Atari Assault Environment

Overview

The player is presented with an alien mother ship, which continually deploys three smaller ships during play.[2] The mother ship and the smaller vessels shoot at a weapon the player is in command of, and the player’s aim is to eliminate the opposition while preventing the weapon from receiving enough fire to destroy it.[2] The player uses a joystick to operate the game, and only one player at a time can play.[1]

Description from Wikipedia

Performances of RL Agents

We list various reinforcement learning algorithms that were tested in this environment. These results are from RL Database. If this page was helpful, please consider giving a star!

Star

Human Starts

Result Algorithm Source
14497.9 A3C LSTM Asynchronous Methods for Deep Reinforcement Learning
14491.7 Rainbow Rainbow: Combining Improvements in Deep Reinforcement Learning
10950.6 PDD DQN Dueling Network Architectures for Deep Reinforcement Learning
7748.5 Prioritized DDQN (prop, tuned) Prioritized Experience Replay
6548.9 Prioritized DDQN (rank, tuned) Prioritized Experience Replay
6060.8 DDQN (tuned) Deep Reinforcement Learning with Double Q-learning
5474.9 A3C FF Asynchronous Methods for Deep Reinforcement Learning
5101.3 Distributional DQN Rainbow: Combining Improvements in Deep Reinforcement Learning
3994.8 DuDQN Dueling Network Architectures for Deep Reinforcement Learning
3746.1 A3C FF 1 day Asynchronous Methods for Deep Reinforcement Learning
3332.3 DQN Massively Parallel Methods for Deep Reinforcement Learning
3081.3 Prioritized DQN (rank) Prioritized Experience Replay
2774.3 DDQN Deep Reinforcement Learning with Double Q-learning
1195.85 Gorila DQN Massively Parallel Methods for Deep Reinforcement Learning
628.9 Human Massively Parallel Methods for Deep Reinforcement Learning
166.9 Random Massively Parallel Methods for Deep Reinforcement Learning

No-op Starts

Result Algorithm Source
29091 IQN Implicit Quantile Networks for Distributional Reinforcement Learning
22012 QR-DQN-1 Distributional Reinforcement Learning with Quantile Regression
19961 QR-DQN-0 Distributional Reinforcement Learning with Quantile Regression
19148.47 IMPALA (deep) IMPALA: Scalable Distributed Deep-RL with Importance Weighted Actor-Learner Architectures
17543.8 Reactor ND The Reactor: A fast and sample-efficient Actor-Critic agent for Reinforcement Learning
14198.5 Rainbow Rainbow: Combining Improvements in Deep Reinforcement Learning
12086.86 IMPALA (shallow) IMPALA: Scalable Distributed Deep-RL with Importance Weighted Actor-Learner Architectures
11477.0 PDD DQN Dueling Network Architectures for Deep Reinforcement Learning
11231 NoisyNet DuDQN Noisy Networks for Exploration
11013.5 Reactor The Reactor: A fast and sample-efficient Actor-Critic agent for Reinforcement Learning
10777.7 ACKTR Scalable trust-region method for deep reinforcement learning using Kronecker-factored approximation
8323.3 Reactor The Reactor: A fast and sample-efficient Actor-Critic agent for Reinforcement Learning
8010 DuDQN Noisy Networks for Exploration
7203 C51 A Distributional Perspective on Reinforcement Learning
5909.0 Distributional DQN Rainbow: Combining Improvements in Deep Reinforcement Learning
5510 NoisyNet DQN Noisy Networks for Exploration
5393.2 DDQN A Distributional Perspective on Reinforcement Learning
5022.9 DDQN Deep Reinforcement Learning with Double Q-learning
4621.0 DuDQN Dueling Network Architectures for Deep Reinforcement Learning
4280.4 DQN A Distributional Perspective on Reinforcement Learning
3595 DQN Noisy Networks for Exploration
3359 DQN Human-level control through deep reinforcement learning
3060 NoisyNet A3C Noisy Networks for Exploration
2879 A3C Noisy Networks for Exploration
2116.32 IMPALA (deep, multitask) IMPALA: Scalable Distributed Deep-RL with Importance Weighted Actor-Learner Architectures
1496.4 Human Human-level control through deep reinforcement learning
1450.41 Gorila DQN Massively Parallel Methods for Deep Reinforcement Learning
742.0 Human Dueling Network Architectures for Deep Reinforcement Learning
628 Linear Human-level control through deep reinforcement learning
537 Contingency Human-level control through deep reinforcement learning
222.4 Random Human-level control through deep reinforcement learning

Normal Starts

Result Algorithm Source
4971.9 PPO Proximal Policy Optimization Algorithm
4653.8 ACER Proximal Policy Optimization Algorithm
1562.9 A2C Proximal Policy Optimization Algorithm