Amazon Web Services recently launched an open-source library that enables developers to implement deep learning models on image, text or tabular data using just a few lines of code.
AutoGluon is designed to be an easy-to-use and easy-to-extend AutoML toolkit, suitable for both machine learning beginners and experts. It enables prototyping deep learning models with a few lines; automatic hyperparameter tuning, model selection and data processing; and automatic utilization of SOTA deep learning models. The AutoGluon framework can also help researchers customise and improve their existing bespoke models and data pipelines.
Developers have traditionally trained deep learning models by manually defining the neural network and specifying the hyperparameters during the training process. Researchers aiming for SOTA performance must invest considerate time deciding on the most efficient parameter updates to minimize errors, selecting the number of layers and how they should be connected, determining how best to classify and format the data, and the list goes on.
AutoGluon aims to eliminate such cumbersome processes and deliver developers a truly hands-off-the-wheel experience. After importing the AutoGluon package, developers can simply specify a task of interest, load the appropriate dataset, and finally have AutoGluon quickly and automatically train many models under thousands of different hyperparameter configurations and then return the best model.
One of the applications of AutoGluon is object detection in images, achieved through target identification and localisation within a bounding box. The author of AutoGluon Tabular Nick Erickson and his colleagues used AutoGluon to train an object detector on a small toy dataset generated using the motorbike category of the VOC dataset. The task was to localize motorbikes in the given pictures. With a single call to fit() and using predict() to test the model, AutoGluon was able to generate a reasonably accurate visualization image (shown below).
The AutoGluon project is on GitHub.
Author: Hecate He | Editor: Michael Sarazen