Showing posts with label Machine learning. Show all posts
Showing posts with label Machine learning. Show all posts

Thursday, September 09, 2010

Google's Prediction API

Technology Review: Google Offers Cloud-Based Learning Engine: the smartest Web services around rely on machine learning--algorithms that enable software to learn how to respond with a degree of intelligence to new information or events...... Google-hosted algorithms could be trained to sort e-mails into categories for "complaints" and "praise" using a dataset that provides many examples of both kinds .... extracting emergency information from Twitter ...... machine-learning black box--data goes in one end, and predictions come out the other ...... "This API could be a way to get a capability cheaply that would cost a huge amount through a traditional route." ... Prediction API .... has the potential to be a leveler between established companies and smaller startups

We went from big, ugly computers - mainframes - to PCs. PCs were simpler. And over time they became pretty powerful. And then the cloud emerged. The internet itself was the cloud. So I agree with Larry Ellison when he claims he has always done the cloud thing.

We went from servers to data farms. And these data farms run by big companies like Google and Facebook are huge, big enough that the electricity costs are a major headache even for these rich companies.

When Larry bought Sun, I threw a challenge in his direction. Can you build data centers that are the size of servers? Or at least small data centers? That might be nano territory. But I figured what the heck? There is never too much drama in Larry's life. What is one more challenge?

One common denominator with these disruptive technologies is they have been democratizing forces. It has always been about making it possible for more and more people, more and more businesses. We basically want everybody to be able to go online.

Google's Prediction API is a step in that same direction, and I am glad. Suddenly even small businesses will be able to make sense of large quantities of data they might end up collecting.

Enhanced by Zemanta

Tuesday, September 07, 2010

Got Wings?

Sunrise - A Song of Two Humans (1927)Image by twm1340 via Flickr
Technology Review: Web Service Goes Date A-Mining: Unlike sites that rely on questionnaires, Wings tries to understand who you are by picking up the social media bread crumbs you leave online. ..... Wings doesn't ask you about yourself. It tells you. The service requires a Facebook account ..... All that data is fed into the service's recommendation engine. ..... Sunil Nagaraj, chief executive and cofounder of Triangulate, the company behind Wings. ..... The company raised $750,000 in July .... the density of one's social network turns out to be an important factor ..... couples tend to be well suited if they have similar percentages of friends from their own country versus other countries. It matters as well whether your Netflix rental or music playlist history tends toward the mainstream or underground. And couples that have lots of overlap in the types of people they follow on Twitter tend to match well ...... collecting and analyzing social data the way Wings does could be a new branch in the evolution of Web services that make smarter recommendations without having to be told something twice --or even once.

This comes across to me as the next wavelet for the social web's evolution rather than finally that it service that will find us all our soulmates that other dating sites or we ourselves can't. Relationships are mysterious things. Statistical analysis might show most white people go for white people, black for black, brown for brown, but recommending the same might reenforce a pattern that perhaps has been steadily breaking.

But it's good to see yet another startup taking a crack at online dating. Online dating sure is a growing phenomenon. A lot of people also in crowded cities like New York seem to find online dating preferable to meeting someone random at some bar.

It is not one or the other. What works for you works for you. It could be online. It could be offline. It could be a friend of a friend. It could be a perfect stranger.

Relationship building though you get to do on your own. There is not yet a site for that.

Technology Review

What's Next for the Netflix Algorithms?: more than 100 million ratings covering almost 18,000 titles from nearly half a million subscribers ..... combining lots of algorithms with machine-learning techniques might be a good approach to handling large datasets in general .... such algorithms could be applied in market trading, fraud detection, spam-fighting, and computer security

Can You Trust Crowd Wisdom?those ratings can easily be swayed by a small group of highly active users. ..... studied voting patterns on Amazon, the Internet Movie Database (IMDb), and the book review site BookCrossings .... In each case, they found that a small number of users accounted for a large number of ratings.

Digging a Smarter Crowd Instead of using the characteristics of articles to run its recommendation engine's algorithms, Digg's system is based entirely on calculating connections between users

Getting Computers Into the Groove Computers have revolutionized the production, distribution and consumption of music, but when it comes to recommending a good tune, they're still sorely lacking..... more automated methods of music search and recommendation could become important as on-demand music becomes more popular, and sites feel increased pressure to help users find new music...... while analyzing music using computers is "a very interesting and promising area of research," it will be hard to create a music search engine that's both general and fully automatic. "Music similarity is such a personal and variable thing," Crawford says. "Two heavy-metal tracks may seem highly similar to a classical-music expert like me, but entirely different to a heavy-metal enthusiast, who may in turn regard the music of Brahms and Tchaikovsky as very similar, which would be laughable to me."

Recommendation Nation The truth is that I now get more good recommendations about more things, more often, from Bayesian algorithms than from my best friends..... Better tech­nology doesn't mean worse friends.

Enhanced by Zemanta