OFA²: A Multi-Objective Perspective for the Once-for-All Neural Architecture Search [arXiv]

@misc{ito2023ofa2,
      title={OFA$^2$: A Multi-Objective Perspective for the Once-for-All Neural Architecture Search}, 
      author={Rafael C. Ito and Fernando J. Von Zuben},
      year={2023},
      eprint={2303.13683},
      archivePrefix={arXiv},
      primaryClass={cs.NE}
}

Jupyter Notebooks

• ofa2.ipynb
  This notebook runs the OFA² multi-objective search for 10,000 generations for three different EMOAs (Evolutionary Multi-objective Optimization Algorithms): NSGA-II, SMS-EMOA and SPEA2. Each algorithm runs 3 times considering 3 different random seeds. The Colab version is a simplified version of this notebook and runs only for 1,000 generations for each algorithm.
• ofa2-debug.ipynb
  This notebook runs the same OFA² search as the previous notebook, but it expands the evolutionary algorithm section and provide more details related to the code implementation. The framework used to implement the multi-objective optimization is the pymoo.