Konstantinos Tsampazis
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Konstantinos Tsampazis has a total of 4 co-authors. Here are the top co-authors by number of shared publications:
- Anastasios Tefas
Aristotle University of Thessaloniki
3 shared publications - Manos Kirtas
Computational Intelligence, Deep Learning Group, AIIA Lab, Department of Informatics, Aristotle University of Thessaloniki, Thessaloniki, Greece
3 shared publications - Nikolaos Passalis
Aristotle University of Thessaloniki, Thessaloniki, Greece
3 shared publications - Loukia Avramelou
Aristotle University of Thessaloniki
2 shared publications
Latest Publications
- 1 January 2025
- Published by Springer Nature
- p. 420-432
Using Part-Based Representations for Explainable Deep Reinforcement Learning
- 1 January 2025
- Published by Springer Nature
- p. 420-432
Using Part-Based Representations for Explainable Deep Reinforcement Learning
- Open Access
Using Part-based Representations for Explainable Deep Reinforcement Learning
Utilizing deep learning models to learn part-based representations holds significant potential for interpretable-by-design approaches, as these models incorporate latent causes obtained from feature representations through simple addition. However, training a part-based learning model presents challenges, particularly in enforcing non-negative constraints on the model's parameters, which can result in training difficulties such as instability and convergence issues. Moreover, applying such approaches in Deep Reinforcement Learning (RL) is even more demanding due to the inherent instabilities that impact many optimization methods. In this paper, we propose a non-negative training approach for actor models in RL, enabling the extraction of part-based representations that enhance interpretability while adhering to non-negative constraints. To this end, we employ a non-negative initialization technique, as well as a modified sign-preserving training method, which can ensure better gradient flow compared to existing approaches. We demonstrate the effectiveness of the proposed approach using the well-known Cartpole benchmark.
Open AccessUsing Part-based Representations for Explainable Deep Reinforcement Learning
Utilizing deep learning models to learn part-based representations holds significant potential for interpretable-by-design approaches, as these models incorporate latent causes obtained from feature representations through simple addition. However, training a part-based learning model presents challenges, particularly in enforcing non-negative constraints on the model's parameters, which can result in training difficulties such as instability and convergence issues. Moreover, applying such approaches in Deep Reinforcement Learning (RL) is even more demanding due to the inherent instabilities that impact many optimization methods. In this paper, we propose a non-negative training approach for actor models in RL, enabling the extraction of part-based representations that enhance interpretability while adhering to non-negative constraints. To this end, we employ a non-negative initialization technique, as well as a modified sign-preserving training method, which can ensure better gradient flow compared to existing approaches. We demonstrate the effectiveness of the proposed approach using the well-known Cartpole benchmark.
- 29 May 2020
- Published by Springer Nature
- Vol. 584, 64-75
Deepbots: A Webots-Based Deep Reinforcement Learning Framework for Robotics
Deep Reinforcement Learning (DRL) is increasingly used to train robots to perform complex and delicate tasks, while the development of realistic simulators contributes to the acceleration of research on DRL for robotics. However, it is still not straightforward to employ such simulators in the typical DRL pipeline, since their steep learning curve and the enormous amount of development required to interface with DRL methods significantly restrict their use by researchers. To overcome these limitations, in this work we present an open-source framework that combines an established interface used by DRL researchers, the OpenAI Gym interface, with the state-of-the-art Webots robot simulator in order to provide a standardized way to employ DRL in various robotics scenarios. Deepbots aims to enable researchers to easily develop DRL methods in Webots by handling all the low-level details and reducing the required development effort. The effectiveness of the proposed framework is demonstrated through code examples, as well as using three use cases of varying difficulty.
- 29 May 2020
- Published by Springer Nature
- Vol. 584, 64-75
Deepbots: A Webots-Based Deep Reinforcement Learning Framework for Robotics
Deep Reinforcement Learning (DRL) is increasingly used to train robots to perform complex and delicate tasks, while the development of realistic simulators contributes to the acceleration of research on DRL for robotics. However, it is still not straightforward to employ such simulators in the typical DRL pipeline, since their steep learning curve and the enormous amount of development required to interface with DRL methods significantly restrict their use by researchers. To overcome these limitations, in this work we present an open-source framework that combines an established interface used by DRL researchers, the OpenAI Gym interface, with the state-of-the-art Webots robot simulator in order to provide a standardized way to employ DRL in various robotics scenarios. Deepbots aims to enable researchers to easily develop DRL methods in Webots by handling all the low-level details and reducing the required development effort. The effectiveness of the proposed framework is demonstrated through code examples, as well as using three use cases of varying difficulty.