deep reinforcement learning

Protected: Research Trends in Deep Reinforcement Learning: Meta-Learning and Transfer Learning, Intrinsic Motivation and Curriculum Learning

Research trends in deep reinforcement learning for digital transformation, artificial intelligence, and machine learning tasks: meta-learning and transfer learning, intrinsic motivation and curriculum learning automatic curriculum generation, automatic task decomposition, task difficulty adjustment, intrinsic reward, robot domain transformation, robot domain transformation, simulator to simulator transfer learning, BERT, Metric/Representation Base, Memory/Knowledge Base, active learning, meta-learning, and robot domain transformation) Robot domain transformation, transfer learning from simulators, BERT, Model-Agnostic Meta-Learning, Active Learning, Metric/Representation Base, Memory/Knowledge Base, Weigh Base, and Learning to Optimize

2023.05.11

アルゴリズム:Algorithmsグラフ理論スパースモデリング幾何学:Geometry強化学習微分積分:Calculus最適化:Optimization機械学習:Machine Learning深層学習:Deep Learning確率・統計:Probability and Statistics線形代数:Linear Algebra

Protected: Implementation of two approaches to improve environmental awareness, a weak point of deep reinforcement learning.

Implementation of two approaches to improve environment awareness, a weakness of deep reinforcement learning used in digital transformation, artificial intelligence, and machine learning tasks (inverse predictive, constrained, representation learning, imitation learning, reconstruction, predictive, WorldModels, transition function, reward function Weaknesses of representation learning, VAE, Vision Model, RNN, Memory RNN, Monte Carlo methods, TD Search, Monte Carlo Tree Search, Model-based learning, Dyna, Deep Reinforcement Learning)

2023.04.27

アルゴリズム:Algorithmsグラフ理論スパースモデリングマルチエージェントシステム幾何学:Geometry強化学習微分積分:Calculus数理論理学:Mathematical logic最適化:Optimization機械学習:Machine Learning深層学習:Deep Learning確率・統計:Probability and Statistics

Protected: Overview of Weaknesses and Countermeasures in Deep Reinforcement Learning and Two Approaches to Improve Environment Recognition

An overview of the weaknesses and countermeasures of deep reinforcement learning utilized in digital transformation, artificial intelligence, and machine learning tasks and two approaches of improving environmental awareness Mixture Density Network, RNN, Variational Auto Encoder, World Modles, Expression Learning, Strategy Network Compression, Model Free Learning, Sample-Based Planning Model, Dyna, Simulation-Based, Sample-Based, Gaussian Process, Neural Network, Transition Function, Reward Function) World Modles, Representation Learning, Strategy Network Compression, Model-Free Learning, Sample-Based Planning Model, Dyna, Simulation-Based, Sample-Based, Gaussian Process, Neural Network, Transition Function, Reward Function, Simulator , learning capability, transition capability

2023.04.13

アルゴリズム:Algorithmsグラフ理論スパースモデリング幾何学:Geometry強化学習微分積分:Calculus最適化:Optimization機械学習:Machine Learning深層学習:Deep Learning確率・統計:Probability and Statistics線形代数:Linear Algebra

Protected: Value Assessment and Policy and Weaknesses in Deep Reinforcement Learning

Value assessment and strategies and weaknesses in deep reinforcement learning used for digital transformation, artificial intelligence, and machine learning tasks poor sample efficiency, difficulty in validating methods as well, impact of implementation practices on performance, library initial values, poor reproducibility, over-training, local optimum, dexterity, TRPO, PPO, continuous value control, image control, policy-based, value-based

2023.03.30

アルゴリズム:Algorithmsグラフ理論スパースモデリング幾何学:Geometry強化学習微分積分:Calculus最適化:Optimization機械学習:Machine Learning深層学習:Deep Learning確率・統計:Probability and Statistics線形代数:Linear Algebra

Protected: Implementation of Model-Free Reinforcement Learning in python (3)Using experience for value assessment or strategy update: Value-based vs. policy-based

Value-based and policy-based implementations of model-free reinforcement learning in python for digital transformation, artificial intelligence, and machine learning tasks

2022.12.02

pythonアルゴリズム:Algorithms強化学習微分積分:Calculus最適化:Optimization機械学習:Machine Learning確率・統計:Probability and Statistics

Protected: New Developments in Reinforcement Learning (2) – Approaches Using Deep Learning

On seven methods for improving deep reinforcement learning used in digital transformation , artificial intelligence , and machine learning tasks (first generation DQN, dual Q learning (dual DQN method), prioritized experience replay, collision Q networks, distributed reinforcement learning (categorical DQN method) noise networks, n-step cutting returns) and alpha zero

2022.02.07

オンライン学習強化学習微分積分:Calculus最適化:Optimization機械学習:Machine Learning深層学習:Deep Learning確率・統計:Probability and Statistics