A | B | C | |
---|---|---|---|
1 | Filename | Demos featured in | Description |
2 | ml_unsupervised_clusterKmeans.m | demo_unsupervised_clustering | Cluster observations in model matrix X using K-means algorithm with optional K-means++ initialization |
3 | ml_unsupervised_DBcluster.m | demo_unsupervised_clustering | Cluster observations in model matrix X using density-based clustering algorithm |
4 | ml_unsupervised_dimRedFA.m | demo_unsupervised_PCA_FactorAnalysis | Reduce dimensionality of model matrix X using Factor Analysis |
5 | ml_unsupervised_dimRedKPCA.m | demo_unsupervised_ISOMAP, demo_unsupervised_kernelPCA | Use kernalized PCA to reduce dimensionality of model matrix X |
6 | ml_unsupervised_dimRedPCA.m | demo_unsupervised_PCA_FactorAnalysis | Use PCA to find best low-rank approximation of model matrix X where best is minimizing error in Frobenius norm |
7 | ml_unsupervised_HMM.m | demo_unsupervised_HMM | Models observed data in input X as generated by a hidden state with a chain structured having the Markov property |
8 | ml_unsupervised_LGSSM.m | demo_unsupervised_KalmanTracking | Models observed data X as generated by a Gaussian distribution whose parameters depend upon a latent variable distribution, also Gaussian distributed, which changes over time according to a linear combination of previous states |
9 | ml_visualize_ISOMAP.m | demo_unsupervised_ISOMAP | Apply multidimensional scaling using geodeisc distance defined over nearest neighbours graph to learn manifold and visualize in lower dimensions |
10 | ml_visualize_MDS.m | demo_unsupervised_MDS_Sammon | Apply MDS with Euclidean distance function |
11 | ml_visualize_Sammon.m | demo_unsupervised_MDS_Sammon | Apply MDS with Sammon's distance function |
12 | ml_visualize_tSNE.m | demo_unsupervised_tSNE | Find low-dimensional representation of model matrix X by minimizing KL-divergence between probability distributions representing similarity of points in high-dimensional space and similarity of points in low-dimensional points |
13 | ml_unsupervised_sparseAutoencoder.m | demo_unsupervised_sparseAutoencoder | Fit a feed-forward neural network with linear neurons using sigmoid activiation function, using reconstruction error (Frobenius norm) as loss function. |
14 | ml_unsupervised_HMM.m | demo_unsupervised_HMM.m | Models observation data in input X as generated by a chain-structured latent variable distribution with the Markov property |
15 | ml_unsupervised_LGSSM.m | demo_unsupervised_KalmanTracking.m | Implements Kalman filtering and Kalman smoothing for directed probabilistic models that have Gaussian latent variables with a Markov transition structure and Gaussian observation structure. |