelligence lab in Palo Alto, California, because his team could only compare accuracy where the other systems all returned results rather than empty strings.

Ng said that while the research is still just research for now, Baidu is definitely considering integrating it into its speech-recognition software for smartphones and connected devices such as Baidu Eye. The company is also working on an Amazon Echo-like home appliance called CoolBox, and even a smart bike.

“Some of the applications we already know about would be much more awesome if speech worked in noisy environments,” Ng said.

Deep Speech also outperformed, by about 9 percent, top academic speech-recognition models on a popular dataset called Hub5’00. The system is based on a type of recurrent neural network, which are often used for speech recognition and text analysis. Ng credits much of the success to Baidu’s massive GPU-based deep learning infrastructure, as well as to the novel way them team built up a training set of 100,000 hours of speech data on which to train the system on noisy situations.

Baidu gathered about 7,000 hours of data on people speaking conversationally, and then synthesized a total of roughly 100,000 hours by fusing those files with files containing background noise. That was noise from a restaurant, a television, a cafeteria, and the inside of a car and a train. By contrast, the Hub5’00 dataset includes a total of 2,300 hours.

“This is a vast amount of data,” said Ng. ” … Most systems wouldn’t know what to do with that much speech data.”

Another big improvement, he said, came from using an end-to-end deep learning model on that huge dataset rather than using a standard, and computationally expensive, type of acoustic model. Traditional approaches will break recognition down into multiple steps, including one called speaker adaption, Ng explained, but “we just feed our algorithm a lot of data” and rely on it to learn everything it needs to. Accuracy aside, the Baidu approach also resulted in a dramatically reduced code base, he added.

You can hear Ng talk more about Baidu’s work in deep learning in this Gigaom Future of AI talk embedded below. That event also included a talk from Google speech recognition engineer Johan Schalkwyk. Deep learning will also play a prominent role at our upcoming Structure Data conference, where speakers from Facebook, Yahoo and elsewhere will discuss how they do it and how it impacts their businesses.

(2个打分, 平均:1.00 / 5)



邢波Eric P. Xing】清华大学物理学、生物学本科;美国新泽西州立大学分子生物学与生物化学博士;美国加州大学伯克利分校(UC,Berkeley)计算机科学博士;现任美国卡耐基梅隆大学(CMU)计算机系教授,2014年国际机器学习大会(ICML)主席。美国国防部高级研究计划局(DARPA)信息科学与技术顾问组成员。(他在中国大数据技术大会上的报告请参考阅读原文链接)

Professor of Carnegie Mellon University Program Chair of ICML2014

Dr. Eric Xing is a Professor of Machine Learning in the Schoolof Computer Science at Carnegie Mellon University. His principal researchinterests lie in the development of machine learning and statisticalmethodology; especially for solving problems involving automated learning,reasoning, and decision-making in high-dimensional, multimodal, and dynamicpossible worlds in social and biological systems. Professor Xing received aPh.D. in Molecular Biology from Rutgers University, and another Ph.D. inComputer Science from UC Berkeley.



【邢波Eric P. Xing】中国的大数据技术与题目跟进国外趋势还做得不错。但在原创性部分有欠缺。也许由于工程性,技术性上的原创工作通常不吸引眼球且风险极大这样的特点,所以没人愿意啃硬骨头。整体不算太差,但缺少领军人物,和领先的理念。还有在导向上,倾向于显著的效益和快的结果,但对于学术本身的追求不是很强烈。如果效果不是立竿见影,愿意碰的人就少。大部分人都这样,就是趋向于平庸。整个生态系统上看,中国大数据发展水平与欧洲、日本比并不差,公众的认知也热烈。整个环境还蛮好。与中国学生有点像,群体不见得差,但缺少特别杰出的领袖,和有胆识的开拓者。




【邢波 Eric P. Xing】人工智能的目标其实是没有上限的。人工智能的目标并不是达到动物或人类本身的智力水平,而是在可严格测量评估的范围内能否达到我们对于具体功能上的期待。例如用多少机器、多长时间能达成多少具体任务?(这里通常不包含抽象,或非客观任务,比如情绪,感情等。)人的智力不好评价,尤其标准、功能、结果及其多元,很多不是人工智能追求的目标。科幻家的浪漫幻想和科学家的严格工作是有区分的。大部分计算机科学家完成的工作可能不那么让人惊叹,但很多任务已经改变世界。例如,飞机自动驾驶装置可能没有人的智能,但它完成飞行的任务,却比人类驾驶员好。






【邢波 Eric P. Xing】很难科普的讲,不使用专业术语。机器学习不过是应用数学方法体系和计算实践的一个结合,包罗万象。比如图模型(深度学习就是其中一种),核(kernel)方法,谱(spectral)方法,贝叶斯方法,非参数方法,优化、稀疏、结构等等。我在CMU的机器学习课和图模型课对此有系统全面的讲解。




【邢波 Eric P. Xing】大数据、深度学习、高性能计算只是接触了机器学习的表层,远远不到收获红利的时候,还要接着往下做。算法的更新和变化还没有深挖,很多潜力,空间还很大。现在还根本没做出像样的东西。另外我要强调,机器学习的所谓红利,远远不仅靠“大数据、深度学习、高性能计算”带来。举个例子,请对比谷歌和百度的搜索质量(即使都用中文),我想即使百度自己的搜索团队也清楚要靠什么来追赶谷歌。



【邢波 Eric P. Xing】不太愿意评价同行的水平。人的智能和人工智能可以平行,也可以交互。



【邢波 Eric P. Xing】基因工程其实突破很多。在美国和全球转基因的食品也有很多。胰岛素等药物也是通过转基因菌株来生产,而不是化学合成。诊断胎儿遗传缺陷的主要手段也基于基因工程技术。但是舆论风向在变,也不理性。例如我小时候读的《小灵通看未来》里,“大瓜子”等神奇食品现在已经通过基因技术实现。从技术上看,我们已经实现了这个需要,但公众是否接受,是个问题。科学家要对自己的责任有所界定。例如造出原子弹,科学家负责设计制造,但怎么用是社会的事。





【邢波 Eric P. Xing】欧洲大脑工程的争议很大,包括目标和经费分配。但这个目标也提升了社会和公众的对于科学的关注,工程的目的不用过于纠结。这个项目就是启发式的,培养人才,培养科学实力的种子项目。





【李志飞】更正一下我的问题: 现有的机器学习算法如深度学习在利用大数据和高性能计算所带来的红利是不是遇到瓶颈了?(至少我所在的机器翻译领域是这样) 接下来会有什么新机器学习算法或原理会把大数据和高性能计算再次充分利用以大幅提升应用的性能?我觉得如果机器学习领域想在应用领域如机器翻译产生更大的影响,需要有更多人做更多对应用领域的教育和培训,或者是自己跨界直接把理论研究跟应用实践结合起来

【邢波 Eric P. Xing】机器学习的算法有几百种,但是目前在深度学习领域基本没有被应用。尝试的空间还很大,而且无需局限在深度学习下。一方面机器学习学者需要跨出自己的圈子去接触实际,另一方面应用人士也要积极学习,掌握使用发展新理论。



【邢波 Eric P. Xing】谷歌是最具有领导性的IT公司。世界上没有人工智能公司,公司不能用技术手段或目标定义名称和性质。人工智能是一个目标,而不是具体的一些手段。所以有一些界定是不严肃的。关于应用领域前面已经谈过了。




【邢波 Eric P. Xing】中国整个IT领域,以至科学界,应该百花齐放,有的观点占领了过多的话语权,别的观点就得不到尊重。目前业界形成一边倒的局面,媒体的极化现象比较严重。建议媒体应该平衡报道。中国目前深度学习话语权比较大,没人敢批评,或者其他研究领域的空间被压缩。这种研究空间的压缩对机器学习整个领域的发展是有害的。学界也存在有人山中装虎称王,山外实际是猫的现象。坦率的说,目前中国国内还没有世界上有卓越影响的重量级人工智能学者,和数据科学学者。中国需要更多说实话,戳皇帝新衣的小孩,而不是吹捧的大臣、百姓和裸奔的皇帝。不要等到潮水退去,才让大家看到谁在裸奔。




【邢波 Eric P. Xing】首先我没有那样说过,我的看法其实相反。即使给那些企业提供了大数据,他们真会玩么?这有点叶公好龙,作为一个严肃的研究,应该把工具造出来。得先有好的技术,别人才会把数据提供给你。有时小数据都没做好,又开始要大数据,没人会给。可以用模拟,更可以自己写爬网器(crawler)自己在网上抓。例如我们的实验室,学生就可以自己去找数据源。研究者的心态有时不正确,好像社会都需要供给他,自己戴白手套。其实人人都可以搭上“大数据”这个列车,但需要自己去勤奋积极努力。



【邢波 Eric P. Xing】希望如此。更客观地说,不是取代。是解决不同的问题,有很好的共生、互补关系。




【邢波 Eric P. Xing】希望中国人工智能学术界要对机器学习、统计学习的大局有所掌控,全面判断和寻找,尚未解决的难题。这需要很多人静下来,慢下来,多读,多想。而不是跟风或被热点裹挟。得有足够的耐心,屏蔽环境的影响和压力。在技术上得重视原创性,如果只把学术看成是一个短时期的比赛,价值就不大。得找有相当难度,而自己有独特资源的方向,就保证了思想的原创性和资源的独特性。要分析清楚自己的优势。





【邢波 Eric P. Xing】这算八卦。原来有几个流派,但现在流派的界限已经非常模糊了。



【邢波 Eric P. Xing】国内的同行思路有些短板,所以研究领域比较割裂。上层不够高,下层也不够深,横向也不宽,因此扎根不够,影响有限。所以比较缺憾,体现为很多割裂的领域。





【邢波 Eric P. Xing】不是没有可能,但也需要好的平台和环境、机缘。这次回国参会,很兴奋的是,学术界和产业界都对机器学习的技术有很大的热情,也有信念去获取成功,相当积极。我个人的观点,通过交流,收获很大。期望这种交流继续,也期待国内的学界、媒体、企业能够共同促进产业生态的发展,利益多样化。可以是金钱的成功,也可以是原创性的增长。而不是被某一个目标来一统天下。


下个月还有机会回国,到时也期待与大家继续交流互动。非常感谢@杨静lillian 提供这个和大家交流的机会。也钦佩她专业敬业。这次结识很多朋友,后会有期!


(3个打分, 平均:5.00 / 5)

The Wall Street Journal吴恩达专访


Six months ago, Chinese Internet-search giant Baidu signaled its ambitions to innovate by opening an artificial-intelligence center in Silicon Valley, in Google’s backyard. To drive home the point, Baidu hired Stanford researcher Andrew Ng, the founder of Google’s artificial-intelligence effort, to head it.

Ng is a leading voice in “deep learning,” a branch of artificial intelligence in which scientists try to get computers to “learn” for themselves by processing massive amounts of data. He was part of a team that in 2012 famously taught a network of computers to recognize cats after being shown millions of photos.

On a practical level, the field helps computers better recognize spoken words, text and shapes, providing users with better Web searches, suggested photo tags or communication with virtual assistants like Apple’s Siri.

In an interview with The Wall Street Journal, Ng discussed his team’s progress, the quirks of Chinese Web-search queries, the challenges of driverless cars and what it’s like to work for Baidu. Edited excerpts follow:

WSJ: In May, we wrote about Baidu’s plans to invest $300 million in this facility and hire almost 200 employees. How’s that coming along?

Ng: We’re on track to close out the year with 96 people in this office, employees plus contractors. We’re still doing the 2015 planning, but I think we’ll quite likely double again in 2015. We’re creating models much faster than I have before so that’s been really nice. Our machine-learning team has been developing a few ideas, looking a lot at speech recognition, also looking a bit at computer vision.

WSJ: Are there examples of the team’s work on speech recognition and computer vision?

Ng: Baidu’s performance at speech recognition has already improved substantially in the past year because of deep learning. About 10% of our web search queries today come in through voice search. Large parts of China are still a developing economy. If you’re illiterate, you can’t type, so enabling users to speak to us is critical for helping them find information. In China, some users are less sophisticated, and you get queries that you just wouldn’t get in the United States. For example, we get queries like, “Hi Baidu, how are you? I ate noodles at a corner store last week and they were delicious. Do you think they’re on sale this weekend?” That’s the query.

WSJ: You can process that?

Ng: If they speak clearly, we can do the transcription fairly well and then I think we make a good attempt at answering. Honestly, the funniest ones are schoolchildren asking questions like: “Two trains leave at 5 o’ clock, one from …” That one we’ve made a smaller investment in, dealing with the children’s homework. In China, a lot of users’ first computational device is their smartphone, they’ve never owned a laptop, never owned a PC. It’s a challenge and an opportunity.

WSJ: You have the Baidu Eye, a head-mounted device similar to Google Glass. How is that project going?

Ng: Baidu Eye is not a product, it’s a research exploration. It might be more likely that we’ll find one or two verticals where it adds a lot of value and we’d recommend you wear Baidu Eye when you engage in certain activities, such as shopping or visiting museums. Building something that works for everything 24/7 – that is challenging.

WSJ: What about the self-driving car project? We know Baidu has partnered with BMW on that.

Ng: That’s another research exploration. Building self-driving cars is really hard. I think making it achieve high levels of safety is challenging. It’s a relatively early project. Building something that is safe enough to drive hundreds of thousands of miles, including roads that you haven’t seen before, roads that you don’t have a map of, roads where someone might have started to do construction just 10 minutes ago, that is hard.

WSJ: How does working at Baidu compare to your experience at Google?

Ng: Google is a great company, I don’t want to compare against Google specifically but I can speak about Baidu. Baidu is an incredibly nimble company. Stuff just moves, decisions get made incredibly quickly. There’s a willingness to try things out to see if they work. I think that’s why Baidu, as far as I can tell, has shipped more deep-learning products than any other company, including things at the heart of our business model. Our advertising today is powered by deep learning.

WSJ: Who’s at the forefront of deep learning?

Ng: There are a lot of deep-learning startups. Unfortunately, deep learning is so hot today that there are startups that call themselves deep learning using a somewhat generous interpretation. It’s creating tons of value for users and for companies, but there’s also a lot of hype. We tend to say deep learning is loosely a simulation of the brain. That sound bite is so easy for all of us to use that it sometimes causes people to over-extrapolate to what deep learning is. The reality is it’s really very different than the brain. We barely (even) know what the human brain does.

WSJ: For all of Baidu’s achievements, it still has to operate within China’s constraints. How do you see your work and whether its potential might be limited?

Ng: Obviously, before I joined Baidu this was something I thought about carefully. I think that today, Baidu has done more than any other organization to open the information horizon of the Chinese people. When Baidu operates in China, we obey Chinese law. When we operate in Brazil, which we also do, we obey Brazil’s law. When we operate in the U.S. and have an office here, we obey U.S. law. When a user searches on Baidu, it’s clear that they would like to see a full set of results. I’m comfortable with what Baidu is doing today and I’m excited to continue to improve service to users in China and worldwide.





由于Spark出自伯克利大学,使其在整个发展过程中都烙上了学术研究的标记,对于一个在数据科学领域的平台而言,这也是题中应有之义,它甚至决定了Spark的发展动力。Spark的核心RDD(resilient distributed datasets),以及流处理,SQL智能分析,机器学习等功能,都脱胎于学术研究论文,如下所示:
Discretized Streams: Fault-Tolerant Streaming Computation at Scale. Matei Zaharia, Tathagata Das, Haoyuan Li, Timothy Hunter, Scott Shenker, Ion Stoica. SOSP 2013. November 2013.
Shark: SQL and Rich Analytics at Scale. Reynold Xin, Joshua Rosen, Matei Zaharia, Michael J. Franklin, Scott Shenker, Ion Stoica. SIGMOD 2013. June 2013.
Discretized Streams: An Efficient and Fault-Tolerant Model for Stream Processing on Large Clusters. Matei Zaharia, Tathagata Das, Haoyuan Li, Scott Shenker, Ion Stoica. HotCloud 2012. June 2012.
Shark: Fast Data Analysis Using Coarse-grained Distributed Memory (demo). Cliff Engle, Antonio Lupher, Reynold Xin, Matei Zaharia, Haoyuan Li, Scott Shenker, Ion Stoica. SIGMOD 2012. May 2012. Best Demo Award.
Resilient Distributed Datasets: A Fault-Tolerant Abstraction for In-Memory Cluster Computing. Matei Zaharia, Mosharaf Chowdhury, Tathagata Das, Ankur Dave, Justin Ma, Murphy McCauley, Michael J. Franklin, Scott Shenker, Ion Stoica. NSDI 2012. April 2012. Best Paper Award and Honorable Mention for Community Award.
Spark: Cluster Computing with Working Sets. Matei Zaharia, Mosharaf Chowdhury, Michael J. Franklin, Scott Shenker, Ion Stoica. HotCloud 2010. June 2010.
在大数据领域,只有深挖数据科学领域,走在学术前沿,才能在底层算法和模型方面走在前面,从而占据领先地位。Spark的这种学术基因,使得它从一开始就在大数据领域建立了一定优势。无论是性能,还是方案的统一性,对比传统的Hadoop,优势都非常明显。Spark提供的基于RDD的一体化解决方案,将MapReduce、Streaming、SQL、Machine Learning、Graph Processing等模型统一到一个平台下,并以一致的API公开,并提供相同的部署方案,使得Spark的工程应用领域变得更加广泛。
目前的Spark版本为1.1.0。在该版本的代码贡献者列表中,出现了数十位国内程序员的身影。这些贡献者的多数工作主要集中在Bug Fix上,甚至包括Example的Bug Fix。由于1.1.0版本极大地增强了Spark SQL和MLib的功能,因此有部分贡献都集中在SQL和MLib的特性实现上。下图是Spark Master分支上最近发生的仍然处于Open状态的Pull Request:
可以看出,由于Spark仍然比较年轻,当运用到生产上时,可能发现一些小缺陷。而在代码整洁度方面,也随时在对代码进行着重构。例如,淘宝技术部在2013年就开始尝试将Spark on Yarn应用到生产环境上。他们在执行数据分析作业过程中,先后发现了DAGSchedular的内存泄露,不匹配的作业结束状态等缺陷,从而为Spark库贡献了几个比较重要的Pull Request。具体内容可以查看淘宝技术部的博客文章:《Spark on Yarn:几个关键Pull Request(》。

机器学习应用–Smart Autofill

机器学习算法被广泛的应用在谷歌家的语言语音处理,翻译,以及视觉图像处理方面的应用上,看起来都是比较拒人千里的东西,但是最近,谷歌把这项技术用在了自家的Google Sheets上,貌似在我朝较难使用Google Sheets,但是感兴趣还是可以在Chrome Webstore里找到这款叫做Smart Autofill的插件试用一下。

那么Smart Autofill是干什么用的呢?顾名思义,它是用来填表的。经常用Excel的读者一定知道一个功能叫自动填充,能够填充的信息包括日历日期,星期,以及有序数字等。Smart Autofill干的是类似的事,但由于融入了机器学习,逼格又稍高,它可以根据表格中与缺失信息栏相关栏中的数据,学习其中的模式,推测出缺失信息栏中缺失的数据。

Smart Autofill使用了谷歌基于云的机器学习服务–Prediction API。这项服务可以训练多种线性或非线性,分类和回归模型。他会通过对比利用cross-validation算出的Misclassification error(针对分类问题)或RMS error(针对回归问题),自动选出最佳的模型,用于数据预测。


在下图的截图中,我们给出车的五个非常简单的数据,分别为使用购买年份,行驶里程,车门数量,车辆类型以及价格。因为车的价格可能和车的这些特质有关,因此可以把那些包含价格的行作为训练数据,用Smart Autofill来估测缺失的价格数据。