Alphabet-owned Waymo has partnered with UK-based computer programmers company DeepMind on a ‘Population-Based Training’ method for pedestrian detection that has the best neural networks like life forms do in natural selection, thus, saving time and effort.
According to the company, Waymo’s self-driving vehicles employ neural networks to perform many driving tasks, from detecting objects and predicting how others will behave, to planning a car’s next moves.
“Now, Waymo, in a research collaboration with DeepMind, has taken inspiration from Darwin’s insights into evolution to make this training more effective and efficient,” Yu-hsin Chen, Senior Software Engineer at Waymo recently wrote in a blog post.
“To make this process more efficient, researchers at DeepMind devised a way to automatically determine good hyper parameter schedules based on evolutionary competition (called ‘Population-Based Training’ or PBT), which combines the advantages of hand-tuning and random search,” Chen added.
Like random search, PBT also starts with multiple networks initiated with random hyper parameters.