https://github.com/kalifou/fitness_landscape_analysis_nas

Fitness Landscape Analysis for NAS
https://github.com/kalifou/fitness_landscape_analysis_nas

automl evolutionary-computation fitness-landscape-analysis neural-architecture-search searching-algorithms

Last synced: 3 days ago
JSON representation

Fitness Landscape Analysis for NAS

• Host: GitHub
• URL: https://github.com/kalifou/fitness_landscape_analysis_nas
• Owner: kalifou
• Created: 2021-11-23T19:02:15.000Z (about 3 years ago)
• Default Branch: main
• Last Pushed: 2021-11-29T10:31:30.000Z (about 3 years ago)
• Last Synced: 2025-01-13T03:59:01.301Z (26 days ago)
• Topics: automl, evolutionary-computation, fitness-landscape-analysis, neural-architecture-search, searching-algorithms
• Language: Jupyter Notebook
• Homepage:
• Size: 83.2 MB
• Stars: 2
• Watchers: 1
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md

Awesome Lists containing this project

README

        
# Fitness Landscape Analysis for Neural Architecture Search:

## Step 1: *Install Nasbench-101 submodule*

## Generate descriptive data of NASbench-101:

  * **cd nasbench/**

  * **python nasbench/nasalg/landscape.py  > experiments/full-landscape.json**

## Analyze the Fitness Landscape of NASBench-101    

  * **Statistics and FDC:** nasbench/experiments/fitness-landscape.Rmd (R-sudio)

  * **Local optima estimation:** card_optima/local_search_for_card_optima.py

  * **Local optima analysis:** notebooks/Card_Optima_-_CIFAR_-_10.ipynb (notebook)

  * **Random Walk Analysis:** notebooks/Random_Walks_Analysis_-_CIFAR-10.ipynb'(notebook)

  * **Persistence:** notebooks/Persistence_-_CIFAR-10.ipynb'(notebook)

            

##Project using this code:

**Fitness Landscape Footprint: A Framework to Compare Neural Architecture Search Problems**,

*arXiv:2111.01584, 

 Kalifou René Traoré, Andrés Camero, Xiao Xiang Zhu.*