Can Machine Learning Save the Whales?: In the 1960’s whales faced mass extinction from whaling. Careful scientific study and regulatory action saved them. Now many whale species are facing another wave of extinction due to poorly understood forces. Is it depletion of their food stock? Collisions with ships? High-energy sonar? The problem is not so obvious this time. I’ll argue that we need breakthroughs in machine learning to figure it out. I’ll discuss two new projects and the impact they had on our expeditions to Alaska in 2017. In the first, we explore the use of new features to identify individual whales. In the second, we use deep learning-based morphometry to study their energy stores, which are vital to their survival and reproduction. Both problems suffer data starvation and challenges unique to the marine environment. I’ll discuss our race to clear these hurdles and save the whales.
Session Summary
Can Machine Learning Save the Whales?
MLconf 2017 San Francisco
Ted Willke
Intel
Director of the Brain-Inspired Computing Lab
Learn more »