Paper
Post / Article
Automating automation: framework for developing deep learning models
Design by evolution: 딥네트워크 아키텍처와 하이퍼파라미터 등을 설정하는데에 Evolutionary Computing, Genetic Algorithm 등이 다시 부각이 되고 있죠? PyTorch를 이용해서 이 내용들을 설명하면서 간단한 예제
Accelerating Deep Learning Research with the Tensor2Tensor Library: open-source system for training deep learning models in TensorFlow.
Automate your Machine Learning in Python – TPOT and Genetic Algorithms
AutoML을 이용한 Architecture Search 소개 및 NASNet 논문 리뷰: 수아랩, 2018.09.28
Material (ppt, pdf)
- automl & autodraw: ppt
Implementation
Evolution Strategies as a Scalable Alternative to Reinforcement Learning: Keras구현코드, Evostra: Evolution Strategy for Python: Evolutio Strategy (ES) is an optimization technique based on ideas of adaptation and evolution.
OptML : 깃허브
Optuna: A Next-generation Hyperparameter Optimization Framework : 2019