Meta-learning: learning to learn. Generalization of neural network capability for task series.

Environment

Neural networks have demonstrated their strong capabilities in various domains, such as recognizing objects into images, translating language, and playing games. However, a neural network is trained only for a specific task and the ability cannot be adopted in another scenario. However, a neural network that can recognize handwriting cannot figure out the differences between cats and dogs.

The meta-learning method describes the philosophy of "learning to learn". Here, a set of tasks are considered together and the generality is learned. A Machine Learning model learns here not only the tasks themselves, but also how to train itself better.

Task

Using classification, regression and reinforcement learning problems as examples, various meta-learning approaches are researched and implemented.

The effects of the meta-learning procedure are then documented and evaluated.

Prerequisites

• Desire and motivation for new technology
• Ideally experience in
  • Programming (Python, C++...)
  • machine learning