Similarity Search
Similarity search provides a way to find the objects that are the most similar, in an overall sense, to the object(s) of interest. A typical example is that of a doctor finding the top 10 past patients who are most similar to the current patient of interest. This could be used for diagnosis, but also adds the human judgement that some other machine learning methods do not necessarily offer. Another example of approximate similarity search is for finding the song in a database corresponding to a given sound sample, or finding the person in a database corresponding to a face photo. Similarity searches can be thought of as multidimensional analogs to SQL queries. SQL queries are composed of conditions on individual variables, for example “Find all customers whose age is within a certain range and whose income is greater than a certain amount”, whereas similarity searches are more like “Find all the customers most like this one”.
