Research map / Reinforcement learning / Policy gradient and actor-critic
Fuzzy Tiling Activations: A Simple Approach to Learning Sparse Representations Online
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| Use FTA instead of ReLU | Online neural-network supervised learning · Piecewise Random Walk Problem | mean squared error | Helped |
| Use FTA instead of ReLU | Online neural-network supervised learning · Piecewise Random Walk Problem with high correlation | mean squared error | Helped |
| Use FTA in DQN | DQN · MountainCar, CartPole, Acrobot, and LunarLander | episodic return | Mixed |
| Omit the target network while using FTA (ablation) | DQN-FTA · Discrete control environments | episodic return | Helped |
| Omit the target network without using FTA (ablation) | DQN · Discrete control environments | episodic return | Hurt |
| Increase the hidden-layer size to match the FTA feature dimension (ablation) | DQN · Discrete control environments | episodic return | Hurt |
| Use tiling bound [-1,1] instead of [-20,20] (ablation) | DQN-FTA · LunarLander | episodic return | Helped |
| Vary the number of tiles k, increasing it (ablation) | DQN-FTA · LunarLander | episodic return | Helped |
| Increase DQN hidden-layer size as k increases (ablation) | DQN · LunarLander | episodic return | No clear effect |
| Use an extremely small or large tiling bound (ablation) | DQN-FTA · LunarLander | episodic return | Hurt |
| Use FTA in the critic and omit its target network | DDPG · Mujoco continuous control | episodic return | Mixed |
| Replace FTA with RBF features | DQN · Discrete control environments | episodic return | Hurt |
Closest papers by meaning
- Random Sparse Lifts: Construction, Analysis and Convergence of finite sparse networks (ICLR 2024 · similarity 0.85)
- Exploring the Benefit of Activation Sparsity in Pre-training (ICML 2024 · similarity 0.84)
- Locality Sensitive Sparse Encoding for Learning World Models Online (ICLR 2024 · similarity 0.84)
- Reinforcement Learning with Action Chunking (NeurIPS 2025 · similarity 0.84)
- Beyond Matryoshka: Revisiting Sparse Coding for Adaptive Representation (ICML 2025 · similarity 0.84)
- Straight-Through Meets Sparse Recovery: the Support Exploration Algorithm (ICML 2024 · similarity 0.84)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.