Running machine learning models

Machine learning and especially deep learning has been a rising topic ever since AlexNet's triumph in 2012 (https://papers.nips.cc/paper/4824-imagenet-classification-with-deep-convolutional-neural-networks.pdf), with the language of choice being mostly Python for its easy-to-use syntax and flexibility. However, the underlying frameworks (TensorFlow, PyTorch, and more) are commonly built using C++, not only for performance reasons but also because accessing hardware (such as a GPU) is a lot easier. Rust has—so far—not been the language of choice to implement lower-level frameworks. Even outside the area of deep learning, Rust lacks library support in several areas including data preparation, classical machine learning, and optimization (progress is tracked here: http://www.arewelearningyet.com/)—so, why bother using Rust...