# A Mixture of Surprises for Unsupervised Reinforcement Learning

arXiv: https://arxiv.org/abs/2210.06702 · HTML: https://mlsanity.com/paper/2210.06702/

## Key facts

- Cited by 9 papers in the map (data through 14 Oct 2026).
- Builds on 42 of its 67 references that are in the map.
- First posted on arXiv 13 Oct 2022.

## Builds on (most cited first, 30 of 42)

- [Language Models are Few-Shot Learners](https://mlsanity.com/paper/2005.14165/) (2020)
- [PyTorch: An Imperative Style, High-Performance Deep Learning Library](https://mlsanity.com/paper/1912.01703/) (2019)
- [Representation Learning with Contrastive Predictive Coding](https://mlsanity.com/paper/1807.03748/) (2018)
- [Masked Autoencoders Are Scalable Vision Learners](https://mlsanity.com/paper/2111.06377/) (2021)
- [Continuous control with deep reinforcement learning](https://mlsanity.com/paper/1509.02971/) (2015)
- [Deep Reinforcement Learning with Double Q-learning](https://mlsanity.com/paper/1509.06461/) (2015)
- [Spectral Normalization for Generative Adversarial Networks](https://mlsanity.com/paper/1802.05957/) (2018)
- [Curiosity-driven Exploration by Self-supervised Prediction](https://mlsanity.com/paper/1705.05363/) (2017)
- [Exploration by Random Network Distillation](https://mlsanity.com/paper/1810.12894/) (2018)
- [DeepMind Control Suite](https://mlsanity.com/paper/1801.00690/) (2018)
- [Deep Reinforcement Learning at the Edge of the Statistical Precipice](https://mlsanity.com/paper/2108.13264/) (2021)
- [Diversity is All You Need: Learning Skills without a Reward Function](https://mlsanity.com/paper/1802.06070/) (2018)
- [Unifying Count-Based Exploration and Intrinsic Motivation](https://mlsanity.com/paper/1606.01868/) (2016)
- [Mastering Visual Continuous Control: Improved Data-Augmented Reinforcement Learning](https://mlsanity.com/paper/2107.09645/) (2021)
- [Dynamics-Aware Unsupervised Discovery of Skills](https://mlsanity.com/paper/1907.01657/) (2019)
- [Variational Intrinsic Control](https://mlsanity.com/paper/1611.07507/) (2016)
- [Self-Supervised Exploration via Disagreement](https://mlsanity.com/paper/1906.04161/) (2019)
- [Efficient Exploration via State Marginal Matching](https://mlsanity.com/paper/1906.05274/) (2019)
- [ViZDoom: A Doom-based AI Research Platform for Visual Reinforcement Learning](https://mlsanity.com/paper/1605.02097/) (2016)
- [Reinforcement Learning with Prototypical Representations](https://mlsanity.com/paper/2102.11271/) (2021)
- [Behavior From the Void: Unsupervised Active Pre-Training](https://mlsanity.com/paper/2103.04551/) (2021)
- [Variational Information Maximisation for Intrinsically Motivated Reinforcement Learning](https://mlsanity.com/paper/1509.08731/) (2015)
- [URLB: Unsupervised Reinforcement Learning Benchmark](https://mlsanity.com/paper/2110.15191/) (2021)
- [Fast Task Inference with Variational Intrinsic Successor Features](https://mlsanity.com/paper/1906.05030/) (2019)
- [Variational Option Discovery Algorithms](https://mlsanity.com/paper/1807.10299/) (2018)
- [Explore, Discover and Learn: Unsupervised Discovery of State-Covering Skills](https://mlsanity.com/paper/2002.03647/) (2020)
- [APS: Active Pretraining with Successor Features](https://mlsanity.com/paper/2108.13956/) (2021)
- [A survey on intrinsic motivation in reinforcement learning](https://mlsanity.com/paper/1908.06976/) (2019)
- [CIC: Contrastive Intrinsic Control for Unsupervised Skill Discovery](https://mlsanity.com/paper/2202.00161/) (2022)
- [Unsupervised Meta-Learning for Reinforcement Learning](https://mlsanity.com/paper/1806.04640/) (2018)

## Built on by (most cited first, 9 of 9)

- [Absolute Zero: Reinforced Self-play Reasoning with Zero Data](https://mlsanity.com/paper/2505.03335/) (2025)
- [METRA: Scalable Unsupervised RL with Metric-Aware Abstraction](https://mlsanity.com/paper/2310.08887/) (2023)
- [Controllability-Aware Unsupervised Skill Discovery](https://mlsanity.com/paper/2302.05103/) (2023)
- [PEAC: Unsupervised Pre-training for Cross-Embodiment Reinforcement Learning](https://mlsanity.com/paper/2405.14073/) (2024)
- [Constrained Intrinsic Motivation for Reinforcement Learning](https://mlsanity.com/paper/2407.09247/) (2024)
- [Surprise-Adaptive Intrinsic Motivation for Unsupervised Reinforcement Learning](https://mlsanity.com/paper/2405.17243/) (2024)
- [Scaffolded Language Models with Language Supervision for Mixed-Autonomy: A Survey](https://mlsanity.com/paper/2410.16392/) (2024)
- [Learning Generalizable Skill Policy with Data-Efficient Unsupervised RL](https://mlsanity.com/paper/2607.00392/) (2026)
- [Augmenting Unsupervised Reinforcement Learning with Self-Reference](https://mlsanity.com/paper/2311.09692/) (2023)

## Abstract (arXiv)

Unsupervised reinforcement learning aims at learning a generalist policy in a reward-free manner for fast adaptation to downstream tasks. Most of the existing methods propose to provide an intrinsic reward based on surprise. Maximizing or minimizing surprise drives the agent to either explore or gain control over its environment. However, both strategies rely on a strong assumption: the entropy of the environment's dynamics is either high or low. This assumption may not always hold in real-world scenarios, where the entropy of the environment's dynamics may be unknown. Hence, choosing between the two objectives is a dilemma. We propose a novel yet simple mixture of policies to address this concern, allowing us to optimize an objective that simultaneously maximizes and minimizes the surprise. Concretely, we train one mixture component whose objective is to maximize the surprise and another whose objective is to minimize the surprise. Hence, our method does not make assumptions about the entropy of the environment's dynamics. We call our method a Mixture Of SurpriseS (MOSS) for unsupervised reinforcement learning. Experimental results show that our simple method achieves state-of-the-art performance on the URLB benchmark, outperforming previous pure surprise maximization-based objectives. Our code is available at: https://github.com/LeapLabTHU/MOSS.