A cluster

Classification

Number of clusters

Single linkage

Algorithm overshoots minimum

Good performance on training but poor on testing

Reduce model complexity and prevent overfitting

Sigmoid

Binary classification

Sigmoid, Tanh

Clustering, Association Rule Mining, Anomaly Detection

Learning Rate, Number of Hidden Layers, Maximum Iterations

L1 Regularization, L2 Regularization, Dropout

Single linkage, Complete linkage, Centroid

Neither underfitting nor overfitting ,Good performance on validation set ,Generalizing well to unseen data