lunes, 23 de diciembre de 2013

Wise.io Debuts Machine Learning Service That Offers To The Public What Google Builds For Itself

Google, LinkedIn and Amazon have thousands of engineers who point their work inward to build better recommendations, search and other Internet-scale features. Wise.io is launching today to offer a similar form of machine learning that does the inverse by pointing its technology outward for people to use.

That's not to say that Wise.io will necessarily compete against these companies. It's just to point out the company's machine learning as a service is something that can be used by anyone to solve problems that now take hundreds or thousands of people to do.

It's the kind of company a scientist studying the great beyond would start. Someone like Joshua Bloom, the founder of Wise.io and a former professor of astrophysics from the University of California at Berkeley who launched his company today at the Alchemist Accelerator Demo Day. The company will now join the Citrix Startup Accelerator program and receive seed funding as well an undisclosed investment from the Alchemist group.

Wise.io works with machine-learning algorithms to make them fast and scalable. To do that, Bloom and his team used their years of studying quasars, black holes and the depths of space and built that knowledge into its technology. He said that in his research at Berkeley he and others would often manage data that was well beyond what conventional tools could offer. They had to invent the technology themselves. Now all that knowledge is getting packaged to offer as a service. It is still pretty complex. The site says Wise.io is built for data scientists and ready for the enterprise. The service features a data scientist marketplace and automated reports. The company has an enterprise offering that allows companies to host Wise.io on a virtual machine.

Wise.io acts as a framework for ingesting data from Hadoop, MongoDB and various file sources. The engine creates multi-dimensional views of the data it ingests. For example, machine learning can analyze every pixel in a picture and correlate its relationship to all the other pixels in the photo. The Wise.io engne can, as well,  process the billions of other signals that any data point relates to. Scale that multi-dimensional view from billions of signals and the benefits can come in any kind of form. A streaming provider might know if the customer using the smartphone is sitting or standing. The Wise.io framework serves as a central brain that takes a holistic look at the data. It has its most useful applications in tasks that require a high level of cognition and intelligence to get the work done.

He said the problem today stems from the man hours it takes to benefit from machine learning. Bloom said Netflix considered a machine-learning option a few years ago, but it would take 100 years man years to implement. By combining their own knowledge, Wise.io has developed a service that takes a holistic view of the data, doing with machines what he says people can do only with very large teams, Bloom said.

The company is targeting markets such an industrial safety, which has historically required analyzing data at scale. It can often take six months to produce a report for a customer about all its safety hazards that it migt have in a customer's factory.  Think of oil refineries or nuclear power plants and all the data that can be analyzed. Historically people have done the analysis. Machine learning can take that data and anaylze it in a fraction of the time. "It takes hundreds of man hours," he said. "We can do it in 20 minutes."

That kind of time savings means fewer people are required to do the job, which makes Wise.io both fascinating and reflective on who we are and the age that is rapidly emerging before us. As Bloom said to me, the Industrial Age saw machines doing the physical lifting. Today data does the intellectual lifting. And that's something to chew on.

Wise.io might very well compete with Google, which is now scaling its Google Cloud platform after making it generally available at Google I/O earlier this month. But the real competitor is the market noise. Data analytics providers, in-memory database companies and the list of competitors goes on and on. But money does talk. Bloom said he expects Wise.io to have $1.3 million in revenues by the end of the year.


No hay comentarios:

Publicar un comentario