Monday, 18 February 2013

Google X's neural network


At a time when Research & Development divisions are frequently seen as luxury for companies, Google had established one of them a few years ago in the most secret way possible. Google X Lab, the young R&D division of Google, is charged with confidential projects such as Google Glass or driverless cars, and only little information is filtered.  Even though those two projects sound very cool and will probably be part of our future, I would like to bring your attention to another project which is the Google Brain project. It consists of creating large scale artificial neural networks, exploiting the ridiculous amount of processing power that Google owns.

Using 1000 machines composed of 16 cores each, Google X Lab created an unsupervised self-learning neural network [1] which was fed with pictures extracted from Youtube videos. After training, the neural network managed to categorise images such as human faces or cats with a satisfying percentage of success. In other words, Google X’s neural network was able to create new concepts without any previous data or human intervention during the learning process, which could mean that the artificial neural network was able to generate new knowledge by itself and somehow “understand” its environment. Of course, this kind of conclusion can sound a bit hasty and ambitious but the results are quite surprising, considering that a normal human brain has around 100 billion neurons for 100 trillion synapses while this network had only 1 million synapses.

It is important to emphasise the fact that the neural network was asked to learn from visual inputs while it would have been easier to make it learn some concepts from audio inputs, which would lead to the possibility of translating concepts from a language to another, instead of translating literally. Nonetheless, Google X Lab is now using the same neural network for speech recognition system so we can expect them to work on such an idea.

We could then wonder what the result would be if they could build a neural network composed of 100 trillion synapses, even though it is true that the cognitive model of artificial neurons does not properly match that of biological neurons.

[1] Le, Quoc V., et al. "Building high-level features using large scale unsupervised learning." arXiv preprint arXiv:1112.6209 (2011).


4 comments:

  1. Here is a blog piece that I found that was posted today entitled 'How Google Retooled Android With Help From Your Brain'.

    http://www.wired.com/wiredenterprise/2013/02/android-neural-network/

    It is hard not to appreciate the work of Google, IBM and Microsoft in relation to neural networks as these advances are really contributing to how we interact and behave around technology. I am really interested in the whole speech recognition area. As it states in the blog post i linked to, "In October, Microsoft Chief Research Officer Rick Rashid showed a live demonstration of Microsoft’s neural network-based voice processing software in Tianjin, China. In the demo, Rashid spoke in English and paused after each phrase. To the audience’s delight, Microsoft’s software simultaneously translated what he was saying and then spoke it back to the audience in Chinese. The software even adjusted its intonation to make itself sound like Rashid’s voice". As the blog goes on, it states that this technology looks very promising and that in a few years we will be able to break down the language barriers between people. Some may say in a sense that the future is already here but with continued advances, it will be very interesting to see what this may mean for learning a new language or what it will mean for our linguistic lives! What will the consequences be when a reinvented Watson attains clear semantic understanding? Just some thoughts.

    ReplyDelete
  2. That’s really interesting. I didn’t realize voice recognition software had reached such a level of sophistication with
    simultaneous translation and interpreting. I watched the video about Rick Rachid, and although it is clear that the system still makes errors (it seems to be especially lousy at detecting whether the speaker is still on the same sentence or starting a new one), it made me wonder about the future of the translation and interpreting business. Will machine translation one day become so efficient as to render human translators and interpreters obsolete? I am particularly interested in how it will perform on translations with a more literary content like poetry.

    And what about language teachers, in a world where translating can be performed effortlessly by machines, will we still need to learn other languages?
    As a first step in that direction, last summer, Will Powell, a UK-based inventor came up with a system that translates both sides of a conversation between English and Spanish speakers—if they are patient, and speak slowly. Each interlocutor wears a hands-free headset linked to a mobile phone, and wears special goggles that display the translated text like subtitles in a foreign film:

    http://www.huffingtonpost.com/2012/07/23/google-glass-inspired-specs-auto-translate_n_1695008.html

    ReplyDelete
  3. Yes we'll still need to learn new languages.... Like java, scheme and church etc! I've been using google translate for years now and since I started using it its come on in leaps and bounds, before I would only use it to confirm whether I was right about a particular word but now you can use it reliably enough to translate whole sentences, as long as none of the terms are overly ambiguous. I do think that translators will be pretty much redundant before too long, but they're always be as much a need for interpreters and for learning new languages as there is today unless google or someone else develops a babel fish
    http://en.wikipedia.org/wiki/Babel_fish_(The_Hitchhiker%27s_Guide_to_the_Galaxy)#Babel_fish

    ReplyDelete
  4. Rumours have been floating about for a while about Google working with the CIA to improve their analysis of data which could be linked to terrorism or other forms of crime. One link I found talks about a company called which looks at trends in such a profound way that it can "predict the future". Google already reads my e-mails. So what will happen if I use the wrong combination of words in my e-mail and get police kicking down my front door? Slight exaggeration, but you get my meaning .
    See link below and the video showing "recorded future" which is quite nice
    http://www.wired.com/dangerroom/2010/07/exclusive-google-cia/

    ReplyDelete