--- title: Supervised Learning tags: ml --- ### Supervised Learning - Predicting labels given input features - Regression - A problem is a regression when the output labels take on arbitrary numerical values - Classification - Common loss function is cross entropy - Hierarchical classification - Not all errors are equal so if we are unsure, it’s better to classify something as one thing over another - Tagging - Predicting classes that are not mutually exclusive so a bounding box or image can be labelled as multiple different things - Search - Display set of relevant items to the user - Recommender systems - Display set of relevant items to the user but personalization to specific users - Sequence learning - Require a model to ingest sequences of inputs or to emit sequences of outputs (or both)