Understanding interestingness – similarities
We started off this chapter discussing Flickr and its interestingness algorithm: the enigma of an algorithm which brings forth some amazing photos for us to explore and enjoy from millions uploaded every day. The Explore page presents a pretty diverse set of photos showcasing different photography styles, clicked using different photography equipment by photographers of varied skills. Yet the interestingness algorithm picks them all!
Through this use case we will try to find out what type of photos are being picked up by the algorithm and understand/uncover if there are certain patterns or similarities between such interesting photos.
Since we do not know much about the algorithm other than the fact that it presents excellent photos to explore, we'll take the unsupervised approach to see what we have here.
Under the unsupervised umbrella of machine-learning algorithms, one of the simplest and most widely used algorithms to get started with is the...