mlsanityData to 7 Oct 2026

KAN v.s. MLP for Offline Reinforcement Learning

arXiv 2409.09653 ↗First posted 15 Sep 2024cs.LG cs.AIcited by 3

On the map: Offline RL and imitation learning from data → Offline RL algorithms

Abstract (arXiv)

Kolmogorov-Arnold Networks (KAN) is an emerging neural network architecture in machine learning. It has greatly interested the research community about whether KAN can be a promising alternative of the commonly used Multi-Layer Perceptions (MLP). Experiments in various fields demonstrated that KAN-based machine learning can achieve comparable if not better performance than MLP-based methods, but with much smaller parameter scales and are more explainable. In this paper, we explore the incorporation of KAN into the actor and critic networks for offline reinforcement learning (RL). We evaluated the performance, parameter scales, and training efficiency of various KAN and MLP based conservative Q-learning (CQL) on the the classical D4RL benchmark for offline RL. Our study demonstrates that KAN can achieve performance close to the commonly used MLP with significantly fewer parameters. This provides us an option to choose the base networks according to the requirements of the offline RL tasks.

Key facts. Cited by 3 papers in the map (data through 7 Oct 2026). Placed in Offline RL algorithms, one of 13 directions in Offline RL and imitation learning from data. Builds on 26 of its 37 references that are in the map. First posted on arXiv 15 Sep 2024.

Builds on

26 of 37 references.

Closest ancestors · 4

Same direction, newest first

  1. 2022A Survey on Offline Reinforcement Learning: Taxonomy, Review, and Open Problems2 Mar
  2. 2020Conservative Q-Learning for Offline Reinforcement Learning8 Jun
  3. 2020Offline Reinforcement Learning: Tutorial, Review, and Perspectives on Open Problems4 May
  4. 2020D4RL: Datasets for Deep Data-Driven Reinforcement Learning15 Apr

Other references · 22

  1. 2018Soft Actor-Critic: Off-Policy Maximum Entropy Deep Reinforcement Learning with a Stochastic Actorcited by 4,635
  2. 2024U-KAN Makes Strong Backbone for Medical Image Segmentation and Generationcited by 176
  3. 2024KAN 2.0: Kolmogorov-Arnold Networks Meet Sciencecited by 116
  4. 2024Kolmogorov-Arnold Networks (KANs) for Time Series Analysiscited by 97
  5. 2024DeepOKAN: Deep Operator Network Based on Kolmogorov Arnold Networks for Mechanics Problemscited by 82
  6. 2024Convolutional Kolmogorov-Arnold Networkscited by 76
  7. 2024TKAN: Temporal Kolmogorov-Arnold Networkscited by 70
  8. 2024Kolmogorov-Arnold Networks for Time Series: Bridging Predictive Power and Interpretabilitycited by 69
  9. 2024fKAN: Fractional Kolmogorov-Arnold Networks with trainable Jacobi basis functionscited by 60
  10. 2024BSRBF-KAN: A combination of B-splines and Radial Basis Functions in Kolmogorov-Arnold Networkscited by 43
  11. 2024Demonstrating the Efficacy of Kolmogorov-Arnold Networks in Vision Taskscited by 39
  12. 2024KANQAS: Kolmogorov-Arnold Network for Quantum Architecture Searchcited by 36
10 more
  1. 2024ReLU-KAN: New Kolmogorov-Arnold Networks that Only Need Matrix Addition, Dot Multiplication, and ReLUcited by 32
  2. 2024GraphKAN: Enhancing Feature Extraction with Graph Kolmogorov Arnold Networkscited by 31
  3. 2024rKAN: Rational Kolmogorov-Arnold Networkscited by 31
  4. 2024A Benchmarking Study of Kolmogorov-Arnold Networks on Tabular Datacited by 25
  5. 2024Predictive Modeling of Flexible EHD Pumps using Kolmogorov-Arnold Networkscited by 23
  6. 2024Smooth Kolmogorov Arnold networks enabling structural knowledge representationcited by 22
  7. 2024Endowing Interpretability for Neural Cognitive Diagnosis by Efficient Kolmogorov-Arnold Networkscited by 12
  8. 2024Initial Investigation of Kolmogorov-Arnold Networks (KANs) as Feature Extractors for IMU Based Human Activity Recognitioncited by 12
  9. 2024On the Robustness of Kolmogorov-Arnold Networks: An Adversarial Perspectivecited by 11
  10. 2024A First Look at Kolmogorov-Arnold Networks in Surrogate-assisted Evolutionary Algorithmscited by 7

Built on by

Cited by year

Citing papers per year: 2024: 3.

3 papers.

  1. 2024Quantum Kolmogorov-Arnold networks by combining quantum signal processing circuitscited by 1
  2. 2024On the Convergence of (Stochastic) Gradient Descent for Kolmogorov--Arnold Networkscited by 0
  3. 2024Human-in-the-Loop Feature Selection Using Interpretable Kolmogorov-Arnold Network-based Double Deep Q-Networkcited by 0

Similar, not cited

Close in wording, no citation either way

  1. 2025KAN-Dreamer: Benchmarking Kolmogorov-Arnold Networks as Function Approximators in World Models
  2. 2020Overcoming Model Bias for Robust Offline Deep Reinforcement Learning
  3. 2025Pretraining a Shared Q-Network for Data-Efficient Offline Reinforcement Learning
  4. 2024MLP-KAN: Unifying Deep Representation and Function Learning
  5. 2021Uncertainty-Based Offline Reinforcement Learning with Diversified Q-Ensemble
  6. 2023ENOTO: Improving Offline-to-Online Reinforcement Learning with Q-Ensembles
  7. 2021A Minimalist Approach to Offline Reinforcement Learning
  8. 2025Offline RLAIF: Piloting VLM Feedback for RL via SFO