AutoKeras

AutoKeras is an open-source AutoML library built on top of Keras and TensorFlow that aims to simplify the development of deep learning models. By combining neural architecture search with hyperparameter optimization, it automates model design and training, enabling users to achieve competitive performance with minimal manual tuning.

AutoKeras provides task-specific components such as ImageClassifier, ImageRegressor, TextClassifier, TextRegressor, StructuredDataClassifier, and StructuredDataRegressor. It also handles preprocessing steps automatically, including image resizing, text tokenization, normalization, and handling missing values for tabular data.

The framework uses a neural architecture search engine to explore a space of model architectures and hyperparameters, selecting promising candidates and training them under user-specified constraints.

Input data can be provided as NumPy arrays or pandas DataFrames, and the resulting model can be exported as a standard Keras model for deployment.

AutoKeras runs on TensorFlow 2.x and supports CPU or GPU execution. It is released as an open-source project under a permissive license and is maintained by a community of contributors.