Research map / Learning theory and algorithms / Kernels and regression theory
Scalable and Stable Estimation of Amari $\alpha$-Divergence using Random Fourier Features
Jiaolong Wang, Fode Zhang, Lingrui Wang
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 |
|---|---|---|---|
| inject fixed Spatial RFF (sigma=5) into the CNN discriminator (ablation) | CNN GAN discriminator · CelebA | FID: 240.47 -> 27.23 | Improved |
| enlarge Spatial-RFF bandwidth from sigma=1 to sigma=5 (ablation) | CNN GAN discriminator · CelebA | FID: 63.56 -> 30.07 -> 27.23 (sigma 1, 2, 5) | Improved |
| enlarge Spatial-RFF bandwidth from sigma=1 to sigma=5 (ablation) | CNN GAN discriminator · CelebA | SWD: 98.30 (sigma=1) -> 127.23 (sigma=2) -> 120.78 (sigma=5) | Mixed |
| inject Spatial RFF to raise high-frequency capture (ablation) | CNN GAN discriminator · CelebA | HF energy (x10^-2): 24.89 -> 32.57 | Improved |
| use RFF linear critic instead of unconstrained MLP critics | Amari alpha-divergence estimator · synthetic truncated Gaussian vs uniform, d in {2,5,10} | relative RMSE | Improved |
| use RFF estimator instead of RKHS-regularized neural critic | Amari alpha-divergence estimator · synthetic divergence estimation | mean relative error | Improved |
Closest papers by meaning
- Globally Convergent Variational Inference (NeurIPS 2024 · similarity 0.87)
- Asymptotics of Alpha-Divergence Variational Inference Algorithms with Exponential Families (NeurIPS 2024 · similarity 0.86)
- Drifting Fields are not Conservative (NeurIPS 2026 · similarity 0.85)
- Non-asymptotic Analysis of Biased Adaptive Stochastic Approximation (NeurIPS 2024 · similarity 0.84)
- Implicit Bias of Mirror Flow for Shallow Neural Networks in Univariate Regression (ICLR 2025 · similarity 0.84)
- Convergence Analysis of Newton's Method for Neural Networks in the Overparameterized Limit (NeurIPS 2026 · 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.
