A Simple Automated Image Classifier illustrates how to implement a simple automated image-classifier in Mathematica. Among other things, the example shows the importance of providing an automated classifier with a sufficiently diverse, and sufficiently large, universe of examples during the classifier’s training phase. If we don’t, the results are likely to be unreliable. For example, iIf we “tell” such a classifier that any given object in the universe is either green cheese or beef jerky, the classifier will identify anything we ask it to classify (including a picture of Donald Trump) as green cheese or beef jerky, and nothing else. This problem is not peculiar to automated image-classifiers: it is one of the sources of sensory illusions and tribalisms.
The file at the link above was last updated on 11 November 2019 at 0800 US Central time.
Three Philosophers Walk into a Classifier illustrates how to implement an automated text classifier in Mathematica.
The file at the link above was last updated on 8 March 2021.
___________________________________________________________
Click here for an illustration of using the Mathematica function FeatureExtraction as an image classifier.
_____________________________________________________________
Click here for a Mathematica demonstration the vertex-degree distribution of the metabolic network of Y. pestis is a power-law.
Click here for an automated classifier of Clovis vs. Solutrean projectile points
This page was last updated on 19 June 2021 at 1011 CT.