Bias Variance

یکی از مقالات مهم در زمینه روشهای value-based در یادگیری عمیق تقویتی مقاله Dueling Network Architectures for Deep Reinforcement Learning است. این روش عملکرد بسیار بهتری از DQN دارد.

In recent years there have been many successes of using deep representations in reinforcement learning. Still, many of these applications use conventional architectures, such as convolutional networks, LSTMs, or auto-encoders. In this paper, we present a new neural network architecture for model-free reinforcement learning. Our dueling network represents two separate estimators: one for the state value function and one for the state-dependent action advantage function. The main benefit of this factoring is to generalize learning across actions without imposing any change to the underlying reinforcement learning algorithm. Our results show that this architecture leads to better policy evaluation in the presence of many similar-valued actions. Moreover, the dueling architecture enables our RL agent to outperform the state-of-the-art on the Atari 2600 domain.

لینک مقاله: https://arxiv.org/abs/1511.06581
لینک پیاده سازی های متفاوت: https://paperswithcode.com/paper/dueling-network-architectures-for-deep


#معرفی_مقاله #یادگیری_عمیق #یادگیری_تقویتی #value_based

🌴 سایت | 🌺 کانال | 🌳 پشتیبانی

Paperswithcode

Papers with Code - Dueling Network Architectures for Deep Reinforcement Learning

🏆 SOTA for Atari Games on Atari 2600 Pong (Score metric)

www.tgoop.com/biasvariance_ir/94

183 viewsBias Variance, edited  Sep 29, 2021 at 18:45

یکی از مقالات مهم در زمینه روشهای value-based در یادگیری عمیق تقویتی مقاله Dueling Network Architectures for Deep Reinforcement Learning است. این روش عملکرد بسیار بهتری از DQN دارد.

In recent years there have been many successes of using deep representations in reinforcement learning. Still, many of these applications use conventional architectures, such as convolutional networks, LSTMs, or auto-encoders. In this paper, we present a new neural network architecture for model-free reinforcement learning. Our dueling network represents two separate estimators: one for the state value function and one for the state-dependent action advantage function. The main benefit of this factoring is to generalize learning across actions without imposing any change to the underlying reinforcement learning algorithm. Our results show that this architecture leads to better policy evaluation in the presence of many similar-valued actions. Moreover, the dueling architecture enables our RL agent to outperform the state-of-the-art on the Atari 2600 domain.

لینک مقاله: https://arxiv.org/abs/1511.06581
لینک پیاده سازی های متفاوت: https://paperswithcode.com/paper/dueling-network-architectures-for-deep


#معرفی_مقاله #یادگیری_عمیق #یادگیری_تقویتی #value_based

🌴 سایت | 🌺 کانال | 🌳 پشتیبانی

BY Bias Variance