Research map / Learning theory and algorithms / Kernels and regression theory
On the Fine-Grained Complexity of Empirical Risk Minimization: Kernel Methods and Neural Networks
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 a Gaussian-kernel hard-margin SVM and require high-accuracy optimization | Kernel SVM · Kernel SVM | running time: almost quadratic time | Hurt |
| Include a bias term in the hard-margin SVM | Kernel SVM · Kernel SVM with bias term | running time: almost quadratic time | Hurt |
| Use soft-margin SVM with bias and sufficiently small regularization | Kernel SVM · Soft-margin kernel SVM | running time: almost quadratic time | Hurt |
| Solve kernel ridge regression for the stated special case | Kernel ridge regression · Kernel ridge regression | running time: almost quadratic time | Hurt |
| Compute the centered kernel matrix eigenvalue sum | Kernel PCA · Kernel PCA | running time: almost quadratic time | Hurt |
| Optimize the final layer of a depth-2 network to high accuracy | Top-layer neural network ERM · Neural network ERM | running time: almost quadratic time | Hurt |
| Compute the empirical-risk gradient for a batch using ReLU | Neural network gradient computation · Neural network gradient, ReLU | running time: O((nm)^{1-o(1)}) | Hurt |
| Compute the empirical-risk gradient for a batch using sigmoid | Neural network gradient computation · Neural network gradient, sigmoid | running time: (nm)^{1-o(1)} | Hurt |
Closest papers by meaning
- On the Efficiency of ERM in Feature Learning (NeurIPS 2024 · similarity 0.90)
- Stochastic Primal-Dual Method for Empirical Risk Minimization with O(1) Per-Iteration Complexity (NeurIPS 2018 · similarity 0.90)
- On the Asymptotic Distribution of the Minimum Empirical Risk (ICML 2024 · similarity 0.89)
- Differentially Private Empirical Risk Minimization with Smooth Non-Convex Loss Functions: A Non-Stationary View (AAAI 2019 · similarity 0.89)
- Differentially Private Empirical Risk Minimization Revisited: Faster and More General (NeurIPS 2017 · similarity 0.89)
- Universal Rates of Empirical Risk Minimization (NeurIPS 2024 · similarity 0.87)
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.