Big Data  

Computer science researchers have come up with a new principle they call 'data smashing' for estimating the similarities between streams of arbitrary data without human intervention, and without access to the data sources.

Simply feeding raw data into a data analysis algorithm is unlikely to produce meaningful results, say the authors of a new Cornell University study funded by DARPA.

Related articles IBM Looks To “Post-Silicon” Era To Continue Moore's Law   George John on the Age of Artificial Intelligence  Biological Underpinnings of Modular Networks May Help Create Advanced Artificial Intelligence

From recognizing speech to identifying unusual stars, new discoveries often begin with comparison of data streams to find connections and spot outliers. But most data comparison algorithms today have one major weakness – somewhere, they rely on a human expert to specify what aspects of the data are relevant for comparison, and what aspects aren’t.

Data smashing applications. Pairwise distance matrices, identified clusters and 3D projections of Euclidean embeddings for epileptic pathology identification (shown in (A)), identification of heart murmur (shown in (B)), and classification of variable stars from photometry (shown in (C)). In these applications, the relevant clusters are found unsupervised.
Image Source - Chattopadhyay and Lipson

In the era of Big Data however, experts aren’t keeping pace with the growing amounts and complexities of data.

Now, Cornell computing researchers have come up with a new principle they call “data smashing” for estimating the similarities between streams of arbitrary data without human intervention, and without access to the data sources. Hod Lipson, associate professor of mechanical engineering and computing and information science, and Ishanu Chattopadhyay, a former postdoctoral associate with Lipson and now at the University of Chicago, have described their method in Royal Society Interface.

"Just as smashing atoms can reveal their composition, "colliding" quantitative data streams can reveal their hidden structure."

Data smashing is based on a new way to compare data streams. The process involves two steps. First, the data streams are algorithmically “smashed” to “annihilate” the information in each other. Then, the process measures what information remained after the collision. The more information remained, the less likely the streams originated in the same source.

"Just as smashing atoms can reveal their composition, "colliding" quantitative data streams can reveal their hidden structure," writes Chattopadhyay.

Any time a data mining algorithm searches beyond simple correlations, a human expert must help define a notion of similarity - by specifying important distinguishing ``features'' of the data to compare, or by training learning algorithms using copious amounts of examples. The data smashing principle removes the reliance on expert-defined features or examples, and in many cases, does so faster and with better accuracy than traditional methods according to Chattopadhyay.

Data smashing principles may open the door to understanding increasingly complex observations, especially when experts do not know what to look for, according to the researchers.

The authors demonstrated the application of their principle to data from real-world problems, including the disambiguation of electroencephalograph patterns from epileptic seizure patients; detection of anomalous cardiac activity from heart recordings; and classification of astronomical objects from raw photometry.

In all cases and without access to original domain knowledge, the researchers demonstrated performance on par with the accuracy of specialized algorithms and heuristics devised by experts.

Chattopadhyay has a demonstration of the data smasher available to try at:  http://home.uchicago.edu/~/ishanu/smash.html.


SOURCE  Cornell University

By 33rd Square

Embed

 Artificial Intelligence  

At the Evolving Artificial Intelligence Lab at the University of Wyoming, director Jeff Clune works to reproduce evolution in robotics. The goal is to ultimately evolve artificially intelligent robots that can rival natural animals in intelligence and agility.

At the University of Wyoming, Jeffrey Clune started up the Evolving Artificial Intelligence Lab last year and since then, four students have published research in peer-reviewed scientific journals. Several of his students have won national awards from Associated for Computing Machinery to NASA space grants.

Evolutionary computing simulates natural selection using a ‘survival of the fittest’ rule. The difference is that, instead of plants and animals competing, different versions of Clune's software are battling for their place in the next generation.

A Darwinian process is set up so that the better programs will have lots of copies (versions similar or identical) in the next generation and less desirable software is eliminated.

What makes some programs 'better' than others is determined by the person setting up the experiment (e.g., the ability to get out of mazes, drive a car without crashing, control a legged robot, etc.). Over time the software gets better and better since mutations (random changes in the programs) and 'sex' (combining a portion of the code of one program with a portion of another) will occasionally produce a program that is a slight improvement over its parents. This slightly better software will thrive for a while until it too is replaced by the next slightly better software.

Given enough generations, writes Clune, "These small changes can add up to produce jaguars, whales, Olympians and poets".

"We’re trying to harness the power of evolution. It’s an extremely creative and powerful design force. Can we use that process to evolve robots? We can harness it, and when we do, evolution comes up with something smarter than humans can design."

Clune and his team also apply the algorithms directly to physical robots. “To me, they look alive, not robotic,” Clune, told the Laramie Boomerang of the robots his team has made. “They’re quirky but still functional. They have that ‘je ne sais quoi’ of nature, with no human input.”

Artificial intelligence in robots is a software limitation, and most robots can’t walk across a floor without tripping, Clune said.

“When I read news like firefighters dying, I think we should be sending in robots to do that,” Clune said. “We’re trying to harness the power of evolution. It’s an extremely creative and powerful design force. Can we use that process to evolve robots? We can harness it, and when we do, evolution comes up with something smarter than humans can design.”

At it's core, the concept is Darwinian evolution and survival of the fittest, according to Clune.

As a visiting scientist at Cornell University, working with Hod Lipson, Clune successfully attempted the same process on a robot. He said other techniques were used to try to program the robot, but evolution was the only process to successfully make it walk.

“Nature already produced these designs,” Clune said. “We want to engineer robots that rival nature and are as agile and smart. If you tried writing code to do it, you probably couldn’t.

“Evolution has figured out how to build self-assembling molecules.”

Along with designing, building and evolving robots, Clune created endlessforms.com, a website named after Darwin's ""... from so simple a beginning endless forms most beautiful and most wonderful have been, and are being evolved."

Related articles MIT Researcher Teaches Robots To Adapt  Dennis Hong On Robot Evolution and The DARPA Robotics Challenge  Can Robots Invent Their Own Language?

At the site users can apply the same evolutionary techniques to common objects such as lamps or jewelry. By selecting a starting point, users then choose which attributes they want in their designs. Once the object looks as desired, it can be printed on a three-dimensional printer using any material.

The evolution of artificial intelligence — raises ethical concerns such as developing robots smarter and more powerful than humans. One of Clune’s lecture days is devoted to the topic of the ethics of his and similar work.

After earning two philosophy degrees, Clune decided to pursue computer science because he was fascinated by how intelligence works. He said he marveled at what nature produced and how a sense of being emerges from a collection of cells, leading him to try to understand evolving artificial intelligence.

“You as the designer get to say what you want and it’s up to evolution to produce and accomplish that goal,” Clune said. “I thought the best way to understand my own brain was to build one. I’m learning a lot about how evolution did it by doing it myself.”




SOURCES  Jeff Clune / Evolving AI Lab and Laramie Boomerang

By 33rd Square

Embed