There’s More To Google’s Artificial Brain Than Finding Cats On The Internet | Fast Company


There’s More To Google’s Artificial Brain Than Finding Cats On The Internet

ebrain

Google’s got a brain. An actual electronic brain.

The New York Times has news that inside Google’s high-tech R&D “X” laboratory the search giant has been creating a simulation of the human brain. And rather than teaching it programs, Google’s staff have been exposing it to information from the Net so that it learns organically, a little like the way we humans do. It’s built by hooking together 16,000 processor cores with over one billion interconnections, in a notional model of the around 86 billion neurons in a typical adult human brain.

Some AI systems are all about code run in very fast computers, simulating the various layers of thought and decision that make up a mind with statistics or logic. But Google’s approach is a bit closer to a natural model where the inspiration isn’t some abstract algorithm for simulating a brain, but instead relies on building a replica of a brain and exposing it to raw information. Google has lots of information at its disposal.

In Google’s case they did no complex training, but simply exposed the brain to around 10 million random digital pictures extracted as thumbnails from YouTube videos and let it do its own thing, adjusting signals from some neurons up and down and strengthening and weakening some of the connections between them. It’s a concept well known to science fiction, and Douglas Adams even used it in The Hitchhiker’s Guide To The Galaxy, to describe the Deep Thought super computer “which was so amazingly intelligent that even before the data banks had been connected up it had started from I think therefore I am and got as far as the existence of rice pudding and income tax before anyone managed to turn it off.”

Google’s brain, more or less undirected through a process of repetition, developed a “concept” of human faces and the different parts of a human body from these images, and also a concept of cats. “Concept” here means a fuzzy ill-understood pattern that it could use to categorize a new image it had not seen before, based on its previous learning. The cats concept was a surprise to the researchers, but given the fact that YouTube is a skewed data set, and that we humans do love Lolcats and their like, perhaps it was inevitable.

So what Google’s done is develop a very simplified digital simulation of a human visual cortex. Given that such power is usually imagined as belonging to some military research facility, why’s Google trying it?

The answer is manyfold. In some sense, it’s a natural progression from much of the semantic web research Google’s been doing–investigating how to best process and interpret really human, natural language inputs so that it can deliver even more relevant web search results. Google’s Knowledge Graph is the most recent example of how powerful semantic search can be. The idea is that if you can better understand what someone actually means when they type data into Google, then you’ve got a better chance of actually delivering a matched set of answers in the search results.

A more complete artificial intelligence is simply the successor to these systems, because it would be able to make a guess at the meaning of a search term like “how many roads must a man walk down?” far beyond merely matching the words to the famous song, perhaps guessing that the milage of metaled roads in the U.S. may be useful data, or even engaging in a little discussion about the meaning of life or even the stupidity of Homer Simpson. Though this is a frivolous example, think about how you sometimes have to trawl through hundreds of Google search answers to find the one you want because searching for it isn’t straightforward. An AI search may well be swifter and more helpful.

But an AI trained like this would also make for an improved image recognition system, and also a much more astute voice recognition system. That could turbocharge the usefulness of search using text or imagery on your Android phone. And given what we know of Project Glass, Google’s effort to get us all wearing augmented reality goggles, a future Glass system hooked up to an AI that recognizes what the wearer sees and what they’re saying would seem an almost inevitable goal. Smarter AI could also help with Google’s self-driving cars project, perhaps resulting in safer drives or more efficient journeys.

Ultimately you have to wonder if Google’s system could plug into its Siri-like service, rumored to be codenamed Majel, to create a genuinely smart digital personal assistant. Fun though that is, we can also guess that Google would most likely use an AI for its own ends, to best work out what kinds of targeted ads to sell to its users.

[Image: Flickr user Saad Faruque, Google via New York Times]

Chat about this news with Kit Eaton on Twitter and Fast Company too.

Regards,

Reda Bouaichi
J2EE Web Developer
reda.bouaichi
773-574-0538

Advertisements


Gigaom

Very few companies know how to scale and deploy cloud applications like Netflix(s nflx), the ginormous movie streaming site. And now it’s making some of that cloud management expertise available to the masses via Github.

On Monday, the company open sourced Asgard,  a Grails and JQuery web interface that Netflix engineers use to deploy code changes and manage resources in the Amazon(s amzn) cloud in a massive way. The technology was named after the home of the Norse gods Norse god of thunder and lightning but was once known as the Netflix Application Console or NAC. And it offers some capabilities that the AWS Console does not.

Asgard, for example, helps engineers track the multiple Amazon Web Service components — AMIs, EC2 instances etc. — used by their application and manage them more efficiently.

As Joe Sondow, the Netflix senior software engineer who leads the project, wrote in the blog:

View original post 206 more words

More Cloud Wars: Another 50gb giveaway by Box.net


A little bit of this, a little bit of that

Box.net is giving away another 50gb free to its existing and new users. This coincides with their launch of the Box for Android app, which introduces bulk upload, commenting and collaboration features. To get this free upgrade, users have to access their Box.net account from an Android device.

 

 

A note – iOS users who availed of the free 50gb offer last year cannot use this offer to get a 100gb account.

 

 

 

 

 

 

Read more at Android Users Get 50GB FREE. Get Your #Box50GB and Enter the Box Mobile Giveaway | The Box Blog and Bonjour! Box for Android Adds New Collaboration Features and a 50GB Upgrade to Boost Productivity | The Box Blog.

 

 

View original post

Mu Sigma offers $108M reasons to believe in big data


Gigaom

Chicago-based big data firm Mu Sigma has closed a $108 million private-equity investment round to expand its analytics-outsourcing business. Armed with a team of data scientists and subject-matter experts, Mu Sigma takes customers’ data and it turns it into business insights, meaning customers don’t have to built their own in-house big data expertise. It’s an already-profitable business that will only get bigger.

The engineering, programming and math skills necessary to do big data analytics are in hot demand but short supply, which is what makes companies like Mu Sigma so appealing. All customers have to do is bring their data and their money, and the analytics experts go to work figuring out the best strategies for getting insights, and then run the big data workloads. Whereas using cloud-based resources eliminates the capital expense of big data projects, outsourcing eliminates the human resources expense, too.

It’s a fine business to be…

View original post 206 more words

4 Business Metrics You Can’t Afford to Ignore | Inc.com


http://pulse.me/s/4o1eH
4 Business Metrics You Can’t Afford to Ignore

OWNERS’ MANUAL

Profit and revenue tell you a lot–but they don’t tell you everything about the health of your business.

4 business metrics you can't afford to ignore

Flickr/MStewartPhotography

Every business focuses on and measures revenue. Every business focuses on profit and loss.

And they should, but there are a few other financial and performance measurements that can provide earlier warning signs of trouble—or early indications of longer-term success.

Here are four metrics your business can’t afford to ignore:

Cost to Acquire Customers (CAC). Also known as customer acquisition cost, this measures the cost of landing a customer. In simple terms, add up the cost of marketing and sales—including salaries and overhead—and divide by the number of customers you land during a specific time frame.

Spend $100 and acquire 10 customers and your CAC is $10.

What’s a good number? That depends on your industry and business model. It’s also important to understand how CAC fits into your overall operating budget. The leaner your operation the more you can afford to spend to acquire a customer.

Also keep in mind that a high CAC makes sense if you also generate a high…

Lifetime Value of a Customer (LTV). Unless your business is truly one-off, some percentage of customers will become repeat customers. The more repeat customers you have, and the more those customers spend, the higher CAC you can afford. (Some business models are built on breaking even on the customer’s first purchase; future purchases will be profitable since the CAC is at or near zero.)

LTV is often tricky to calculate and does involve making a few assumptions, at least during the startup phase. But once you’ve built a little history you can start to spot customer retention and spending trends. Then the math gets a lot easier: Determine what the average customer spends over a specific time period and calculate the return on your original CAC investment. Sense-check that against your profit and loss statement. Roughly speaking, the greater the LTV, the higher CAC you can afford.

Why do these two metrics matter so much? A rising CAC means you’ll need to start cutting costs or raising prices—or do a better job in marketing and sales. A falling LTV indicates the same measures are necessary… and means you’re failing to leverage the most important and least expensive customers you have: current customers.

Churn rate. Every business gains and loses customers; that’s a fact of business life. But still, lost customers are like failed investments. You spent money to acquire them, service them, and try to retain them… and now they’re gone.

A rising churn rate could be caused by a number of factors: Dissatisfaction with your products and services, new competition in your market, or even the coming end of a product or service cycle.

Churn rate is a solid indicator of rising CAC and lower LTV. In fact, all three are great leading indicators of problems—or successes—to come, both in other metrics and for your business overall.

Revenue percentages. Very few businesses only have one source of revenue. Most have multiple sources, and changes in the contribution percentage each makes can indicate problems are ahead.

Take wedding photography, a business I know something about. To keep things simple, say 80 percent of revenue historically comes from the initial wedding package sold to couples, 10 percent from additional sales after the wedding to the couple, and 10 percent from post-wedding sales to friends, family, etc. If post-wedding sales fall off that will impact overall profit levels since almost all marketing and sales costs go into booking weddings so margins on additional sales are naturally much higher.

Changes in revenue percentages can often signal not only changes in customer spending habits but also broader trends in your industry and market.

If you have other key metrics your business follows, share them in the comments below.

Read more:

5 Jobs You Should Have Before Starting a Business

Best Way to Become an Expert

Best 2011 Books for Entrepreneurs
jeffhaden.bucket_11526.jpg

Jeff Haden

Jeff Haden learned much of what he knows about business and technology as he worked his way up in the manufacturing industry from forklift driver to manager of a 250-employee book plant. Everything else he picks up from ghostwriting books for some of the smartest innovators and leaders he knows in business. He has written more than 30 non-fiction books, including four Business and Investing titles that reached #1 on Amazon’s bestseller list. He’d tell you which ones, but then he’d have to kill you.

Regards,Reda Bouaichi

Facebook to reach a value of $100bn in 2012 | •…BitterBeam…•


http://bitterbeam.wordpress.com/2011/11/29/facebook-to-reach-a-value-of-100bn-in-2012/

Facebook to reach a value of $100bn in 2012

Facebook

Facebook is planning to launch onto the stock market in the second quarter of 2012 and it is expected that this initial public offering will take the company value of facebook over $100bn. YES! you read it right, 100 Billion American Dollars.

Mark Zuckerberg

Facebook CEO Mark Zuckerberg

Facebook CEO Mark Zuckerberg has managed to keep his company off the stock market than many anticipated he would. However the growing number of shareholders in the company has forced him to change his mind. The company will be crossing the mark of 500 shareholders by the end of the year, which means it will be obliged to publish its finances in April.

However the critics suggest that this move will definitively increase the company potential to do major expands in the future in the current market while increasing the company value.

•…BitterBeam…•

Aside

•…BitterBeam…•


USB Drives come in different shapes and sizes, but they all give you the same 2 choices. You can either buy one big enough to store all your data or you can keep a bag full of drives to keep your data separately. But  Designer Hyunsoo Song’s creative mind has other thoughts as he proposed an alternative with this so-called Amoeba modular USB flash drive, which enables people to store data on individual USB drives that can be used both on their own or together as one large USB drive. The Idea behind this is you can keep the drives together most of the time detach the appropriate section if you want to share only a specific data section. Although this is still at the “Concept” Stage all we can say is that this concept is not too far from a future possibility.

•…BitterBeam…•

Via engadget
Source yankodesign

View original post