Tensorflow is one of the best things to happen for anyone who wants to leverage machine learning to solve some sort of problem.
In case you didnít know, machine learning is a specialized technique where computer software and hardware essentially create their own programming to find particular things. For example, by looking at millions of examples of skin cancer, itís been possible to create an algorithm that can diagnose skin cancer as well (or better) than a dermatologist. These same techniques can also be used to make predictions about how things will turn out in real life, based on observations of historical data. Thatís what machine learning does in general. It looks at heaps and heaps of examples of something and then discovers patterns and insights that can be used to build useful tools.
Our computer systems are now powerful enough to do some amazing things, but thatís only half of the story. You need some pretty advanced software to imbue those electronic brains with the knowledge of how to look for these patterns. Thatís where machine learning libraries come into play and thatís exactly what Tensorflow is.
The Library is Open
You can think of Tensorflow as a set of machine learning tools that anyone who has a problem solvable through those methods can use. You donít need to create the base techniques and technology from scratch. Instead Google, who created Tensorflow, has made it free and Open Source. Theyíve also made it one of the easiest ML libraries to learn and implement, which is why itís so popular.
Fundamentally, in order to use the software, you only need to learn the Python programming language. The user-facing side of Tensorflow is fully integrated into this relatively easy language. In the background itís all high-speed, complex C++ performing the complicated calculations.
This opens these powerful tools to just about anyone, whether a massive corporation or a small app development team. By using this incredible toolset you donít need to be an expert in how the low level process of machine learning works. You just need to define your overall problem and data flow, provide the hardware horsepower to run it on and let Ďer rip.
Power to the People
While Tensorflow itself is free and Open Source, you need to have some hefty hardware to make the best of it. Google themselves offer Tensorflow-optimized cloud-based hardware, but you can do a lot with a powerful workstation or local cluster of computers. Often itís a mix of both, with the cloud systems doing the big initial work and local workstations crunching data by applying the generated algorithms to them. All we know is that Tensorflow is a hot topic among our customers and weíve sold plenty of AAA hardware destined for this machine learning suite.
Operating systems supported by TensorFlow
Ubuntu 16.04 or later (64-bit)
Windows 7 or later (64-bit) (Python 3 only)
Raspbian 9.0 or later.
ADDITIONAL COMPATIBLE WORKSTATIONS