TensorFlow is an open-source machine learning software library for numerical computation using data flow graphs. It was originally developed by Google.
Microsoft SQL is ubiquitous, while MySQL runs under the hood all over the place. Microsoft SQL is the platform taught in colleges and certification courses and is the one most likely to be used by businesses because it is backed by Microsoft. Its interface is friendly (well, as pleasant as SQL can be) and has been used by so many for so long that resources are freely available if you encounter any issues.
TensorFlow is great for most deep learning purposes. This is especially true in two domains: 1. Computer vision: image classification, object detection and image generation via generative adversarial networks 2. Natural language processing: text classification and generation. The good community support often means that a lot of off-the-shelf models can be used to prove a concept or test an idea quickly. That, and Google's promotion of Colab means that ideas can be shared quite freely. Training, visualizing and debugging models is very easy in TensorFlow, compared to other platforms (especially the good old Caffe days). In terms of productionizing, it's a bit of a mixed bag. In our case, most of our feature building is performed via Apache Spark. This means having to convert Parquet (columnar optimized) files to a TensorFlow friendly format i.e., protobufs. The lack of good JVM bindings mean that our projects end up being a mix of Python and Scala. This makes it hard to reuse some of the tooling and support we wrote in Scala. This is where MXNet shines better (though its Scala API could do with more work).
Microsoft SQL Server Enterprise edition has a high cost but is the only edition which supports SQL Always On Availability Groups. It would be nice to include this feature in the Standard version.
Licensing of Microsoft SQL Server is a quite complex matter, it would be good to simplify licensing in the future. For example, per core vs per user CAL licensing, as well as complex licensing scenarios in the Cloud and on Edge locations.
It would be good to include native tools for converting Oracle, DB2, Postgresql and MySQL/MariaDB databases (schema and data) for import into Microsoft SQL Server.
Theano is perhaps a bit faster and eats up less memory than TensorFlow on a given GPU, perhaps due to element-wise ops. Tensorflow wins for multi-GPU and “compilation” time.
We understand that the Microsoft SQL Server will continue to advance, offering the same robust and reliable platform while adding new features that enable us, as a software center, to create a superior product. That provides excellent performance while reducing the hardware requirements and the total cost of ownership of our solution.
SQL Server mostly 'just works' or generates error messages to help you sort out the trouble. You can usually count on the product to get the job done and keep an eye on your potential mistakes. Interaction with other Microsoft products makes operating as a Windows user pretty straight forward. Digging through the multitude of dialogs and wizards can be a pain, but the answer is usually there somewhere.
We managed to handle most of our problems by looking into Microsoft's official documentation that has everything explained and almost every function has an example that illustrates in detail how a particular functionality works. Just like PowerShell has the ability to show you an example of how some cmdlet works, that is the case also here, and in my opinion, it is a very good practice and I like it.
Community support for TensorFlow is great. There's a huge community that truly loves the platform and there are many examples of development in TensorFlow. Often, when a new good technique is published, there will be a TensorFlow implementation not long after. This makes it quick to ally the latest techniques from academia straight to production-grade systems. Tooling around TensorFlow is also good. TensorBoard has been such a useful tool, I can't imagine how hard it would be to debug a deep neural network gone wrong without TensorBoard.
Other than SQL taking quite a bit of time to actually install there are no problems with installation. Even on hardware that has good performance SQL can still take close to an hour to install a typical server with management and reporting services.
[Microsoft] SQL Server has a much better community and professional support and is overall just a more reliable system with Microsoft behind it. I've used MySQL in the past and SQL Server has just become more comfortable for me and is my go to RDBMS.
Keras is built on top of TensorFlow, but it is much simpler to use and more Python style friendly, so if you don't want to focus on too many details or control and not focus on some advanced features, Keras is one of the best options, but as far as if you want to dig into more, for sure TensorFlow is the right choice
Increased accuracy - We went from multiple users having different versions of an Excel spreadsheet to a single source of truth for our reporting.
Increased Efficiency - We can now generate reports at any time from a single source rather than multiple users spending their time collating data and generating reports.
Improved Security - Enterprise level security on a dedicated server rather than financial files on multiple laptop hard drives.
