Performance stories

Below are five stories describing the performance of different kinds of AI systems and problems:

(A) A regression solution to predict the price (in thousand dollars) of houses for sale, based on their size, location, and other features.

(B) An e-commerce recommender engine that takes the recent purchase history of a customer and generates a short list of suggested products for them to buy next.

(C) A cat species classification model with a comparison between the predicted class and actual class label for a validation set of 200 cat images, 190 of which have been correctly classified.

(D) A spam e-mail classifier, with a weekly evolution of its performance that shows a stable declining trend over time.

(E) A Proof-of-Concept Generative AI system that generates realistic portrait images of people and has been trained with millions of portrait images of people from only one specific race or ethnicity.