"I'm sorry Dave, I'm afraid I can't do ...
Document type :
Communication dans un congrès avec actes
Title :
"I'm sorry Dave, I'm afraid I can't do that" Deep Q-Learning From Forbidden Actions
Author(s) :
Seurin, Mathieu [Auteur]
Scool [Scool]
Sequential Learning [SEQUEL]
Centre de Recherche en Informatique, Signal et Automatique de Lille - UMR 9189 [CRIStAL]
Preux, Philippe [Auteur]
Scool [Scool]
Sequential Learning [SEQUEL]
Pietquin, Olivier [Auteur]
Google Brain, Paris
Scool [Scool]
Sequential Learning [SEQUEL]
Centre de Recherche en Informatique, Signal et Automatique de Lille - UMR 9189 [CRIStAL]
Preux, Philippe [Auteur]
Scool [Scool]
Sequential Learning [SEQUEL]
Pietquin, Olivier [Auteur]
Google Brain, Paris
Conference title :
Internationnal Joint Conference on Neural Networks
City :
Glasgow
Country :
Royaume-Uni
Start date of the conference :
2020-07-17
English keyword(s) :
Deep Reinforcement Learning
Safety
constraints
Q-Learning
Safety
constraints
Q-Learning
HAL domain(s) :
Informatique [cs]/Intelligence artificielle [cs.AI]
Informatique [cs]/Apprentissage [cs.LG]
Informatique [cs]/Réseau de neurones [cs.NE]
Informatique [cs]/Apprentissage [cs.LG]
Informatique [cs]/Réseau de neurones [cs.NE]
English abstract : [en]
The use of Reinforcement Learning (RL) is still restricted to simulation or to enhance human-operated systems through recommendations. Real-world environments (e.g. industrial robots or power grids) are generally designed ...
Show more >The use of Reinforcement Learning (RL) is still restricted to simulation or to enhance human-operated systems through recommendations. Real-world environments (e.g. industrial robots or power grids) are generally designed with safety constraints in mind implemented in the shape of valid actions masks or contingency controllers. For example, the range of motion and the angles of the motors of a robot can be limited to physical boundaries. Violating constraints thus results in rejected actions or entering in a safe mode driven by an external controller, making RL agents incapable of learning from their mistakes. In this paper, we propose a simple modification of a state-of-the-art deep RL algorithm (DQN), enabling learning from forbidden actions. To do so, the standard Q-learning update is enhanced with an extra safety loss inspired by structured classification. We empirically show that it reduces the number of hit constraints during the learning phase and accelerates convergence to near-optimal policies compared to using standard DQN. Experiments are done on a Visual Grid World Environment and Text-World domain.Show less >
Show more >The use of Reinforcement Learning (RL) is still restricted to simulation or to enhance human-operated systems through recommendations. Real-world environments (e.g. industrial robots or power grids) are generally designed with safety constraints in mind implemented in the shape of valid actions masks or contingency controllers. For example, the range of motion and the angles of the motors of a robot can be limited to physical boundaries. Violating constraints thus results in rejected actions or entering in a safe mode driven by an external controller, making RL agents incapable of learning from their mistakes. In this paper, we propose a simple modification of a state-of-the-art deep RL algorithm (DQN), enabling learning from forbidden actions. To do so, the standard Q-learning update is enhanced with an extra safety loss inspired by structured classification. We empirically show that it reduces the number of hit constraints during the learning phase and accelerates convergence to near-optimal policies compared to using standard DQN. Experiments are done on a Visual Grid World Environment and Text-World domain.Show less >
Language :
Anglais
Peer reviewed article :
Non
Audience :
Internationale
Popular science :
Non
Collections :
Source :
Files
- https://hal.inria.fr/hal-02387419v2/document
- Open access
- Access the document
- https://hal.inria.fr/hal-02387419v2/document
- Open access
- Access the document
- https://hal.inria.fr/hal-02387419v2/document
- Open access
- Access the document
- document
- Open access
- Access the document
- Dave_IJCNN.pdf
- Open access
- Access the document