About

Research Interests (Lay Summary)

Eleni started her PhD in I-Form in August 2019 and is investigating the use of artificial intelligence to improve decision-making aspects of 3D printing. It is very common that during the manufacturing process many decisions on safety and product quality need to be made. The main idea of this research is to monitor the manufacturing process through sensors, manipulate the obtained data in order to locate and understand any flaws in the process, and implement corrective actions using intelligent decision-making.

Technical Summary

The area of dynamic programming focuses on sequential decision-making when the outcome of each decision is not fully predictable. The problem is usually formalised as a Markov Decision Process and the objective is to determine the optimal course of action that minimises a certain cost or maximises a reward. In general, we are trying to make a decision under uncertainty, which can be caused by various sources. Machine learning algorithms are used to define the scope of that uncertainty where exact probabilistic models are not available.

The main approach that will be used in this research is deep reinforcement learning (RL). With the use of deep neural networks as powerful function approximators that have great representation learning properties (i.e. feature extraction), reinforcement learning can scale decision-making problems that were previously intractable, such as problems with high-dimensional state and action spaces. Thanks to its ability to learn different levels of abstractions from data, deep RL has been successful in complicated tasks with low prior knowledge of the data.