쉬엄쉬엄블로그

이 색깔은 주석이라 무시하셔도 됩니다. 1. Overview 1.1 Conventional DL Training Pipeline Data Engineering Data Cleansing, Preprocessing Feature Engineering Select ML Algorithm Select Backbone Model(ResNet, MobileNet, EfficientNet, ...) Set Hyperparameters Loss, Optimizer, Learning rate, Batchsize, ...) … 반복적인(Human in the Loop) Tuning 과정 좋은 configuration을 찾을 때까지 반복 1.2 Objectives of AutoML AutoML: "True" End-to-..