Most nonparametric algorithms have hyperparameters that do not describe the model itself, but rather tweak the way the algorithm works and thereby indirectly affect the model it learns. A typical neural network is nonparametric because the user does not directly specify or control the way it learns, but it nonetheless has a number of hyperparameters including, for example, the number of neurons in each layer.
- used by
- ALG_Adaptive resonance theory network ALG_Association rule learning ALG_Autoencoder ALG_Convolutional neural network ALG_DBSCAN ALG_Decision tree ALG_Deep Q-network ALG_Discriminant analysis ALG_Expectation maximization ALG_Hopfield network ALG_Latent semantic indexing ALG_Learning Vector Quantization ALG_Local outlier factor ALG_Local regression ALG_Long short-term memory network ALG_Nearest Neighbour ALG_Neural actor-critic ALG_Perceptron ALG_Policy gradient estimation ALG_Probabilistic latent semantic indexing ALG_Projection pursuit ALG_Radial basis function network ALG_Random forest ALG_Restricted Boltzmann machine ALG_k-means ALG_k-medians ALG_k-medoids