The Minds Mastering Machines (m3) Conference, a conference on Machine Learning, took place in Cologne last week and the topics covered from basics of Machine Learning to different Machine Learning optimizers and how they work and behave, but some fun parts, practical reports and ethics as topic was also covered. Thus, the opening keynote addressed the question of whether turn robot cars, service robots and chat bots into moral and immoral machines. The second keynote on the second day dealt with the question which problems AI can currently solve and where there are still difficulties.
In addition to practical reports, such as Artificial Intelligence in production systems or software processes such as testing and deployment for ML models in production, there was also a presentation on the attempt to explain machine learning models. Local Interpretable Model-Agnostic Explanations (LIME) is one attempt to make these complex models at least partially understandable.
Also interesting was the presentation on deep learning for recommendation systems, in which deep neural networks in different domains improve their recommendations. This was done using an example of an automobile sales platform.
Evolutionary programming was also an topic. The programming of a robot, which marks fields, was created with evolutionary programming.
There was also some fun, like the lecture: AI meets creativity – why every company needs an artificial muse! The Muse is a machine-learning system that stimulates the creativity of the painter Roman Lipski, inspires and provokes him and sets new impulses for his creative process.
Not to forget my presentation on CoreML and reinforcement learning with Minecraft. Boris Adryan tweeted parts of my lecture nearly live:
… and I had my first USB-C experience at a conference. I don’t think Machine Learning can solve this problem, only either a second laptop or a ton of adapters can do it.