Wednesday, December 14, 2016

Has Deep Learning Made Traditional Machine Learning Irrelevant?

Here is a great article in DataScienceCentral by William Vorhies, about neural networks vs traditional machine learning techniques (kNN, random forest, naive bayes, boosting techniques,  etc.).

It answers the simple question: is deep learning the future.  His answer, of course, is no.  Regardless of what is seen in tech media articles and Kaggle competitions, a lot of problems, especially with structured and understood data, can be solved more simply using normal machine-learning algorithms.

We mentioned earlier that there are at least 28 unique architectures for ANNs, many of which are quite specialized like the many-hidden layers necessary in CNNs and RNNs.  If your business has image or NLP unstructured data that needs to be analyzed then using CNNs and RNNs is the way to go.
But keep in mind:
  • CNNs and RNNs are very difficult to train and sometimes fail to train at all.
  • If you are building a CNN or RNN from scratch you are talking weeks or even months of development time.
  • CNNs and RNNs require extremely large amounts of labeled data on which to train which many companies find difficult or too costly to acquire.
In fact the barriers to de novo CNN and RNN creation are so steep that the market is rapidly evolving toward prebuilt models available via API from companies like Amazon, Microsoft, IBM, Google, and others.

I predict (no pun intended) that the webservice offerings by Amazon and Google will continue to evolve and become a bigger deal in the near future (like 2017!).