mlsanityData to 7 Oct 2026

Adaptive Rational Activations to Boost Deep Reinforcement Learning

arXiv 2102.09407 ↗First posted 18 Feb 2021cs.LGcited by 23

Not on the map. Why?

Abstract (arXiv)

Latest insights from biology show that intelligence not only emerges from the connections between neurons but that individual neurons shoulder more computational responsibility than previously anticipated. This perspective should be critical in the context of constantly changing distinct reinforcement learning environments, yet current approaches still primarily employ static activation functions. In this work, we motivate why rationals are suitable for adaptable activation functions and why their inclusion into neural networks is crucial. Inspired by recurrence in residual networks, we derive a condition under which rational units are closed under residual connections and formulate a naturally regularised version: the recurrent-rational. We demonstrate that equipping popular algorithms with (recurrent-)rational activations leads to consistent improvements on Atari games, especially turning simple DQN into a solid approach, competitive to DDQN and Rainbow.

Key facts. Cited by 23 papers in the map (data through 7 Oct 2026). Builds on 35 of its 73 references that are in the map. First posted on arXiv 18 Feb 2021.

Builds on

35 of 73 references.

References · 35

  1. 2015Deep Residual Learning for Image Recognitioncited by 51,809
  2. 2020Language Models are Few-Shot Learnerscited by 32,199
  3. 2017Proximal Policy Optimization Algorithmscited by 14,441
  4. 2015You Only Look Once: Unified, Real-Time Object Detectioncited by 4,829
  5. 2018Soft Actor-Critic: Off-Policy Maximum Entropy Deep Reinforcement Learning with a Stochastic Actorcited by 4,635
  6. 2016Asynchronous Methods for Deep Reinforcement Learningcited by 2,218
  7. 2015Deep Reinforcement Learning with Double Q-learningcited by 1,704
  8. 2016OpenAI Gymcited by 1,684
  9. 2016Neural Architecture Search with Reinforcement Learningcited by 1,489
  10. 2020Mastering Atari with Discrete World Modelscited by 990
  11. 2017Sigmoid-Weighted Linear Units for Neural Network Function Approximation in Reinforcement Learningcited by 935
  12. 2018IMPALA: Scalable Distributed Deep-RL with Importance Weighted Actor-Learner Architecturescited by 866
23 more
  1. 2017Rainbow: Combining Improvements in Deep Reinforcement Learningcited by 774
  2. 2017Deep Reinforcement Learning that Matterscited by 727
  3. 2020Image Augmentation Is All You Need: Regularizing Deep Reinforcement Learning from Pixelscited by 573
  4. 2017Progressive Neural Architecture Searchcited by 483
  5. 2018Quantifying Generalization in Reinforcement Learningcited by 345
  6. 2018SNAS: Stochastic Neural Architecture Searchcited by 288
  7. 2020Never Give Up: Learning Directed Exploration Strategiescited by 210
  8. 2020SPACE: Unsupervised Object-Oriented Scene Representation via Spatial Attention and Decompositioncited by 190
  9. 2018Generalization and Regularization in DQNcited by 117
  10. 2016Understanding and Improving Convolutional Neural Networks via Concatenated Rectified Linear Unitscited by 86
  11. 2020Revisiting Rainbow: Promoting more Insightful and Inclusive Deep Reinforcement Learning Researchcited by 78
  12. 2020Rational neural networkscited by 75
  13. 2019Pad'e Activation Units: End-to-end Learning of Flexible Activation Functions in Deep Networkscited by 65
  14. 2020Counterfactual Credit Assignment in Model-Free Reinforcement Learningcited by 60
  15. 2016Highway and Residual Networks learn Unrolled Iterative Estimationcited by 52
  16. 2020MushroomRL: Simplifying Reinforcement Learning Researchcited by 50
  17. 2016Bridging the Gaps Between Residual Learning, Recurrent Neural Networks and Visual Cortexcited by 42
  18. 2017Neural networks and rational functionscited by 40
  19. 2019Learning Activation Functions: A new paradigm for understanding Neural Networkscited by 33
  20. 2016Parametric Exponential Linear Unit for Deep Convolutional Neural Networkscited by 32
  21. 2018Learning Combinations of Activation Functionscited by 20
  22. 2020Non-linear Neurons with Human-like Apical Dendrite Activationscited by 5
  23. 2015Small-footprint Deep Neural Networks with Highway Connections for Speech Recognitioncited by 1

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2022: 1; 2023: 4; 2024: 9; 2025: 5; 2026: 4.

23 papers.

Sort:

By area:

  1. 2023Convolutional Neural Operators for robust and accurate learning of PDEscited by 207
  2. 2023Understanding plasticity in neural networkscited by 170
  3. 2023Interpretable and Explainable Logical Policies via Neurally Guided Symbolic Abstractioncited by 45
  4. 2023OCAtari: Object-Centric Atari 2600 Reinforcement Learning Environmentscited by 30
  5. 2024Streaming Deep Reinforcement Learning Finally Workscited by 24
  6. 2024Interpretable Concept Bottlenecks to Align Reinforcement Learning Agentscited by 23
  7. 2022Boosting Object Representation Learning via Motion and Object Continuitycited by 17
  8. 2024BlendRL: A Framework for Merging Symbolic and Neural Policy Learningcited by 14
  9. 2024HackAtari: Atari Learning Environments for Robust and Continual Reinforcement Learningcited by 11
  10. 2024PEDRO: Parameter-Efficient Fine-tuning with Prompt DEpenDent Representation MOdificationcited by 8
  11. 2025Activation by Interval-wise Dropout: A Simple Way to Prevent Neural Networks from Plasticity Losscited by 6
  12. 2025Activation Function Design Sustains Plasticity in Continual Learningcited by 6
11 more
  1. 2025Spectral Collapse Drives Loss of Plasticity in Deep Continual Learningcited by 5
  2. 2024Mitigating the Stability-Plasticity Dilemma in Adaptive Train Scheduling with Curriculum-Driven Continual DQN Expansioncited by 2
  3. 2024Graph Neural Networks Need Cluster-Normalize-Activate Modulescited by 2
  4. 2025Balancing Expressivity and Robustness: Constrained Rational Activations for Reinforcement Learningcited by 2
  5. 2024Hadamard Representation: Scaffolding Performance Across Model-free RLcited by 1
  6. 2026Rational Neural Networks have Expressivity Advantagescited by 1
  7. 2026Beyond Single-Model Optimization: Preserving Plasticity in Continual Reinforcement Learningcited by 1
  8. 2024Zorro: A Flexible and Differentiable Parametric Family of Activation Functions That Extends ReLU and GELUcited by 0
  9. 2025A Simple Baseline for Stable and Plastic Neural Networkscited by 0
  10. 2026Rational Sparse Autoencodercited by 0
  11. 2026Sustaining Plasticity via Learnable Wavelet Activations in Continual Learningcited by 0