mlsanityData to 7 Oct 2026

Transfer Dynamics in Emergent Evolutionary Curricula

arXiv 2203.10941 ↗First posted 3 Mar 2022cs.AI cs.LG cs.NEcited by 6

Not on the map. Why?

Abstract (arXiv)

PINSKY is a system for open-ended learning through neuroevolution in game-based domains. It builds on the Paired Open-Ended Trailblazer (POET) system, which originally explored learning and environment generation for bipedal walkers, and adapts it to games in the General Video Game AI (GVGAI) system. Previous work showed that by co-evolving levels and neural network policies, levels could be found for which successful policies could not be created via optimization alone. Studied in the realm of Artificial Life as a potentially open-ended alternative to gradient-based fitness, minimal criteria (MC)-based selection helps foster diversity in evolutionary populations. The main question addressed by this paper is how the open-ended learning actually works, focusing in particular on the role of transfer of policies from one evolutionary branch ("species") to another. We analyze the dynamics of the system through creating phylogenetic trees, analyzing evolutionary trajectories of policies, and temporally breaking down transfers according to species type. Furthermore, we analyze the impact of the minimal criterion on generated level diversity and inter-species transfer. The most insightful finding is that inter-species transfer, while rare, is crucial to the system's success.

Key facts. Cited by 6 papers in the map (data through 7 Oct 2026). Builds on 28 of its 60 references that are in the map. First posted on arXiv 3 Mar 2022.

Builds on

28 of 60 references.

References · 28

  1. 2022Generative Adversarial Networkscited by 5,603
  2. 2017Model-Agnostic Meta-Learning for Fast Adaptation of Deep Networkscited by 4,861
  3. 2018Soft Actor-Critic: Off-Policy Maximum Entropy Deep Reinforcement Learning with a Stochastic Actorcited by 4,635
  4. 2017Curiosity-driven Exploration by Self-supervised Predictioncited by 1,173
  5. 2019Meta-World: A Benchmark and Evaluation for Multi-Task and Meta Reinforcement Learningcited by 1,168
  6. 2019Dota 2 with Large Scale Deep Reinforcement Learningcited by 1,084
  7. 2014Deep Visual-Semantic Alignments for Generating Image Descriptionscited by 966
  8. 2017Evolution Strategies as a Scalable Alternative to Reinforcement Learningcited by 555
  9. 2015Illuminating search spaces by mapping elitescited by 406
  10. 2020Curriculum Learning for Reinforcement Learning Domains: A Framework and Surveycited by 350
  11. 2018Learning to Adapt in Dynamic, Real-World Environments Through Meta-Reinforcement Learningcited by 270
  12. 2017StarCraft II: A New Challenge for Reinforcement Learningcited by 268
16 more
  1. 2017A Unified Game-Theoretic Approach to Multiagent Reinforcement Learningcited by 265
  2. 2018Characterizing and Avoiding Negative Transfercited by 226
  3. 2014Robots that can adapt like animalscited by 203
  4. 2020Enhanced POET: Open-Ended Reinforcement Learning through Unbounded Invention of Learning Challenges and their Solutionscited by 103
  5. 2019Generalized Inner Loop Meta-Learningcited by 92
  6. 2018Illuminating Generalization in Deep Reinforcement Learning through Procedural Level Generationcited by 80
  7. 2016Loss is its own Reward: Self-Supervision for Reinforcement Learningcited by 72
  8. 2019AlphaStar: An Evolutionary Computation Perspectivecited by 50
  9. 2018General Video Game AI: a Multi-Track Framework for Evaluating Agents, Games and Content Generation Algorithmscited by 45
  10. 2020Rapidly Adaptable Legged Robots via Evolutionary Meta-Learningcited by 42
  11. 2018StarCraft Micromanagement with Reinforcement Learning and Curriculum Transfer Learningcited by 25
  12. 2020Rotation, Translation, and Cropping for Zero-Shot Generalizationcited by 23
  13. 2020Co-generation of game levels and game-playing agentscited by 14
  14. 2019Evolvability ES: Scalable and Direct Optimization of Evolvabilitycited by 7
  15. 2020Action Guidance: Getting the Best of Sparse Rewards and Shaped Rewards for Real-time Strategy Gamescited by 3
  16. 2018On the potential for open-endedness in neural networkscited by 1

Built on by

Cited by year

Citing papers per year: 2022: 2; 2023: 3; 2024: 1.

6 papers.

  1. 2024Automated Design of Agentic Systemscited by 360
  2. 2023OMNI: Open-endedness via Models of human Notions of Interestingnesscited by 66
  3. 2022Deep Surrogate Assisted Generation of Environmentscited by 40
  4. 2023Stabilizing Unsupervised Environment Design with a Learned Adversarycited by 17
  5. 2023Surrogate Assisted Generation of Human-Robot Interaction Scenarioscited by 8
  6. 2022Augmentative Topology Agents For Open-Ended Learningcited by 3