AI gets a boost from curiosity

As data analytics become increasingly driven by artificial intelligence (AI),

photo credit: https://c1.staticflickr.com/3/2332/2083892100_3e015d810a_b.jpg

researchers search for a way to drive machine learning. The key ingredient its future development may be a dash of curiosity.

There are all kinds of AI systems currently used by various businesses with different names like Alexa and Albert to personalize then. Perhaps it's time for an AI system named George after the monkey whose curiosity propels him into various adventures.

That would be an apt choice for the Intrinsic Curiosity Module (ICM) developed by a group of four researchers at University of California, Berkeley. The attempt to inject curiosity to achieve self-motivated advances in machine learning was the subject of their paper, Curiosity-driven Exploration by Self-supervised Prediction, that was just submitted to the 34th International Conference on Machine Learning (ICML 2017).

Their premise is that external rewards for learning are of necessity limited and actually rather rare in real life. That doesn't mean that people stop exploring or seeking out answers even when there are no prizes for doing. They are motivated by their own human curiosity. Infusing that kind of motivation in a virtual agent gets it to test things out for itself even when not directed to do so. The test of the effect was done in monitoring how far it would proceed in two video games, VizDoom and Super Mario Bros. as you see in the demo video here:

Read more in Machine Learning Taps Power of Curiosity

Missions in space may come and go, but the National Aeronautics and Space Administration has always stuck to a mission of bringing in data.

(

One of its early achievements in this field was sending a spacecraft close enough to Venus to get accurate readings of its surface and atmosphere. On Dec. 14, 1962, the Mariner 2 spacecraft got within 34,762km (21,600 miles) of the planet. Over a 42-minute period, it was able to pick up many points of data that proved Venus, which had been thought of as Earth's twin, would be uninhabitable, with a surface temperature of 425°C (797°F) and a toxic atmosphere.

This picture (from NASA's site) of the data gathered in that mission is cropped. The paper showing the data that was gathered is actually much longer, as this uncropped version shows.

Back then, the data covered a roll of paper, but the data NASA handles today takes supercomputing power to process. As Nick Skytland wrote in NASA blog post in October:

In the time it took you to read this sentence, NASA gathered approximately 1.73 gigabytes of data from our nearly 100 currently active missions! We do this every hour, every day, every year -- and the collection rate is growing exponentially...
In our current missions, data is transferred with radio frequency, which is relatively slow. In the future, NASA will employ technology such as optical (laser) communication to increase the download and mean a 1000x increase in the volume of data. This is much more then we can handle today and this is what we are starting to prepare for now. We are planning missions today that will easily stream more [than] 24TB's a day. That's roughly 2.4 times the entire Library of Congress -- EVERY DAY. For one mission.

read more at 

Big Data on the Final Frontier