AI Paper Atlas

Research map / Benchmarks / Breakout

Breakout leaderboard

10 papers at ICML, NeurIPS, ICLR, CVPR, AAAI, ICCV and ECCV, and the journals TPAMI, IJCV and JMLR report results on Breakout. Best score: 532.8 (NGU(N=32), ICLR 2020).

Numbers are as each paper reports them, read by AI from its results tables and not checked by hand. Papers can differ in setting (training data, model size, extra data); the topic column shows what each paper is about. Open the paper before you compare or quote a number.

Breakout · score

10 papers, each with its best reported number for its own method; higher is better.

#score ↑MethodPaperTopic
1532.8NGU(N=32)
R2D2
Never Give Up: Learning Directed Exploration Strategies
ICLR 2020 · Dense reward results table
Visual and representation RL
2423MMN (B_h=64, B_f=100), after fine-tuning
RNN
Learning Finite State Representations of Recurrent Policy Networks
ICLR 2019 · Atari table
Visual and representation RL
3400Knowledge Flow (expert teachers)
A3C
Knowledge Flow: Improve Upon Your Teachers
ICLR 2019 · Table 2
Knowledge distillation
4372.8V-MCTS
EfficientZero
Spending Thinking Time Wisely: Accelerating MCTS with Virtual Expansions
NeurIPS 2022 · Table: Results for Atari games
Combinatorial and search optimization
5360.5MPOMaximum a Posteriori Policy Optimisation
ICLR 2018 · Table: Results on a subset of
Policy gradient and actor-critic
6286.7Converted SNN
DQN
Strategy and Benchmark for Converting Deep Q-Networks to Event-Driven Spiking Neural Networks
AAAI 2021 · Table 1
Spiking networks
7225.3Mega-reward
PPO
Mega-Reward: Achieving Human-Level Play without Extrinsic Rewards
AAAI 2020 · Table: Comparison of mega-rewa
RL from feedback
8221.3T-REX
PPO
Extrapolating Beyond Suboptimal Demonstrations via Inverse Reinforcement Learning from Observations
ICML 2019 · Table 1
Imitation learning
9100DQN + preferences
DQN
Reward learning from human preferences and demonstrations in Atari
NeurIPS 2018 · Comparison with policy gradien
Imitation learning
1087.63DT + TLB
Decision Transformer
Temporal Latent Bottleneck: Synthesis of Fast and Slow Processing Mechanisms in Sequence Learning
NeurIPS 2022 · Atari table
RNNs and normalization

Not affiliated with any conference, journal or the benchmark's authors. Corrections: contact@latenticlabs.com.