Abstract
Deep Learning is an exciting field that created a multi-billion industry. But it’s not the only approach to Machine Learning (ML) and AI, nature-inspired algorithms are making a comeback. There’s a renewed interest in Evolutionary Algorithms, and Neuroevolution in particular, yet there are relatively few modern libraries and frameworks compared with Deep Learning and Reinforcement Learning.

Darwin is an open source C++ framework ([ Ссылка ]) intended to make EA & Neuroevolution experiments easy, quick and fun. It provides building blocks, samples and tooling to avoid the repetitive (and potentially complex) scaffolding required to research new ideas.

In this talk I’ll introduce Evolutionary Algorithms, go through a quick overview of the Darwin Framework and I’ll share a few engineering lessons learned from designing and implementing it.
Speaker Bio
By day, Leonard Mosescu is a compiler engineer, having worked on various parts of the Microsoft Visual C++ toolchain, then on the Android runtime and dev tools for Google, and currently working on the Nvidia CUDA toolchain. By night, he’s pursuing an old interest in AI and Evolutionary Algorithms.

свернуть