A music search engine that uses a novel technique to classify songs,will go into beta this week.

I wrote about the system a few months ago. It was designed by researchers from the University of California, San Diego, including assistant professor Gert Lanckriet. The researchers have trained the search using information contributed by Facebook users, via an application called HerdIt. The goal is to train the system to tag songs automatically--using statistical analysis applied to the waveform patterns that represent each song:

About 90 percent of the time, Lanckriet says, the system identifies patterns that are ordinarily hidden. For example, the patterns that identify a hip-hop song might include a typical hip-hop beat, but also elements that the listener wouldn't recognize as a pattern within the song. "On average, these automatic tags predict other humans' [tags] pretty much as accurately as a given human person can do," Lanckriet says.[...] He envisions a system that could take an unfamiliar song--from an independent band, or even something recorded in a user's garage--and then analyze it on the fly and suggest appropriate tags and similar music.

I'm looking forward to trying it out. See the video below for a more detailed explanation of the project.

Courtesy of PNAS

Researchers at the Ecole Polytechnique Fédérale de Lausanne in Switzerland have found that robots equipped with artificial neural networks and programmed to find "food" eventually learned to conceal their visual signals from other robots to keep the food for themselves. The results are detailed in an upcoming PNAS study.

The team programmed small, wheeled robots with the goal of finding food: each robot received more points the longer it stayed close to "food" (signified by a light colored ring on the floor) and lost points when it was close to "poison" (a dark-colored ring). Each robot could also flash a blue light that other robots could detect with their cameras.

"Over the first few generations, robots quickly evolved to successfully locate the food, while emitting light randomly. This resulted in a high intensity of light near food, which provided social information allowing other robots to more rapidly find the food," write the authors.

The team "evolved" new generations of robots by copying and combining the artificial neural networksof the most successful robots. The scientists also added a few random changes to their code to mimic biological mutations.

Because space is limited around the food, the bots bumped and jostled each other after spotting the blue light. By the 50th generation, some eventually learned to not flash their blue light as much when they were near the food so as to not draw the attention of other robots, according to the researchers. After a few hundred generations, the majority of the robots never flashed light when they were near the food. The robots also evolved to become either highly attracted to, slightly attracted to, or repelled by the light.

Because robots were competing for food, they were quickly selected to conceal this information," the authors add.

The researchers suggest that the study may help scientists better understand the evolution of biological communication systems.

Courtesy of UCSD

Researchers at the University of California, San Diego (UCSD), who demoed a realistic-looking robot Einstein at the TED Conference last February, have now gone a step farther, infusing the robot with the ability to improve its own expressions through learning.

Previously, the head of the robot--designed by Hanson Robotics--could only respond to the people around it using a variety of preprogrammed expressions. With 31 motors and a realistic skinlike material called Frubber, the head delighted and surprised TED conference goers last winter.

Inspired by how babies babble to learn words and expressions, the UCSD researchers have now given the Einstein-bot its own learning ability. Instead of being preprogrammed to make certain facial expressions, the UCSD robot experiments in front of a mirror, gradually learning how its motors control its facial expressions. In this way, it learns to re-create particular expressions. The group presented its paper last month at the 2009 IEEE Conference on Development and Learning.

According to a press release from the university,

Once the robot learned the relationship between facial expressions and the muscle movements required to make them, the robot learned to make facial expressions it had never encountered.

Such an expressive robot could be useful as an assistant or teacher, or just as a means of learning more about how humans develop expressions. But a robot that watches itself in a mirror, practicing and improving how it looks, seems like another step into uncanny valley.