Lipschitzness is all you need to tame off-policy generative adversarial imitation learning

DOI10.1007/s10994-022-06144-5OpenAlexW3038629022MaRDI QIDQ2163202

Alexandros Kalousis, Lionel Blondé, Pablo Strasser

Publication date: 10 August 2022

Published in: Machine Learning (Search for Journal in Brave)

Full work available at URL: https://arxiv.org/abs/2006.16785

zbMATH Keywords

reinforcement learningimitation learningdeep learningLipschitz-continuitygenerative adversarial networks

Mathematics Subject Classification ID

Learning and adaptive systems in artificial intelligence (68T05)

Uses Software

• OpenAI Gym
• PyTorch
• Adam
• MuJoCo
• mixup
• Waymo
• Safety Gym
• Wasserstein GAN

Cites Work

• Unnamed Item
• Unnamed Item
• Unnamed Item
• Unnamed Item
• Unnamed Item
• Unnamed Item
• Batch mode reinforcement learning based on the synthesis of artificial trajectories
• \({\mathcal Q}\)-learning
• Regularisation of neural networks by enforcing Lipschitz continuity
• Reinforcement Learning in Robust Markov Decision Processes
• On Upper-Confidence Bound Policies for Switching Bandit Problems
• Flat Minima
• A massively parallel architecture for a self-organizing neural pattern recognition machine
• Sample Selection Bias as a Specification Error
• Optimal Exploration–Exploitation in a Multi-armed Bandit Problem with Non-stationary Rewards
• A survey on concept drift adaptation
• Robust Control of Markov Decision Processes with Uncertain Transition Matrices
• Some aspects of the sequential design of experiments