Research map / Learning theory and algorithms / Kernels and regression theory
Random Feature-based Online Multi-kernel Learning in Environments with Unknown Dynamics
Yanning Shen, Tianyi Chen, Georgios B. Giannakis
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (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 |
|---|---|---|---|
| add multiple Raker instances with variable learning rates (AdaRaker over Raker) | Raker · Energy regression | MSE (x10^-3): 19.3 -> 13.8 | Improved |
| add multiple Raker instances with variable learning rates (AdaRaker over Raker) | Raker · Air quality regression | MSE (x10^-3): 2.0 -> 1.3 | Improved |
| add random feature approximation (Raker instead of OMKL) | OMKL · Energy regression | MSE (x10^-3): 28.1 -> 19.3 | Improved |
| add random feature approximation (Raker instead of OMKL-B budget) | OMKL-B · Energy regression | MSE (x10^-3): 73.3 -> 19.3 | Improved |
| tune stepsize for Raker instead of 1/sqrt(T) (ablation) | Raker · Twitter regression | MSE (x10^-3): 3.0 -> 2.9 | Improved |
| tune stepsize for OMKL-B instead of 1/sqrt(T) (ablation) | OMKL-B · Air quality regression | MSE (x10^-3): 35.9 -> 7.1 | Improved |
| use diminishing stepsize 0.5/sqrt(t) instead of constant 1/sqrt(T) (ablation) | Raker · Air quality regression | MSE (x10^-3): 2.0 -> 29.1 | Hurt |
| increase random features D from 10 to 50 (ablation) | Raker · Air pollution regression | MSE (x10^-3): 10.1 -> 2.0 | Improved |
| increase random features D from 50 to 100 (ablation) | AdaRaker · Twitter (Large) regression | MSE (x10^-3): 2.7 -> 2.7 | No clear effect |
| increase random features D from 10 to 50 (ablation) | AdaRaker · Tom's hardware regression | MSE (x10^-3): 7.0 -> 1.9 | Improved |
| add adaptive kernel combination over simple averaging (OMKL instead of AvgMKL) | AvgMKL · Air quality regression | MSE (x10^-3): 11.9 -> 6.4 | Improved |
| add multiple Raker instances (AdaRaker vs Raker), default stepsize | Raker · Movement classification | classification error (%): 9.74 -> 1.10 | Improved |
Closest papers by meaning
- Online Multi-Kernel Learning with Graph-Structured Feedback (ICML 2020 · similarity 0.89)
- Graph-Aided Online Multi-Kernel Learning (JMLR 2023 · similarity 0.87)
- From Kernels to Features: A Multi-Scale Adaptive Theory of Feature Learning (ICML 2025 · similarity 0.85)
- Adaptive kernel predictors from feature-learning infinite limits of neural networks (ICML 2025 · similarity 0.85)
- Multi-View Randomized Kernel Classification via Nonconvex Optimization (AAAI 2024 · similarity 0.85)
- Joker: Joint Optimization Framework for Lightweight Kernel Machines (ICML 2025 · 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.
