Leading artificial intelligence experts are investigating ways to commercialize a rapidly emerging sub-field of research known as “deep learning.” This month, a research team under renowned scientist Geoffrey E. Hinton’s tutelage won a prize sponsored by Merck to design software to uncover molecules that are most likely to be good candidates for new drugs.

The win was a particularly impressive feat given that the team entered at the last minute and was working with relatively small data-sets. Click here to read more about “how they did it.”

In a story originally reported by the New York Times, even skeptical scientists admit this is a significant advancement. “The kind of jump we are seeing in the accuracy of these systems is very rare indeed,” NYU computer scientist Yann leCun told the Times.

Silicon Valley’s technology companies have used artificial intelligence technology for several years. Deep learning is yielding new discoveries in fields like speech recognition and computer vision. It is already used in Apple’s Siri virtual personal assistant and Google Street View.

This is just the beginning. Today’s image recognition systems do not use human-familiar concepts like ‘mouth, or ‘eyes’ but statistical properties derived from the image. Deep learning is based on learning several levels of representations, and higher-level concepts are defined from lower-level ones.

“The innovation of deep learning is that it not only arranges these properties into hierarchies, but it works out how many levels of hierarchy best fit the data,” wrote Tom Stafford and Matt Webb, neuroscience researchers from Mind Hacks.

Recently, deep learning systems have been able to outperform humans. A team at the Swiss AI Lab at the University of Lugano won a pattern recognition test to identify images in a database of traffic signs against a human expert.

In the future, science writer John Markoff posits, deep learning will make surveillance technologies cheaper and more accessible, help marketers comb through data to identify consumer buying patterns, and may also pave the way for self-driving cars and robots that can replace human workers.

Machine Image via Shutterstock

Cortica’s image recognition technology fuses neuroscience and computer science by giving computers the ability to comprehend visual content on the web in real-time. It was developed at Technion, a premier technology institute in Tel Aviv, by a crack team of neuroscientists, digital media experts, and former military intelligence personnel. The technology functions similarly to the human cortex and can identify patterns, make classifications, and understand James Joyce. Maybe not the last one.

With the money, members of the Cortica team will relocate to the US, with people in New York and Silicon Valley.

July 9-10, 2013
San Francisco, CA

Early Bird Tickets on Sale

“Wine is, in truth, the only antidote to the bane of whiskey” said Thomas Jefferson. Copenhagen-based Vivino helps you identify and locate your favorite antidote for future consumption. The company just received an angel investment from Skype co-founder Janus Friis as part of its strategy to go global.

You can use the free Vivino app to take a photograph of any wine label. Vivino uses image recognition to match the label against its database of 450,000 wines and return you all the information available about the wine including, hopefully, where you can buy it locally. The app saves the details of all wines you have matched and you can also record whether you liked a particular wine or not. The business model is based on a revenue share with wine distributors.

“We’re the bridge between tasting a great wine and buying it” says Vivino co-founder Theis Sondergaard. Songergaard and co-founder Heini Zachariassen claim to reflect their target audience quite accurately. “We really like wine, but we don’t know that much about it. We know what we like, when we taste it, but we don’t have a wine-cellar, just a row of bottles in the kitchen.” continues Sondergaard. “We use the Vivino app to educate ourselves a bit about the wines we drink, remember what wines we’ve enjoyed and want to try again, and the ones to avoid. “

Swiss company Kooaba provides the image recognition technology. Kooaba can match against an image-set of 2 million labels in less than a second. Wine labels shot from different angles or in different lighting conditions can be matched; an important feature when dealing with images taken in restaurants or low-lit rooms.

Currently Vivino can automatically match 60 percent of wine labels and the company aims to get up to 80 or 90 percent. When a label can’t be matched Vivino’s data team goes into Sherlock Holmes mode and tracks down the details within 24 hours to make a manual match. The most common matching error (2-4 percent of matches)  is the vintage since this is often in small font on the labels. The data team also checks every automatic match to correct such errors. Vivino plans to combine object character recognition, basically a form of text recognition, with Kooaba’s image recognition to improve matching further.

Vivino isn’t alone in this space. Social wine review site Snooth launched its own wine recognition app in September, 2010. Snooth has a larger database of wines than Vivino and has a complementary and well-established review web site. However, it is very focused on the U.S. market. Snooth’s app is also only available for the iPhone whereas Vivino also supports Android, Blackberry and Windows 7. If you are an Android-owner who lives in Amsterdam, where Vivino lists several locations to buy the wines it identifies, there isn’t much contest.

“Nobody has really claimed the top spot for being THE wine app.” says Sondergaard. Being the top global wine app is Vivino’s goal. Vivino’s most comprehensive coverage of wine outlets is currently in its home market of Denmark. Sondergaard told me that the Danish market is relatively complex, with many small players and exclusivity deals, making it a good test market.

I asked Sondergaard about Vivino’s global ambitions. “This is a global concept but we need to roll out the business market by market like Groupon. Unlike Groupon we acquire customers via our free app. We start by creating the order and then figure out how to fulfill it (via local wine distributors)”. Vivino is already a global company in that the development and data teams are located in Ukraine, Macedonia and India. According to Sondergaard “Being based in Europe gives us a certain advantage in European markets”. Every European country has a different language, taxes, shipping options and even currency. European developers have to internationalize from day one.

However, Vivino’s next big target is the U.S. where it plans to roll out the “buy” feature in the next few months. The angel investment it recently received from Skype co-founder Janus Friis will help. “In 2002, Heini and I, along with Morten Lund, founded BullGuard, a consumer internet security company. Back then, we partnered with Kazaa, and even though Janus was no longer involved in Kazaa at that time, we got to know him. So when it was time to get an angel investor for Vivino, we knew who our first choice was.”

Cracking the U.S. market is the company’s biggest goal for the next year. New features like special or group buying offers, based on which wines are popular in particular markets, are also on the agenda. But it won’t be all work and no play. “We want to drink some amazing wine and have a blast while we do it” concludes Sondergaard.