Research map / Architectures and efficiency / RNNs and normalization
Large-Scale Evolution of Image Classifiers
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 |
|---|---|---|---|
| Disable selection and reproduce and kill random individuals (ablation) | Evolution · CIFAR-10 | test accuracy: 94.6 -> 87.3 | Hurt |
| Disable weight inheritance so individuals train from random weights (ablation) | Evolution · CIFAR-10 | test accuracy: 94.6 -> 92.2 | Hurt |
| Increase population size (ablation) | Evolution · CIFAR-10 | test accuracy | Helped |
| Increase training steps per individual (ablation) | Evolution · CIFAR-10 | test accuracy | Helped |
| Raise the mutation count from one to five during a phase of evolution (ablation) | Evolution · CIFAR-10 | population fitness: 8 out of 10 | Helped |
| Reset all population weights three successive times after a plateau (ablation) | Evolution · CIFAR-10 | population fitness: p < 0.001 | Helped |
| Add recombination of mutation probabilities, trained weights, or parent structures (ablation) | Evolution · CIFAR-10 | test accuracy | No clear effect |
| Apply the CIFAR-10 evolutionary algorithm without changes to CIFAR-100 | Evolution · CIFAR-100 | test accuracy | Helped |
| Ensemble the validation-top 2 models from each population | Evolution · CIFAR-10 | test accuracy: 94.6% -> 95.6% | Helped |
Closest papers by meaning
- Regularized Evolution for Image Classifier Architecture Search (AAAI 2019 · similarity 0.91)
- Hierarchical Representations for Efficient Architecture Search (ICLR 2018 · similarity 0.88)
- Runtime Analysis of Evolutionary NAS for Multiclass Classification (ICML 2025 · similarity 0.86)
- EVOS: Efficient Implicit Neural Training via EVOlutionary Selector (CVPR 2025 · similarity 0.85)
- From Kernels to Features: A Multi-Scale Adaptive Theory of Feature Learning (ICML 2025 · similarity 0.84)
- EvolveDirector: Approaching Advanced Text-to-Image Generation with Large Vision-Language Models (NeurIPS 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.