Predictions using ANNs and CNNs

  • Donald J. Norris


In the previous chapters, I have repeatedly demonstrated how ANNs and CNNs can classify a variety of objects including handwritten numbers and clothing articles. In this chapter I will explore how ANNs and CNNs can predict an outcome. I have noticed repeatedly that DL practitioners often conflate classification and prediction. This is understandable because these tasks are closely intertwined. For instance, when presented with an unknown image, a CNN will attempt to identify it as belonging to one of the classes it has been trained to recognize. This is clearly a classification process. However, if just view this process from a wider perspective, you could say the CNN has been tasked to predict what the image represents. I choose to take the narrower view and restrict my interpretation of prediction, at least as far as it concerns ANNs and CNNs to the following definition:

Copyright information

© Donald J. Norris 2020

Authors and Affiliations

  • Donald J. Norris
    • 1
  1. 1.BarringtonUSA

Personalised recommendations