AmWebTriangle is a solution for experimenting with Machine Learning. The idea is to have an easy and unlimited source of datasets, that can feed models, that can classify samples with measurable quality, requiring minimal user input. Each sample, in each dataset, is a trio of numbers, corresponding to a triangle's internal angles, properly classified, adequate for supervised learning applications. The datasets are JSON files, ready to train models, then usable for the classification of new samples. It is intended to incentivize educational testing with automatic learning, making it easy to generate and visualize samples and datasets, and to train models, via tools with Web and CLI interfaces, with the freedoms of "libre" software. The user follows his own workflow, including the option to adapt the trio of numbers to another reality with a different quantity of features.
Marques, Artur and Madeira, Filipe, "AmWebTriangle: a toolkit for experimenting with Machine Learning" (2022). CAPSI 2022 Proceedings. 52.