Research map / Learning theory and algorithms / Combinatorial and search optimization
GENO -- GENeric Optimization for Classical Machine Learning
Soeren Laue, Matthias Mitterreiter, Joachim Giesen
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 |
|---|---|---|---|
| Use GENO instead of SAGA for the ℓ1-regularized logistic-regression regularization path | GENO · Iris | running time (seconds): 4.3 seconds -> 0.5 seconds | Improved |
| Use GENO instead of CVXPY for the ℓ1-regularized logistic-regression path | GENO · Iris | running time (seconds): 13.5 seconds -> 0.5 seconds | Improved |
| Use GENO instead of LIBLINEAR for the original ℓ1-regularized logistic-regression formulation | GENO · Iris | accuracy of regularization path | Improved |
| Use GENO rather than ECOS for small ℓ2-regularized logistic regression | GENO · Logistic regression (ℓ2), heart and a5a | running time (seconds): 0.005 -> 1.999; 0.062 -> 57.467 | Improved |
| Use GENO rather than LIBSVM for the Gaussian-kernel SVM | GENO · Gaussian-kernel SVM, w8a | running time (seconds): 78.6 -> 25.7 | Improved |
| Use GENO rather than LIBSVM for the Gaussian-kernel SVM | GENO · Gaussian-kernel SVM, ionosphere and a5a | running time (seconds): 0.005 -> 0.009; 1.0 -> 1.6 | Hurt |
| Use GENO rather than glmnet for elastic net regression | GENO · Elastic net, m=1000, n=5000 | running time (seconds): 0.62 -> 1.74 | Hurt |
| Use GENO rather than glmnet for elastic net regression | GENO · Elastic net, m=5000, n=10000 | running time (seconds): 4.75 -> 4.11 | Improved |
| Use GENO rather than ECOS for non-negative least squares | GENO · Non-negative least squares, m=2000, n=6000 | running time (seconds): 689.7 -> 4.8 | Improved |
| Use GENO rather than SCS for non-negative least squares | GENO · Non-negative least squares, m=2000, n=6000 | function value: 0.07116707 -> 0.01306327 | Improved |
| Use GENO rather than SymANLS and SymHALS for symmetric NMF | GENO · Symmetric non-negative matrix factorization | convergence speed | Improved |
| Use GENO rather than S-AdaNCG and AdaNCD-SCSG for non-linear least squares | GENO · Non-linear least squares, w1a and a1a | convergence speed | Improved |
Closest papers by meaning
- Optimization for Classical Machine Learning Problems on the GPU (AAAI 2022 · similarity 0.87)
- Generation as Search Operator for Test-Time Scaling of Diffusion-based Combinatorial Optimization (NeurIPS 2025 · similarity 0.86)
- Gradient descent with generalized Newton’s method (ICLR 2025 · similarity 0.85)
- Towards Constituting Mathematical Structures for Learning to Optimize (ICML 2023 · similarity 0.85)
- Symbolic Learning to Optimize: Towards Interpretability and Scalability (ICLR 2022 · similarity 0.84)
- LLMOPT: Learning to Define and Solve General Optimization Problems from Scratch (ICLR 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.
