Uni-Logo

Lecture: Automated Machine Learning

Background

Based on machine learning (ML), AI achieved major breakthroughs in the last years. However, applying machine learning and in particular deep learning (DL) in practice is a challenging task and requires a lot of expertise. Among other things, the success of ML/DL applications depends on many design decisions, including an appropriate preprocessing of the data, choosing a well-performing machine learning algorithm and tuning its hyperparameters, giving rise to a complex pipeline. Unfortunately, even experts need days, weeks or even months to find well-performing pipelines and can still make mistakes when optimizing their pipelines. The course will discuss meta-algorithmic approaches to automatically search for, and obtain well-performing machine learning systems by means of automated machine learning (AutoML). Such AutoML systems allow for faster development of new ML/DL applications, require far less expert knowledge than doing everything from scratch and often even outperform human developers. In this lecture, you will learn how to use such AutoML systems, to develop your own systems and to understand ideas behind state-of-the-art AutoML approaches.

Requirements

We strongly recommend that you know the foundations of

  • machine learning (ML)
  • and deep learning (DL) in order to attend the course.

We further recommend that you have hands-on experience with:

  • Python (3.6+)
  • machine learning
  • deep learning

The participants should have attended at least one other course for ML and DL in the past.

Topics

The lectures are partitioned in several parts, including:

  • Algorithm Selection
  • Meta-Learning
  • Hyperparameter Optimization
  • Neural Architecture Search
  • Learning2Learn
  • Dynamic Configuration
  • Analysis of AutoML
  • Algorithm Configuration

Dynamics

The course will be in English.

The course will be taught in a flipped-classroom style. We will meet weekly once for a combined Q/A session and an exercise. Roughly every week, there will be a new exercise sheet. Most exercises will be practical and involve teamwork (teams of 2 students!) so that you learn how to apply AutoML in practice.

  • Lecture/Exercise: Tuesday 14:00 (s.t.) - 15:30am, online-only!

Exam

In the end, everyone (no teamwork!) has to implement a larger project which is the base of the final oral exam. In the first 15 minutes of the oral exam, you have to present your project and in the second 15 minutes, we will ask you questions about further course material.