Machine Learning Notes
Just some terms and concepts about machine learning that I want to remember for later.
Binary relevance is a way of transforming a multi-label classification problem into a binary classification problem. A binary classification model is trained for each label, and then the union of positive labels are assigned to a new input.
Multi-label classification is when each example can have 0 or more labels. An example might be choosing words to describe an image. A picture of a dog might have the words “pet”, “dog”, and “mammal” all associated with it.
Sam Stevens, 2020