Multiclass classification is a supervised learning task that assigns each input to exactly one of more than two possible classes. The model learns from labeled examples and predicts the class of new, similar inputs. section 3.1.1 section 5.4
This differs from binary classification, which chooses between two outcomes, and multilabel classification, which can assign several labels to one instance. The distinction determines what the prediction represents. section 3.1.1 section 5.5
Compound prediction
A hERG model that predicts whether a compound blocks the channel is binary because it has positive and negative classes. A task with more than two mutually exclusive compound outcomes would be multiclass. section 3.1.1 section 5.4


