July 9-10, 2013
San Francisco, CA

Tickets On Sale Now

Windows announced Bing Translator for Windows today. It’s a touch-enabled version of Bing Translator for Windows Phone that is now available for Windows 8 laptops, desktops, and tablets.

Bing Translator will translate more than 40 languages on web pages, in images, or in what Microsoft calls an augmented reality translation via a live video stream from your device’s camera:

The app, which Microsoft says is based on years of research into advanced machine learning, is free. It does use Internet access, but because that can be difficult or expensive when traveling, users can download offline language packs for the most common languages, such as French, Chinese, German, Italian, and Spanish. More offline languages are coming soon, Microsoft says.

Perhaps the most interesting thing about Bing Translator is that with this app, Microsoft has enabled fully-system-integrated translation.

“The Share Charm lets you quickly translate highlighted text in any Windows 8 app; with Snap View you can multitask while browsing, chatting, or more by snapping Bing Translator to the right or left of your screen,” Vikram Dendi, Microsoft’s director for Bing Translator wrote in an announcement.

And, of course, when on the go with your Windows tablet, you can also use Bing Translator to speak the translation out loud in certain languages, allowing you to communicate back to people even if you don’t know their language. Google Translate for Android has had this feature for several months now.

Bing Translator supports 42 languages — 43 including Klingon — eight fewer than Google Translate.

Here’s the full list:

Arabic, Bulgarian, Catalan, Chinese Simplified, Chinese Traditional, Czech, Danish, Dutch, English, Estonian, Finnish, French, German, Greek, Haitian Creole, Hebrew, Hindi, Hmong Daw, Hungarian, Indonesian, Italian, Japanese, Klingon, Klingon (Kronos), Korean, Latvian, Lithuanian, Malay, Norwegian, Persian, Polish, Portuguese, Romanian, Russian, Slovak, Slovenian, Spanish, Swedish, Thai, Turkish, Ukrainian, Urdu, Vietnamese

Image credit: Microsoft

Berkeley-based Wise.io has launched its first product for data scientists and developers as part of a goal to make machine learning technology more accessible.

The idea for the initial “machine learning as a service” product came from the founders’ experiences on Kaggle, the website that 70,000 data scientists use to compete to solve tricky math problems.

CEO Joshua Bloom noticed that data scientists were using a specific algorithm on Kaggle to win competitions, but it was clunky to use. “We wanted to make it as fast as possible and in-memory efficient,” Bloom told me. The Wise.io team [above] have some impressive credentials: They have Ph.Ds in statistics, computer science, and astrophysics.

An added benefit: Many of the data scientists that are using Kaggle have day jobs at large companies. To reach enterprise customers, Bloom is hoping that these developers will push their IT departments to buy a site license.

“For now, the lowest barrier to entry are those who are using machine learning algorithms in their daily lives,” said Damian Eads, Wise.io’s cofounder and engineering lead. Eads would not disclose the existing customers of the new product, but he said the company is already profitable, a rare feat in Silicon Valley.

The company also plans to make money by offering professional consulting services to large companies who are looking to invest in a machine learning implementation.

Wise.io doesn’t have any immediate plans to raise institutional funding, and is an entirely bootstrapped effort.

This is a guest post by Walt Maguire, Analytics Director at ParAccel

The emphasis on “big data” has grown mightily over the last year, as more companies strive to draw useful intelligence out of increasingly massive data volumes from web clickstreams, sensor data, social media data and other large datasets.

One technology approach has dominated the discussion: MapReduce. MapReduce is open-source technology used for distributed programming, and its current incarnation “Hadoop” (named for its inventor’s son’s stuffed elephant), has been trumpeted as the new solution on the scene, the silver bullet for getting value from big data.

But while MapReduce and Hadoop are interesting and useful, the approach is nothing new, nor a panacea. While often cost-effective for inexpensive data storage and lightweight data processing, running analytics on Hadoop data has been challenging. Early adopters report that analytics in Hadoop are very slow to process — a big problem for analyzing giant data sets — and complex to write, due to not supporting SQL (structured query language, the lingua franca of analysts.)

Newsflash: Other technologies that solve many of the same problems have existed for decades, namely Massively Parallel Processing (MPP) databases, which are known for speedy processing of analytics and robust SQL support.

However, the MPP and Hadoop approaches are not mutually exclusive. Hadoop and MPP databases are increasingly used together by forward-thinking companies for a complete big data infrastructure that is cost-effective and leverages the best of both technologies.

Let’s compare the two approaches and look at a few specific examples of how they can be combined.

Map/Reduce and Hadoop evolution

At the heart of MapReduce are two functions called, unsurprisingly, Map and Reduce. A Map function’s role in life is to take some input data such as a list of words, apply some function and then map those inputs to output data. A Reduce function will take the outputs from a Map, and apply a function to reduce the input data into usable output data.

In the world of big data, divide and conquer is a must if we’re to cope with the data volumes generated today.

Example of a Map/Reduce function:

The initial driver behind the development of MapReduce was a paradigm shift in computer programming during the 1990s towards an approach called “functional programming.” Not long after it was first used at Google to speed up its indexing of the World Wide Web in 2004, the open source MapReduce platform, Hadoop, was developed. Hadoop delivered a reasonably complete way to develop distributed MapReduce programs. It had numerous gaps, but for those analyzing 10,000x as much data as they were five years ago, it helped.

The recent uptake of Hadoop has been driven in part by necessity. With the exponential growth of the Internet, machine data and the trend toward “saving everything,” organizations have more data than ever before, much of it in unstructured forms. So an innovation first created as a programming technique has been pressed into service as a specialized platform for distributed data processing. While it’s good at functions traditionally performed by ETL tools, it’s not as good at providing fast answers to questions.

Organizations risk finding themselves with a large repository of data in Hadoop that they can’t analyze very well.

Massively Parallel Processing (MPP) evolution

For many of the tasks necessary in processing and analyzing big data today, the Massively Parallel Processing (MPP) database is better. MPP databases also split up complex, large volume jobs into units processed across multiple nodes. While they don’t act exactly like MapReduce, they accomplish many of the same things, and are far better at some things.

MPP databases provide things taken for granted by database users for decades such as ACID compliance — meaning you will get predictable answers to questions. This isn’t enforced in Hadoop. Also, MPP databases include cost-based optimizers and monitor the distribution of data within the system; and as a result, they are generally an order of magnitude more efficient than Hadoop. So you can do things ten times more quickly, or do the same thing with one-tenth the infrastructure.

MPP databases do not solve every problem. For example, when the structure of incoming data is unknown or variable, an MPP database requires that this be structured at load time. So a measure of data manipulation must take place to prepare it. Also, appliance-based MPP systems can be difficult and costly to expand, whereas Hadoop is designed to run on any hardware. Software-based MPP database solutions don’t have this problem.

The following table compares and contrasts Hadoop/MapReduce with MPP databases.

Comparisons from the real world

Many firms have brought in both technologies for their big data infrastructure.

One large retailer found that Hadoop and an MPP platform are complementary. The company ingests large amounts of unstructured data and archives it at low cost with Hadoop; it then loads the data needed for analytics into the MPP platform via the vendor’s proprietary, high-speed integration module. Now, this retailer can run jobs 200x faster than its previous data warehouse, enabling more granular market basket analysis and customer segmentation. It leveraged Hadoop for low-cost storage and an analytic platform for doing the actual analysis,  cost-effectively solving a number of key problems with this combination of technologies.

This is a common model these days. Evernote, a Redwood City, Calif.-based developer of note taking and organization software, has a similar architecture, using Hadoop for low-cost data storage and processing of web application log data, combined with an MPP platform for analytics. For them, it was faster to move the data to a platform purpose-built for analytics than to try to run the analytics within Hadoop.

Evernote CTO Dave Engberg provides much more detail and a summary in the company’s Tech Blog:

Hadoop is great for cheaply storing a ton of data and performing parallel batch processing jobs in minutes instead of hours (or days)…But it’s not particularly quick for more complicated analyses that combine multiple different sets of data.…

Overall, the new infrastructure has met our goals. We can load and transform hundreds of millions of records in two hours instead of 10+, we’re generating far more (and far better) reports, and we can safely perform much more complex analyses of user trends than we could before.

In summary, there is a useful place for MapReduce and Hadoop in the big data landscape, but MPP technologies also offer significant advantages. Companies should strongly consider using both together to deliver big data infrastructures.

A startup called Sift Science says it can help your website battle fraud in 30 minutes or less.

And the San Francisco-based company closed a $5.5 million funding round today from quite a list of high-profile tech investors.

The cofounders have developed an algorithm that pinpoints one million patterns that help predict, and potentially prevent, incidences of fraud. For instance, those who type text in all caps are almost five times more likely to be fraudsters.

Investors included Salesforce CEO Mark Benioff, Expedia founder Richard Barton, Posterous’ Garry Tan, former Paypal CTO Max Levchin (who led the seed round), Tapjoy cofounder Lee Linden, Y Combinator, First Round Capital, SV Angel, Union Square Ventures, and Reddit’s Ohanian.

Fraud image via Blazej Lyjak// Shutterstock

July 9-10, 2013
San Francisco, CA

Tickets On Sale Now

HG Data uses a data-driven approach to help salespeople build lists of potential enterprise customers, and is ready to disrupt the legacy IT market.

The Santa Barbara-based startup today closed a $2 million funding round led by EPIC Ventures.

“While LinkedIn maps the connections between people in business, HG Data maps the connections between businesses themselves – who sells to whom, who buys what, and who partners with whom,” said Craig Harris, CEO and founder of HG Data in a statement.

The company has been building out its algorithms for the past two years. More specifically, it uses machine learning technology to updates its databases with information that is publicly available from press releases, white papers and the like. One of HG Data’s customers might use the tool to pinpoint a company that still runs on a specific type of legacy hardware, and pitch them an alternative.

The company claims its customers can use the data to bolster lead generation, market share analysis and marketing automation.

Angel investors in the round include Kevin O’Connor, founder of DoubleClick and current CEO of FindTheBest; Eric Kanowsky and AJ Rice, both co-founders of Software.com; and Tim Baskerville, former CEO of JupiterResearch.

Every tech startup and large company is on the hunt for a data scientist. The Harvard Business Review dubbed it the sexiest job of the 21st century.

A San Francisco-based startup called Kaggle is taking on the talent shortage, whether through its data science competitions or a new service launching today, “Kaggle Connect.” 

“We want to offer a real genuine source of work for a lot of data scientists and something to be proud of,” said Kaggle cofounder and CEO Anthony Goldbloom in an interview with VentureBeat.

Kaggle’s data scientists are usually Ph.D students that work on tricky challenges and earn cash rewards. The company is best known for its data science competitions, which companies like GE and Pfizer sponsor. One such competition involves identifying patients who will be admitted into hospital in the next year based on historic claims data. A cash prize of $3 million is up for grabs for the winning data scientist.

But not all problems are solved with a competition or by making a new hire. With Kaggle Connect, companies are directly matched with a top-tier data scientist, who takes on a temporary consulting role. Kaggle’s team of developers have built a secure cloud-based workspace and a suite of tools that incorporate the best practices they have seen in previous competitions.

Kaggle charges upward of $30,000 a month for the service (shared between the company and the data scientist), which may seem like a lot. However, Goldbloom said they are solving a problem that is causing companies’ pain. Developers are increasingly rebranding themselves as data scientists to earn a higher salary, but some don’t have the requisite skill set.

Kaggle’s secret is its network of high-performing data scientists and statisticians from around the world. Goldbloom told me that they regularly use their “Kaggle ranking” as a recognized credential when applying for a job.  

“Companies are getting bitten by hiring a data scientist who isn’t really a data scientist,” said Goldbloom. “That’s why we feel that Kaggle Connect is a good fit for the market.”

This is a guest post by technology executive John De Goes

“Big data” is dead. Vendors killed it. Well, industry leaders helped, and the media got the ball rolling, but vendors hold the most responsibility for the painful, lingering death of one of the most overhyped and poorly understood terms since the phrase “cloud computing.”

Any established vendor offering a storage or analytics product for a tiny or a large amount of data is now branded as big data, even if their technology is exactly the same as it was 5 years ago (thank you, marketing departments!). Startups, too, lay claim to the moniker of “big data app” or “big data startup,” eager to soak up some of the big data money floating around in big data-focused VC funds.

The phrase “big data” is now beyond completely meaningless. For those of us who have been in the industry long enough, the mere mention of the phrase is enough to induce a big data headache — please pass the big data Advil. (Editor’s note: We couldn’t agree more!)

If you want proof, witness the rising tide of backlash against the term:

• “Every so often a term becomes so beloved by media that it moves from ‘instructive’ to ‘hackneyed’ to ‘worthless,’ and Big Data is one of those terms….” Roger Ehrenberg
• “Every product by every vendor supports big data… and every ‘industry leader’ with every talk needs to include the phrase in the title of their talk and repeat it as many times as possible. So every data warehouse pitch is rehashed as a big data pitch, every data governance, master data management, OLAP, data mining, everything is now big data.” Rob Klopp
• “Big data as a technological category is becoming an increasingly meaningless name.” Barry Devlin

RIP Big Data.

Now that big data is dead, we’re free to move onto the next chapter of our lives. Which, from a data perspective, means we can stop worrying about the volume, variety, velocity, veracity and verisimilitude of data (just put it in Hadoop already!), and begin focusing on ways to impact bottom-line metrics by leveraging the talent, tools, and technologies that are slowly making their way into mainstream.

As the industry matures, there won’t be a single term that replaces the big data moniker. Instead, different tools and technologies will carve out different niches, each more narrowly focused and highly specialized than the universal sledgehammer that was big data.

I’m going to talk about some of the niches you’re going to hear about again and again. Alas, some of these will be spun into buzzwords that, like big data, accumulate so much “momentum” they eventually lose meaning. But for now, they should give you a glimpse into what lies ahead in the future of data storage, processing and analysis.

Meaningful jargon for 2013

I’ve identified six different aspects of data that you’re going to hear more frequently in 2013. Each of these terms actually conveys useful information, and cuts across slices and use cases that fall under the rubric of “big data”.

Smart Data?

Various industry leaders, writers, speakers, and pollsters (see here, here, here, here, here, here, and here) have started using the term “smart data” to refer to an increasingly common pattern emerging in the big data scene involving the productization of persistent data through predictive analytics.

In essence, companies are moving beyond BI, which relies on humans to interpret data, and are looking to monetize their vast troves of machine-captured data through predictive analytics (which relies on advanced techniques in statistics and machine learning to recognize and exploit patterns). These predictive analytics are often deployed as revenue-generating, intelligent features inside products, such as fraud detection, recommendations, personalizations, ad targeting, and much more. Examples of companies leveraging smart data include Netflix, Amazon, Rich Relevance, Gravity, LinkedIn, SailThru, and many more.

Data Science?

Data science is a new field that employs advanced techniques in statistics, machine learning, natural language processing, and computer science to extract meaning from large amounts of data (sometimes with the goal of creating new data products — arguably the reason data science was created). Though still meaningful, this term is starting to come under abuse from vendors due to its skyrocketing popularity. Metamarkets, for example, touts the benefits of its “data science platform”, but their core technology is a slice & dice aggregator. Similarly, many people who know SQL and MicroStrategy are now claiming to be data scientists. I fear this term may become a victim of its own success and suffer the same fate as big data.

NewSQL?

NewSQL is a moniker for describing highly-scalable, horizontally-distributed SQL systems. Drawntoscale, VoltDB, SpliceMachine, SQLFire, Impala, Redshift, Clustrix, NuoDB, and Hadapt are a few of the many solutions that combine the scalability of NoSQL platforms with SQL and the strong ACID guarantees of legacy relational databases. NewSQL doesn’t mean NoSQL will die, it just means that companies who want scalability and SQL can have their cake and eat it, too.

Many companies will continue choosing NoSQL systems because they support non-relational data and can offer higher performance because they don’t provide ACID guarantees.

Predictive Analytics?

After many years of relative obscurity, predictive analytics are coming into their own. Core to both data science and smart data, predictive analytics are the flip side to historical analytics, and involve using historical data to predict future events. If you can predict the future, you can also change it.

Indeed, predictive analytics are behind everything from recommendation engines (which recommend items that are predicted to maximize the chance of a conversion), to fraud detection, to yes, predicting which parolees are most likely to commit murder. The field calls upon techniques in statistics, machine learning, modeling, and other fields to identify and exploit patterns.

Trends that didn’t make the cut, but are worth paying attention to, include stream processing and streaming analytics, NLP (which seems well on its way to entering mainstream, thanks in no small part to technology vendors like AlchemyAPI), image and video mining (including face, gesture, and emotion detection), machine learning, in-memory storage and computing grids, and graph databases, which offer a completely different way of solving problems in data analysis.

Life after Big Data

Big data as a term has seen its heyday. While many of the challenges that gave rise to the term are valid, storing virtually infinite amounts of multi-structured data is no longer novel or even mildly interesting.

Moreover, widespread and proliferating abuse of the term by vendors means that it means less and less with each passing month.

Increasing sophistication in the storage, processing, and use of data means we’re probably not going to see a single term replace big data. Instead, we’ll see the most common use cases forge on ahead, adopting terminology more restrictive and more descriptive.

Welcome to the post-big data era! It’s going to be one hell of a ride.

John A. De Goes is CEO of Precog, a company focused on making it easy to develop and deploy advanced analytics. He founded Precog after struggling with big data challenges as VP Engineering at LivingSocial. 

De Goes is an established and best-selling technical author and a major open source contributor. He has been working in distributed systems design and development for more than a decade.

Frustrated journalist receiving another “big data” pitch // Joshua Minso, Shutterstock 

July 9-10, 2013
San Francisco, CA

Tickets On Sale Now

‘Big data’ is one of the most over-used buzzwords in the startup vernacular, and founders rarely have the goods to back it up. So you’ll understand that I was intrigued — but highly skeptical — when an email with the subject line “using data to cure cancer” popped into my inbox.

But Ayasdi, a startup that closed $10 million in venture funding today, doesn’t just talk the talk. Stanford researchers have been baking the complex algorithms behind Ayasdi (its quirky name means “to seek,” in Cherokee) for over a decade, with the goal of unlocking the hidden value in human genetic data.

In 2008, the founders, Gurjeet Singh, Dr. Gunnar Carlsson, and Harlan Sexton, decided to commercialize the technology. With the government stepping up its funding for science, they were able to pull in $3.5 million in grants from DARPA, the department of defense agency responsible for building new technology for the military, and the National Science Foundation. The result? A synthesis of machine learning technology and topological data analysis (TDA) that has impressed a score of Silicon Valley investors. Rather than typing in search-style queries, the tools allow for automated discovery of information.

As Dr. Carlsson explained in an interview, “The idea is to answer questions that you didn’t know to ask.”

Ayasdi CEO, Gurjeet Singh.

This year, the 30-person team of engineers will expand its marketing and sales efforts with funding from Khosla Ventures, Floodgate, Data Collective’s Matt Ocko, serial entrepreneur, Steve Blank, and more. Storied investor Vinod Khosla, who rocked the medical world with the statement that 80 percent of doctors would be replaced by machines, said Ayasdi’s “machine powered intelligence” has the potential to unearth “previously unattainable insights that will help solve some of our most pressing global, social, and economic issues.”

Eric Schadt, Director of the Institute for Genomics and Multiscale Biology, has a team of researchers using the technology to identify the genetic predispositions of many diseases, including cancer, which they hope will help them “glean new insights that will lead to breakthrough drug therapies.”

Related: In the burgeoning field of genomics, entrepreneurs aim to deliver more personalized treatments for life-threatening diseases.

Ayasdi is working with the nation’s top hospitals and medical researchers to uncover more targeted treatments for disease. Singh, the company’s CEO, told me that hospitals and big pharmas are routinely pulling data from public sources — medical researchers are required to publish their data  – and combine it with private data to yield new insights.

A data visualization depicting 14 variants of breast cancer

The data isn’t anything new — it’s the technology that has evolved. “We have automated the discovery of knowledge from data,” said Singh in a phone interview. “We were able to discover a new type of breast cancer without asking questions.”

Singh was referring to a recent breakthrough where Ayasdi mapped 14 variants of breast cancer. Using data collected during a 15 year period, and studied by thousands of scientists, the algorithms discovered a sub-group of patients that have a higher chance of survival based on their genetic profile. If a patient falls into this sub-group, it is unlikely that they will require chemotherapy.

Ocko witnessed this medical breakthrough first-hand in the minutes before he opened his check book. “The massive deficit that corporations and governments face as the amount of data gets exponentially larger is not more intelligence, it’s insights and intuition,” he explained by phone. The data-focused investor said the company’s ability to provide human analysts with “intuition from vast amounts of data in very short periods of time” is virtually unprecedented.

In another recent partnership, with Mount Sinai Medical Center, Ayasdi was used to point to targeted treatment options for E. coli sufferers. E. coli affects more than 265,000 people in the U.S. every year, and millions around the world. It is known in the medical community for developing resistance to many drugs, and doctors are never 100 percent sure if a treatment will work or not. Mount Sinai is using Ayasdi to analyze the entire E. coli genome sequence, which includes more than 1 million DNA variants. This will further our understanding of why some types of E. coli develop resistance to antibiotics and how we can combat the spread of the bacteria.

Singh, a former researcher at Stanford, told me that the company has secured 20 customers in the oil and gas, government, pharmaceutical, and healthcare sectors. Big name customers include Merck, the Food and Drug Administration, and the U.S. Department of Agriculture.

Imagine a world where every machine — from the smart meter to the cell phone — is designed to mimic the human brain by getting smarter over time.

A new “big-data” product called Grok promises to make those brainy machines a reality. It’s the first product from Numenta, which Palm cofounders Jeff Hawkins and Donna Dubinsky started in 2005. Numenta recently announced its technology, and now it’s brought new leadership on board prior to going from beta testing to general release. New chief executive Rami Branitzky (pictured above, right) is a former executive at enterprise software behemoth SAP.

But the real brains behind the technology is Hawkins (pictured above, left), a neuroscientist and technology entrepreneur, who I spoke to recently by phone. He painted a vision of an urban environment where “everything is intelligent, everything is learning.”

Hawkins has a knack for predicting the future. Two decades ago, he told a group of engineers at Intel that momentum would shift to the mobile space, and people would carry tiny computers in their pockets. “That talk was one of the worst-received talks I ever gave. I don’t think anyone believed me,” he revealed during a recent keynote at the International Symposium of Computer Science. During his subsequent 15-year-long “side-track” into mobile computing, Hawkins proved critics wrong when he invented the Palm Pilot.

Hawkins’ latest invention, Grok, stems from years of neuroscience research.

Grok uses algorithmic frameworks to find trends in scattered data and make predictions about what patterns come next. Instead of analyzing mountains of data that companies are storing, it is one of the first to work off streams of real-time information from sensors.

“There is an explosion of things that are wide and connected [and] an emphasis on storage and processing all that data,” Hawkins said. “People are storing a lot of data but not solving the problem of what to do with it.”

He refers to technology giants like EMC and NetApp that specialize in storing vast quantities of noisy data. The latest technology products to hit the market are designed to go one step further than this by making this data useful to businesses and government agencies. From SAP with its next-generation database to a venture-backed startup like Platfora, there are dozens of contenders in the increasingly crowded market.

But what’s unique about Grok is the automation layer, and its continuous ability to learn. The product can be fed into companies’ existing control systems to automate processes that previously required manual adjustment. Instead of reacting to problems, users can anticipate them and optimize accordingly.

Grok is currently being tested on a few customers in the ad-tech, video processing, and energy management worlds. One global energy management company is using Grok to predict energy consumption in a building every hour for 24 hours and adjust output accordingly. Likewise, publishers and mobile app developers are leveraging the product to boost ad revenues by ensuring that inventory is optimally placed across the various ad networks.

Hawkins told me the most exciting challenges are “green-field,” meaning there’s no legacy technology solution to displace. “Once we have the data source, within half an hour we will begin to generate results,” he said.

This isn’t too disimilar to the way the human neocortex, the part of the brain that includes the perception and reasoning functions, discovers sensory data. In his body of work, Hawkins argues that the brain is a big data engine– it’s a sophisticated predictive modeling system. The brain has to learn everything by processing a continuous data stream of events. Just like an intelligent computer, it develops a model to make predictions, detect anomalies, and take action.

Given Hawkins’ reputation in the tech community, customer acquisition shouldn’t be a problem. He became widely-known with the success of his book On Intelligence, which delves into the basics of neuroscience and draws parallels with next-generation computing. “When I wrote it seven years ago, I was approached by technology companies — potential customers — but I didn’t have anything to sell,” he said.

So Hawkins founded Numenta, the Redwood City-based predictive software company, and brought on former Palm Computing CEO Dubinsky (pictured above, center). With an expanded team and new product, the internally-funded company is also on the look-out for traditional sources of capital and strategic partners.

New CEO Branitzky will be steering the company to new sources of revenue. ”I have seen a lot of enterprise software,” he told me. “I know where the gaps are in the space.”

For its first customers, Grok has improved top-line performance for its first customers by at least 10 percent. With positive feedback from the beta, the team feels ready to start selling the product in the first half of 2013. They are planning to market to small- to medium-sized businesses that do not have the budget for a data scientist or data service team. According to Branitzky, these companies typically have some kind of “prediction problem,” which is where Grok comes in.

Echoing the rise of mobile technologies, venture capital firms and large companies poured millions of dollars into big data technologies this year. And yet, Hawkins feels he’s an early arrival to the party, and we haven’t seen anything yet in terms of what this technology can do. “We’re definitely not entering into a market that has been around for a while,” he said. “This is an unmet need and our customers have no solution.”

July 9-10, 2013
San Francisco, CA

Tickets On Sale Now

Prominent futurist and author Ray Kurzweil has accepted a position as director of engineering at Google, where he plans to work on technology developments in language processing, machine learning, and other areas.

Kurzweil is well-known for authoring several books about the future of technology and “the singularity,” a period when he says humans will merge with intelligent machines. He believes we have made discernible progress with artificial intelligence but have much further to go. For example, he’s publicly criticized Apple’s Siri as “primitive” and thinks it needs a lot of work. Perhaps that means he’ll work to improve Google’s competing Now service.

“I’m thrilled to be teaming up with Google to work on some of the hardest problems in computer science so we can turn the next decade’s ‘unrealistic’ visions into reality,” Kurzweil said in a statement.

Google has indeed made some of his predictions a reality. The most notable example would likely be Google’s self-driving cars, which, to date, have logged many thousands of miles without an accident.

“In 1999, I said that in about a decade we would see technologies such as self-driving cars and mobile phones that could answer your questions, and people criticized these predictions as unrealistic,” Kurzweil said. “Fast forward a decade — Google has demonstrated self-driving cars, and people are indeed asking questions of their Android phones. It’s easy to shrug our collective shoulders as if these technologies have always been around, but we’re really on a remarkable trajectory of quickening innovation, and Google is at the forefront of much of this development.”

While he has taken on this role at Google, Kurzweil said he will also continue acting as a tech thought leader by hosting academic lectures and talking with press.

For more on Kurzweil, check out an interview we had with him back in October:

Ray Kurzweil at DEMO via Stephen Brashear

Is your startup struggling to solve a tricky data challenge? Data science company Kaggle is launching a program today to rally its community of 70,000 statisticians from around the world to solve your most enduring problems.

The only catch? The new program known as “let the crowd be your cofounder” is only open to small startups, and founders will need to set aside $5,000 in prize money or 0.5 percent in company equity for the winning submission. Still, it’s a darn sight cheaper than hiring a full data services team.

Kaggle is a San Francisco-based startup that hosts data science competitions, typically for large companies and nonprofits like the Hewlett Foundation, AllState, and GE. For instance, The Hewlett Foundation offered up $100,000 in prize money to the first person that could create an algorithm for scoring essays the same way a human grader would.

Related: Read more about cool Kaggle competitions here.

Kaggle’s CEO Anthony Goldbloom told me he realized that fast-growing tech startups are storing vast volumes of consumer data but often don’t know what to do with it.

“I got really excited about the idea of data-driven startup just as I was starting Kaggle,” said Goldbloom in an interview. Before launching the program to the public, Goldbloom worked closely with three startups to solve their data challenges:

July 9-10, 2013
San Francisco, CA

Tickets On Sale Now

Startups often say their technology will change the world — a claim that seems wildly optimistic at best.

But step into the San Francisco offices of Change.org, the online petitioning site that is arming ordinary people with the tools to start revolutions, and you’ll be swept up by the mission. You won’t be alone. The movement is about 25 million people strong and is growing at a rate of 2 million memberships per month.

Can an individual spark a revolution?

Governments and corporations, they say, will be forced to listen if hit by thousands of signatures on Change.org, an army of Twitter and Facebook fans, and a few well-placed newspaper articles.

I meet Ben Rattray, the company’s charismatic CEO (pictured above), in his corner office. He’s weary from frequent travel: we joke about the “networking opportunities” during the red-eye flight from New York to San Francisco on Sundays. It’s typically jam-packed with tech execs returning from East Coast trips.

Directly behind his chair, a single photograph is framed in an otherwise sparsely decorated room. It’s the face of Molly Katchpole, the 22-year-old nanny who took on Bank of America. In November 2011, Katchpole’s victory was affirmed when 300,000 people signed her petition and the bank dropped its new $5 monthly debit card fee.

This is one of many recent campaigns that caught the attention of the global media. On Change.org, a trio of high school girls secured CNN’s Cindy Crowley as the first female presidential debate moderator in 20 years. One of the site’s members rallied hundreds of thousands of people to put pressure on the South African government to crack down on “corrective rape,” a term used to describe the horrific practice of a man raping a lesbian with the aim of ‘turning’ her heterosexual.

Something has shifted in recent years. Political activism is no longer the realm of the youthful and naive. Obama’s election and re-election were fueled by social media, and protests against SOPA and PIPA were staged on Twitter. Social media has created a culture of transparency that is making a site like Change.org truly effective.

“The strong thesis we hold is that people aren’t born apathetic, they are trained to be,” said Rattray.

“If you show them that their voice matters, you can radically change behavior.”

The company is hiring at a rate of two people per week and forecasts revenues of $15 million this year.

The blurry line between profit-making and change-making

And this is where Change.org has faced some criticism. Some say the company has crossed the line between change-making and profit-making; Clay Johnson, author of “The Information Diet” told the Wall Street Journal he has “huge problems” with the site as it is a “lead-generation business disguised as a social-change organization for whoever is willing to pay them for the email addresses.” He continued, “It’s dangerous to monetize ‘change’ because there’s an economic incentive to sensationalize.”

Certified as a B Corporation, the company makes its money through their cause-based ad model. For instance, the World Wildlife Fund might place a sponsored ad on a campaign that supports the environment, or might pay for Change.org to promote an existing campaign. It’s not dissimilar to sponsored tweets or sponsored links on Google.

Rattray claims the company is less concerned with the profit motive than with maximizing impact. To deflect critics, he uses the example of an ongoing engineering project to build a mobile offering that can be deployed in developing countries. This is not intended to make money.

He says Change.org, as a revenue-generating and angel-funded business has been able to offer its employees a competitive salary and benefits. Meanwhile, nonprofits are struggling to hire and maintain talent.

Why work at Change.org?

This week, the company brought on a high-profile CTO, Tom Hughes-Croucher, to lead the engineering efforts. The new hire has a strong reputation in the open-source community, and has worked as a technical evangelist at Joyent and Yahoo.

On the engineering roadmap is a plan to incorporate machine learning so members will be matched instantly to a petition based on their pre-existing interests.

“I’m really excited to be starting at Change.org because I know we have such an amazing opportunity to have an impact on the world here as well as some really interesting engineering challenges that come at the scale we operate,” Hughes-Croucher said.

Still, the company has not been able to lure quality engineers that are interested in a sky-high salary and large chunk of equity. “Eighty percent really care about the equity upside and a massive exit,” Rattray told me. “For the 20 percent that care about the mission, we are the fastest growing company in this space.”

The company has openings in communications and marketing, business development, customer support, and engineering. Check out the full list here.

“Startup culture” is a new series that highlights what it’s really like to work at a Bay Area startup. We’ll be profiling a startup every two weeks. Please send your suggestions for the most rockin’ office spaces, startup happy hours, or company perks via email (christina@venturebeat.com) or Twitter (@chrissyfarr). 

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

July 9-10, 2013
San Francisco, CA

Tickets On Sale Now

The day before Thanksgiving, Prior Knowledge was gobbled up (gobble gobble) by Salesforce.

Prior Knowledge is the company behind Veritable, a predictive database that provides insight into the process of building applications. Referred to as “infer-structure,” the API examines the underlying infrastructure of an application to find causal relationships. It then tells developers what they might be missing. Using this system, developers without deep knowledge of statistics or analytics can make data-driven predictions.

Since the product entered public beta in July and made it to the finals at TechCrunch Disrupt, there has been a lot of excitement surrounding the technology and its MIT PhD holding founders.  Salesforce has been on an aggressive buying spree this year of mobile, social and cloud startups, and while P(k) does not fit into any of those categories, it could help Salesforce take on enterprise behemoth Oracle. Now, Salesforce has an “oracle” of its own, a database that can predict the future.

With this acquisition, Prior Knowledge will shut down the Veritable API and service on December 5. The details of the purchase have not been disclosed. Read the company’s blog post.  

This is a guest post by Laura Teller

In the typical office, machines will increasingly do more work, automating functions that were once performed manually.

Machines that are capable of learning seem smart to us today.  Consider how Internet radio service Pandora is able to better-understand your music taste over time.

There are copious reports that machine learning on “big data” will replace human domain expertise.

However, even the smartest machines still need teachers, and those teachers are human experts. We shouldn’t fear that they will replace people in the workplace.

["Big data" and machine learning is a focus at VentureBeat's upcoming CloudBeat conference.]

These machine learning solutions that include closed-loop feedback require someone to correct their mistakes, so the machines can learn from them. But there’s a bigger reason humans will never be replaced by machines, and it has to do with the three levels of cognition, developed by Prof. Terrence Deacon, Ph.D., Chair of the Department of Anthropology at University of California, Berkeley.

The top level is iconic

At this level, a computer can identify something for what it is. The perfect example of this is music identification apps for smart phones, such as MusicID, Shazam, or SoundHound, which can identify recorded music by listening through a mobile phone’s microphone.

The second level is indexic

This refers to the mind’s ability to make associations. For example, pointing a finger to a given person means that you’re talking about that person. Other examples include how we group certain things, such as peanut butter and jelly, hats and scarves, or time and money.

Machines are very good at these two levels because they are driven by clearly defined patterns and boundaries (in the case of the iconic) and simple rules (in the case of indexic). In other words, they can be programmed. But the third level is not so clearly defined, and thus it is where human virtuosity becomes essential.

The third level is symbolic

The human mind uses abstractions, which allow us to complete a story or picture that has missing parts. If the amount of information present is only 2 percent of the total data available, we will complete the picture with what we’re given as if it’s 90 percent. Some simple examples for this would be how our brains are able to complete words in Wheel of Fortune or how the smell of pine needles represents Christmas. Like the missing letters in that iconic TV trivia show, the Christmas tree is the missing link between the pine smell and the holiday.

Machine learning’s Achilles heel

Symbolic thinking is also what allows us to look forward in terms of innovations. We take the information available and the goal we’re trying to achieve and fill in the blanks to bring ideas to fruition.

Machines are not capable of symbolic thinking because it is outside the realm of data and programmable logic. Instead, the computer combines iconic and indexic cognition at increasingly greater scale. But each time it does this, the human mind expands ever further, pushing the limits of the machine even more.

Symbolic thinking allows humans to wonder, create, dream, question, and so forth. When we get a new piece of information and respond, “that raises more questions than answers,” that’s an example of symbolic thinking. Receiving information that is only a small piece of a large puzzle causes our brains to try to fill in the pieces using a number of faculties, including logic, emotion, experience, foresight, and other human-specific attributes depending on the situation.

Machines’ power to fill in such blanks ends at programmable logic and basic pattern recognition.

In the business world, humans will continue to ask the big questions, look at the big picture, innovate, dream, wonder, and so on. They will also establish emotional connections and relationships with clients.

On a more mundane level, they’ll be the ones to provide the feedback that makes learning possible for machines. But most important, humans will always be the ones making the decisions. No matter how smart the machine is, the buck can’t stop at a computer.

Laura Teller is the CSO at Opera Solutions. She has more than 22 years of management consulting experience in marketing strategy and market opportunity assessment and has played a vital role in putting Opera Solutions at the forefront of the Big Data revolution. Since joining Opera Solutions in 2009, Laura has become a leading expert on Big Data and the dramatic positive impact it can have on business.

The author of “Small Business, Big Savings”, Laura holds a BA from Yale and an MBA from Harvard University, where she was a Baker Scholar.

Top image via Shutterstock

July 9-10, 2013
San Francisco, CA

Tickets On Sale Now

At 12 years old, Nick D’Aloisio began his career as a mobile programmer when he downloaded the Apple Developer Kit. By 15, one of his hacks, an app known as “Trimit,” caught the eye of an investor in Asia offering to fund “the company’s” next project.

What company? D’Aloisio, a school boy from South London, admitted to this venture capitalist that it was merely an after-school creation.

A whizz at the programming language C, D’Aloisio had created the app to test his skills in machine-learning technology, a branch of artificial intelligence that yields trends and patterns in a mass of data, and makes smarter predictions over time.

Summly’s founder demoes the app in San Francisco

Less than a year later, the “Internet boy wonder,” as he was dubbed by the European tech media after an appearance at Le Web’s conference, had developed a keen interest in web summarization technology. It’s a simple enough problem, but the execution is tricky. How do we take a meaty piece of content and whittle it down to its bare bones?

Today, D’Aloisio is ready to take his iPhone app public. Summly, available to download for free, reduces full-blown articles into snippets, making it easier to skim the news on a mobile device.

From ESPN to VentureBeat, the app includes almost all of the major news outlets. You lightly tap the screen to browse news, and if a summarized story piques your interest, you can share it with your friends via SMS, email, Twitter, or Facebook, or click a link the bottom of the screen to access the original, full-length story.

The founder told me he hit on the idea for the novel way to scan the news while studying for his History “GCSE” (an exam that all British school kids sit before their junior year). He quickly became frustrated by thousands of web search results that were slow to load and sometimes impossible to access on mobile.

“There has to be a way to summarize this information into bite-size,” he said. Trimit, the app he developed, summarized content into 140, 500 or 1,000 character summaries. With a little finessing, he moved on to Summly, which can automatically break down full pages of text into bullet-points. 

Most summarization experts have yet to apply their knowledge to a mainstream market. In a highly strategic move, D’Aloisio is applying the technology to the news, since, as he said, “No one had figured out how to perfect the news-reading experience on mobile.” In many cases, it’s relatively simple: Most journalists still write in the Associated Press (AP) style, so the first paragraph contains the crux of the news.

It’s not nearly as straight-forward as cropping the first few sentences, however. In the background, the technology works by scraping content, pulling out the most important bits, categorizing each story by topic, and displaying the news in a highly visual way. The summarized stories fit comfortably on your screen (tailored to both the iPhone 4 and 5).

At a coffee shop in San Francisco, D’Aloisio led me through a demo of the app: It’s a highly intuitive with the algorithm working to recommend news stories to you based on your pre-existing set of interests. As you shift location (from the UK to the U.S., for instance), the stories will change as well.

In person, D’Aloisio is wise beyond his years. It’s hard to believe this media-savvy, sharp-talking business executive is a teenager taking a leave of absence from school. He is both a high-school student and an employer. No longer a solo coder, he now has a team of seven people working for him. In addition, half a dozen researchers at SRI, the non-profit technology research institute based in Silicon Valley, are working with D’Aloisio to perfect the algorithm.

Some major technical hurdles  remain: Magazine-style features are tougher to summarize, and are often thrown out by the algorithm. The team is currently working on auto-translation (notoriously problematic and obvious to anyone who has ever used Google Translate, BabelFish and the like), and they have plans to integrate French and Spanish into the app. If they get it right, it could be a neat language-learning tool.

The team will also butt heads with Pulse, Flipboard, and a host of other beautifully-designed news-reading apps, especially since it is in the process of designing an interface for tablet devices. To succeed, it will need to win over voracious news consumers who routinely use web-based aggregators like Google Reader. The competition for these users is heating up: Circa is experimenting with ways to re-frame the news on mobile, and NewsBlur, a Y Combinator alum, recently launched its iPad app.

Summly is backed by several investors, including Horizons Ventures, Ashton Kutcher, Betaworks, Brian Chesky, Hosain Rahman, Joanna Shields, Josh Kushner, Mark Pincus, Matt Mullenweg, Stephen Fry, Troy Carter, Yoko Ono and many more.

Top image: Wired UK

AutoGrid, a Bay Area-based startup founded in 2011, pulled in $9 million to further the development of its software to help utilities and businesses control their energy usage.

Its technology works by analyzing data collected by smart meters to allow operators to adjust to meet supply and demand. Its major customers include the City of Palo Alto Utilities and Sacramento Municipal Utility District, and it is reaching out to municipalities on the East Coast.

The data is collected from smart meters through sensors that are placed on the electric grid. By incorporating machine learning, which gets smarter over time, it can make predictions about energy consumption patterns inside buildings and across service regions. When enough data has been processed, it can forecast how much electricity might be needed in the coming hours or days.

“The capacity of the [smart] grid is not deployed efficiently,” said founder and chief executive, Dr. Amit Narayan (pictured, left). Smartgrids are the computer-based remote control and automation mechanisms that are used today to deliver electricity. In the past, workers had to manually gather data by reading meters and measuring voltage, and keeping an eye out for flawed equipment.

With the advent of analytics tools, we can begin to make predictions about the future and further optimize energy usage. “Up to 20 percent of the power-generating assets in some regions only get deployed ten or fewer days a year,” said Dr. Narayan.

Prior to founding AutoGrid, he taught on the topic of electronic design automation software at Berkeley and Stanford. He founded Berkeley Design Automation, which is used by more than 20 of the top 25 chip developers.

“AutoGrid is creating the brains for the smart grid. If you can analyze all of the data, you can predict what the electrical parameters of the grid will be under any situation and use that to remove inefficiencies from the electricity supply chain.” said Dan Ahn, managing director at Voyager Capital in a statement.

What’s interesting from a competitive standpoint is the company’s focus on pattern-recognition and predictive analysis. It faces competition from Enernoc, a real-time energy-management tool, and Silver Spring Networks, an energy networking supplier that is used by over 15 utilities across the U.S.

The company’s investors include Foundation Capital,  Voyager Capital and Stanford University. Last year, it won a $5 million grant from the U.S. Department of Energy program to devote to improving the product.

Image via Shutterstock

July 9-10, 2013
San Francisco, CA

Tickets On Sale Now

Skytree, a “big data” company that came out of stealth mode in February, has filed for additional funding from existing investors, Javelin Venture Partners.

The startup has a tool to bring machine learning techniques to a mainstream market and has raised $1 million of a $2 million round, according to the security and exchange commission (SEC).

Machine learning lets systems get smarter over time, which helps data scientists and developers uncover patterns in a chaotic mess of data. Skytree’s eponymous product is known as “Skytree Server.”

The company has hired a team of machine-learning Ph.D.s to design the server and integrate with existing data stores, including Hadoop. The hope is that it will process and query data 10 times faster than is currently possible.

The startup is interesting to investors as it has taken on a meaty technical challenge. Machine learning is a particularly complex approach to big data that has not been fully leveraged. Only a handful of tech companies, government agencies, and financial services firms have integrated machine learning.

The San Jose-based company raised its series A round from Javelin Venture Partners in February, 2012. Read the SEC filing.

Image via Shutterstock

If you’re feeling achy and unwell, the last thing you need is the headache of tracking down a nearby doctor that can see you in short order.

A startup called BetterDoctor has already won over tens of thousands of people with the simple elevator pitch: “We’ll help you find the right doctor.”

The company is the brainchild of Ari Tulli, formerly the head of Nokia’s App Studios, who teamed up with Nokia’s Chief Architect Tapio Tolvanen to fix some of the gaping holes in our healthcare system. ”On average it takes a patient 20 days to get to see a doctor,” Tulla told me.

On BetterDoctor, perform a simple web search, and you’ll be able to access profiles of doctors with offices in your neighborhood. They are listed by speciality, such as primary care or obstetrics. If you’re regular physician can’t see you immediately, the site will point you to alternatives and provide their contact info.

What’s interesting about this startup is that it’s a tool not only to find an available doctor, but to find the best doctor.

The company has spent 18 months building a verification service and has deliberately excluded doctors with negative ratings and those who are fighting malpractice lawsuits. Physicians listed on the site have been included based on their experience, education, and more.

In the future, Tulla told me, they’ll incorporate machine learning (think ‘Pandora for healthcare’), so they’ll be able to recommend a doctor based on your gender, location, insurance type, prior physician, and pre-existing conditions. Also in the product roadmap is a scheduling service for users to book an appointment online. This will put the company in a stronger position to compete with up-and-coming healthcare startup, ZocDoc.

The product has been beta-tested in San Francisco for several months, and after several revisions, it is launching nationwide today. It has already signed up hundreds of doctors in the Bay Area, and is adding more each day.

On the site, you’ll find 600,000 doctors’ profiles. The majority of these doctors will not have registered with the service or even heard of BetterDoctor. The startup’s team of programmers have scraped Yelp, hospital listings, private practice sites, and the like to create custom profiles for doctors with reviews, location information, and more.

Tulla told me it’s a great marketing tool for doctors, who don’t have to lift a finger to get free publicity. In the Bay Area, many will have noticed a recent, mysterious spike in patient bookings before doing their homework and finding BetterDoctor. “There are some doctors who have 10,000 views on their BetterDoctor profile,” Tulla told me.

For now, doctors can join the site on an invitation-only basis, but they can request to “claim their profile,” meaning that they can fill it out with custom, additional information. The founders plan to make money by asking these doctors for a cut when a BetterDoctor user — a new patient — walks into their practice.

BetterDoctor’s web app is already live, and the iPhone app is currently under production and will be released in the coming months. The company launched its private beta at Founders Den, a shared office and private club for San Francisco-based entrepreneurs.

Top image via Shutterstock

You can imagine that, at some point, a young lawyer would start to fantasize about using a computer to summarize those contracts instead of burning the midnight oil.

That’s exactly what the founders of eBrevia did. The company, which launched today at the DEMO Fall 2012 conference here in the heart of Silicon Valley, uses machine learning to automatically summarize contracts and other legal documents.

“We’ve spent many nights as corporate attorneys scanning computer screens at 3 a.m. to extract and summarize legal provisions as part of due diligence in mergers and acquisitions,” founder and chief executive Ned Gannon told VentureBeat. “We have literally felt the ‘pain’ of the process ourselves.” Now, Gannon and his cofounders hope, they can help law firms and legal departments cut costs and save time by semi-automating the contract examination process.

People who use eBrevia’s web-based eDiligence Accelerator service specify what kinds of provisions they’re looking for. They upload legal documents and, in return, receive a summary chart that shows all the uploaded documents along with a more detailed template for each document, with special attention to the provisions that the user is looking for. Readers can easily move back and forth between the summary, the document templates, and the original documents, Gannon said.

The software uses machine learning to improve its summaries and charts as time goes on.

“It is exciting that the software gets smarter and smarter with the more examples it sees,” Gannon said.

Not everyone is so enthusiastic, however. Onstage at DEMO, American Express Ventures managing partner Harshul Sanghi said, “There’s a reason these documents are as long as they are … The devil is in the details. I’m not sure you can shorten that.”

True Ventures founder Tony COnrad concurred, saying, “I love their technology, but I don’t think legal is the place to use it.”

The company hasn’t settled on exact pricing yet but plans to give its customers — primarily law firms and corporate legal departments — a choice between per-document and monthly subscription fees.

After the launch of its eDiligence Accelerator product, the company plans to launch products for contract management, document drafting, and consumer applications. Whether it succeeds in doing that will depend on how well its summaries work. It’s hard to imagine natural language analysis that’s capable of outstripping the contextual knowledge and legal savvy of even the most exhausted lawyer. But perhaps with enough raw data and a well-defined subject area, eBrevia will help save lawyers time — even if it doesn’t eliminate the need for them.

Gannon and cofounder Adam Nguyen started eBrevia in July, 2011. The company is based in Stamford, Conn. Its presentation at DEMO is the prize for winning a national contest sponsored by Startup America and DEMO. It is funded by its founders, with the addition of the $25,000 Startup America/DEMO prize.

“Connecticut has a strong and growing entrepreneurial ecosystem, and we’re thrilled to represent the state in the Startup America/DEMO Competition,” Gannon said.

The founders are now seeking $400,000 in seed funding.

eBrevia is one of more than 75 companies chosen by VentureBeat to launch at the DEMO Fall 2012 event taking place this week in Silicon Valley. After we make our selections, the chosen companies pay a fee to present. Our coverage of them remains objective.

Photo credit: eBrevia. From Left to right: Ned Gannon, Jake Mundt, Adam Nguyen.

The startup AfterCollege is relaunching with a new matching algorithm to help seniors at 2,300 colleges and universities find the perfect entry-level job or internship. The company claims its technology was developed for graduating students to stand out in this fiercely competitive job market.

The new website is far more technical than the original, which was a simple online job board for college students that competed with sites like Indeed and CollegeBuilder. With its relaunch, the site is rebranding itself as a “LinkedIn for college students,” and users are encouraged to build out their professional profiles.

AfterCollege has a tough road ahead if it aims to take on LinkedIn and career networking startups like BranchOut that have spent years developing a presence at college campuses and marketing to students.

Most interesting is the built-in job search tools (a resume and cover-letter builder) and the patented-machine learning technology that matches seniors and recent grads with opportunities, based on their skill-sets, personal and academic interests.

The algorithm works by inputting a user’s academic affiliations (school, educational program, department, groups) and searching the database for relevant opportunities. AfterCollege has been around since 1999 and has been used by 3 million students, so it has access to a wealth of insider information.

For instance, the algorithm would know if the computer science department at San Francisco State has a different focus than the program at Stanford University. At Stanford, Human Computer Interaction (HCI) might be a strong interest for many students, while at SF State, students are skilled at Java or social applications — useful information for engineering recruiters and tech companies.

“The algorithm is also smart enough to distinguish between different programs that may be perceived to be similar,” explained AfterCollege CEO and founder Roberto Angulo. He started the company as a junior at Stanford. In 13 years, AfterCollege has connected students with some half a million jobs and internships.

AfterCollege also optimizes chances for student applicants as they can view jobs that graduates with similar skill-sets have succeeded in landing.

AfterCollege’s algorithm is a useful tool for employers, who fear they’ll miss out on the best candidates. No doubt, this will be a major money-making opportunity for AfterCollege; the site will likely begin selling targeted advertising to employers.

The startup raised its most recent round of funding in 2011, a sub-million-dollar sum from Flywheel Ventures.

Top Image via Shutterstock

Google X Lab, the search-engine giant’s supersecret laboratory for Reed Richards-style experiments, made an interesting discovery in its efforts to create a machine capable of learning on its own. The team created one of the largest neural networks for machine learning — aka a brain –  by connecting 16,000 computer processors with over a billion neural connections. It then exposed it to 10 million randomly selected YouTube video thumbnails over the course of three days. After feeding it a random sampling of 20,000 images, the brain began recognizing pictures of cats using a deep-learning algorithm that didn’t include any prior suggestions or information regarding felines and what they look like.

“We never told it during the training, ‘This is a cat,’” Google Fellow Jeff Dean told the New York Times. “It basically invented the concept of a cat.”

The overly simplistic explanation here? Cats actually do power the Internet, and this is now a serious theory rather than a humorous observation.

The slightly more credible explanation for the results is due to the overwhelming number of cat videos uploaded to YouTube, thus making it easier for an artificial brain to recognize cats. The brain was able to correctly identify cat images with 74.8 percent accuracy. It fared better with identifying human faces (81.7 percent) and human body parts ( 76.7 percent). The brain’s performance is roughly twice as accurate than any previous machine learning test to date.

Google presents its findings at a conference in Edinburgh, Scotland this week. The company said its machine-learning findings could help advance speech and image recognition software, as well as translation services. For now, the research project is moving from the Google X laboratory to other areas of the company.