Research map / Learning theory and algorithms / Prediction and decision losses
Feature Selection using Stochastic Gates
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 Gaussian-based STG gates instead of logistic-based HC gates (ablation) | neural-network feature selection · MNIST (3s and 8s) | test accuracy: 91.7% -> 92.4% | Helped |
| Use Gaussian-based STG gates instead of logistic-based HC gates (ablation) | neural-network feature selection · MNIST (3s and 8s) | variance in number of selected features: 3.8 -> 1 | Helped |
| Add stochastic noise to deterministic non-convex feature selection (ablation) | linear regression feature selection · synthetic linear regression | probability of recovering informative features | Helped |
| Use non-convex regularization instead of convex LASSO regularization (ablation) | linear regression feature selection · synthetic linear regression | probability of recovering informative features | Helped |
| Use stochastic gates instead of deterministic non-convex gates (ablation) | linear regression feature selection · synthetic linear regression | probability of recovering informative features | Helped |
| Use Gaussian-based STG gates instead of HC gates (ablation) | linear regression feature selection · correlated synthetic design | probability of recovering the correct support | Helped |
| Add stochastic gates with an expected-count penalty to neural-network input features | Cox-STG · METABRIC | C-index: 0.612 (0.009) -> 0.633 (0.005) | Helped |
| Add stochastic gates with an expected-count penalty to neural-network input features | Cox-STG · METABRIC | number of selected features: 221 (All) -> 2 (0) | Helped |
Closest papers by meaning
- SAND: One-Shot Feature Selection with Additive Noise Distortion (ICML 2025 · similarity 0.88)
- Contextual Feature Selection with Conditional Stochastic Gates (ICML 2024 · similarity 0.86)
- The Power of Random Features and the Limits of Distribution-Free Gradient Descent (ICML 2025 · similarity 0.85)
- BGDB: Bernoulli-Gaussian Decision Block with Improved Denoising Diffusion Probabilistic Models (AAAI 2025 · similarity 0.84)
- A Study of Bayesian Neural Network Surrogates for Bayesian Optimization (ICLR 2024 · similarity 0.84)
- Critical feature learning in deep neural networks (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.