AI-based optimization: development of learning algorithms for complex tasks

Context

Optimization problems are fundamental in science and engineering: from tuning physical systems, to controlling robots, and training deep neural networks. Traditionally, optimization has been approached using rule-based solvers such as gradient descent or evolutionary algorithms.

In recent years, AI-based optimization is becoming a powerful alternative to traditional methods. Instead of solving every task from scratch, learning-based approaches aim to discover adaptable optimization strategies that generalize across problems.

In this thesis, you will explore AI-driven strategies for solving optimization problems. You may investigate how learning-based methods can be applied to structured decision problems, continuous optimization, or dynamic control scenarios with a focus on designing systems that can adapt and generalize across tasks.

Tasks

The concrete tasks are related to the topic that you select, but the general process of the work is summarized as follows:

• Literature research of the SOTA approaches in AI-based optimization
• Design and implementation of a learning framework for optimization tasks (could be but not limited to regression or reinforcement learning problems)
• Training and evaluation of the system using simulated or real-world optimization scenarios, including comparison with traditional optimization methods
• Investigation of generalization and adaptability across task variations (such as MAML or few shot learning)
• Evaluation and analysis of the results

Requirements

• High motivation to learn new technologies
• Knowledge of machine learning algorithms, ideally reinforcement learning
• Experience with programming languages such as Python, C++ and Java
• Analytical, problem-solving, and communication skills