Machine Learning algorithms, especially the computational intense artificial neural networks, have shown great success in the analysis and prediction of big data sets in recent years. This has driven the development of a data analysis ecosystem in Python, which lead to the appearance of several deep learning libraries, TensorFlow being one of the most prominent ones. Those frameworks specialize on efficient, large scale numerical computations while being lightweight to use. While their main purpose is the building and training of neural networks, they are also used in a wider array of computational intense tasks. An example of such is likelihood model fitting in High Energy Physics as done by the zfit project and other libraries.
The talk covered the principles, functionality and possibilities of the TensorFlow library and its extensions beyond its usual application for machine learning.