ML and Neural Networks are quickly moving past the novelty or fad phase and increasingly becoming an ubiquitous part of the software around us. From location tracking to correction algorithms in our TV sets, chances are that all our devices are already using a form of machine learning to perform everyday calculations.
While we have experience with both Python’s sci-kit and Google’s Tensorflow, we prefer to build our own back-propagation algorithms to enable neural networks to perform specific tasks very efficiently. While ML might not be suitable for all projects, our data-intensive clients are surprised to find that there’s a lot of potential for value-added insight when we build very limited, task-specific neural networks into their applications.