Algorithm Support

MLDB Procedures are used to train models, which can then be applied via Functions.

Supervised Machine Learning

• The classifier.train procedure type can train:
  • Generalized Linear Models
  • Logistic Regressions
  • Decision Trees (including Random Forests)
  • Neural Networks
  • Naive Bayes models
  • with Bagging
  • with Boosting
• The svm.train procedure type can train Support Vector Machines (SVM)
• The probabilizer.train procedure type can calibrate classifiers

Deep Learning

• The tensorflow.graph function type can execute TensorFlow models

Clustering

• The kmeans.train procedure type can train K-Means models

Dimensionality Reduction, Visualization & Manifold Learning

• The svd.train procedure type can perform Truncated Singular Value Decompositions (SVD)
• The tsne.train procedure type can perform t-distributed Stochastic Neighbor Embedding (t-SNE)

Feature Engineering

• The import.sentiwordnet procedure type can import SentiWordNet models
• The import.word2vec procedure type can import Word2Vec embeddings
• The tfidf.train procedure type can train Term-Frequency/Inverse-Document-Frequency (TF-IDF) models
• The statsTable.train procedure type can assemble tables of counts to assemble count-based features
• The feature_hasher function type can be used to do feature hashing, which is a way to vectorize features