classification - predictor variable is categorical
binary classification -
multi-ary classification -
sentiment classification -
ont-to-one -
one-to-many -
image captioning -
one-to-many -
many-to-many -
transcription - unstructured data to discrete textual form (e.g. Optical Character Recognition)
machine translation - sequence-to-sequence data
structured output - broad category that subsumes transcription and machine translation
anomaly detection - finds unusual data (e.g. credit card fraud)
synthesis & sampling - generate new examples similar to training examples (e.g. speech synthesis)
imputation of missing values - predict the values of missing entries
denoising - given corrupted example obtained an unknown corruption process from a clean example, predict the clean example. or more generally predict the conditional probability distribution 𝐏(clean example|corrupted example)
probability mass/density function estimation - learn the joint probability of training examples (can solve other tasks like imputation)
---cognitive-computing---machine-intelligence/ai---subfields/machine-learning-(ml)---pattern-recognition/ml---tasks／problems/machine-learning-task-types.png)