Intelligent Decision Support on Networks - Data-driven Optimization, Augmented and Explainable AI in Complex Supply Chains

Permanent URI for this collectionhttps://hdl.handle.net/10125/107436

Browse

Recent Submissions

Now showing 1 - 3 of 3
  • Thumbnail Image
    Item type: Item ,
    Detection of Important States through an Iterative Q-value Algorithm for Explainable Reinforcement Learning
    (2024-01-03) Milani, Rudy; Moll, Maximilian; De Leone, Renato
    To generate safe and trustworthy Reinforcement Learning agents, it is fundamental to recognize meaningful states where a particular action should be performed. Thus, it is possible to produce more accurate explanations of the behaviour of the trained agent and simultaneously reduce the risk of committing a fatal error. In this study, we improve existing metrics using Q-values to detect essential states in Reinforcement Learning by introducing a scaled iterated algorithm called IQVA. The key observation of our approach is that a state is important not only if the action has a high impact but also if it often appears in different episodes. We compared our approach with the two baseline measures and a newly introduced value in grid-world environments to demonstrate its efficacy. In this way, we show how the proposed methodology can highlight only the meaningful states for that particular agent instead of emphasizing the importance of states that are rarely visited.
  • Thumbnail Image
    Item type: Item ,
    Machine Learning in Vehicle Travel Time Estimation: A Brief Technological Perspective and Review
    (2024-01-03) Pham, Truong Son; Nistor, Marian Sorin; Cao, Loi; Gerschberger, Markus; Moll, Maximilian
    A precise Estimated Time of Arrival (ETA) finds applications in various domains, such as navigation and logistics systems. This problem has gained a lot of attention from the research community. Machine learning has recently been applied and has shown promising results for ETA. Machine learning approaches can be divided into two categories, which are route-based and origin-destination-based methods. The first one divides the route into segments and predicts the ETA based on the information of these segments. The last one predicts ETA based on a few natural information, such as the origin, the estimation, and the departure time. In this paper, we aim to review recent studies of the mentioned machine learning approaches for ETA to determine the necessary input for an ETA forecasting model, the critical factors, and suitable approaches for ETA. Furthermore, we will discuss promising research directions to improve ETA, such as formulating ETA as a time series forecasting problem, including uncertainty or using ensemble learning models.
  • Thumbnail Image
    Item type: Item ,