The Wolfram Language includes a wide range of state-of-the-art integrated machine learning capabilities, from highly automated functions like Predict and Classify to functions based on specific methods and diagnostics, including the latest neural net approaches. The functions work on many types of data, including numerical, categorical, time series, textual, image and audio.

Learning from Input and Output

Classify — classify data into categories using a built-in classifier or learning from examples

Predict — predict values from data using a built-in predictor or learning from examples

ClassifierFunction — symbolic representation of a classifier to be applied to data

PredictorFunction — symbolic representation of a predictor to be applied to data

ClassifierMeasurements, PredictorMeasurements — performance on test data

Learning from Sequences

SequencePredict — predict subsequent elements from sequence examples

SequencePredictorFunction — symbolic representation of a sequence predictor

Specific Methods

Nearest  ▪  FindFit  ▪  NonlinearModelFit  ▪  FindHiddenMarkovStates  ▪  ...

Automated Structure Discovery

LearnDistribution — learn the underlying distribution of any type of data

DimensionReduction — find how to project data onto lower-dimensional space

LearnedDistribution  ▪  DimensionReduce  ▪  DimensionReducerFunction

FeatureExtraction — find how to extract features from data

FeatureExtract  ▪  FeatureExtractorFunction  ▪  FeatureNearest

ClusterClassify — classify data into clusters

FindClusters  ▪  ClusteringTree  ▪  ClusteringComponents

FindDistribution — find a representation for data in terms of named distributions

Visualization

FeatureSpacePlot — visualize dimension-reduced feature space in 2D

FeatureSpacePlot3D — visualize dimension-reduced feature space in 3D

Dendrogram — visualize hierarchical clusters

Structure-Oriented Data Processing

AnomalyDetection — learn an anomaly detector function from data

FindAnomalies  ▪  DeleteAnomalies  ▪  AnomalyDetectorFunction  ▪  RarerProbability

SynthesizeMissingValues — fill in missing values by imputing from existing data

Specific Methods

SingularValueDecomposition  ▪  FindGraphCommunities  ▪  SmoothKernelDistribution  ▪  ...

Learning from Actions

BayesianMinimization — model-based minimization of arbitrary objective functions

ActiveClassification — learn a classifier by actively probing a system

ActivePrediction — learn a predictor by actively probing a system

ActiveClassificationObject  ▪  ActivePredictionObject

Neural Networks »

NetGraph — represent an arbitrary neural network structure

NetChain  ▪  LinearLayer  ▪  ConvolutionLayer  ▪  GatedRecurrentLayer  ▪  ...

NetTrain — train any neural network on CPUs, GPUs, etc.

NetModel — collection of trained and untrained models

Machine Learning Options

FeatureExtractor — how to extract features to learn from

FeatureTypes — feature types to assume for input data

PerformanceGoal — whether to optimize for memory, quality, or speed

TimeGoal — how long to allocate for training, etc.

RandomSeeding — how to seed randomization

ComputeUncertainty — return values including uncertainty (as Around)

MissingValuePattern — specify how missing values are represented in data

Preparing Data »

DeleteMissing — delete missing elements in data

Standardize — transform data to have zero mean and unit variance

Clip  ▪  Rescale  ▪  Threshold  ▪  LogisticSigmoid  ▪  ImageAdjust

CountsBy  ▪  GroupBy  ▪  SortBy  ▪  DeleteDuplicates

Filtering Data »

MovingAverage — compute moving averages of lists, time series, etc.

GaussianFilter  ▪  MeanFilter  ▪  MeanShiftFilter  ▪  LowpassFilter  ▪  ...

Machine Vision Applications

ImageIdentify — recognize objects in images

ImageCases  ▪  FindFaces  ▪  TextRecognize  ▪  ImageGraphics  ▪  ...

Natural Language Processing Applications

LanguageIdentify  ▪  TextStructure  ▪  TextCases  ▪  ...

FindTextualAnswer — find answers to questions from text

Creative Applications

ImageRestyle — restyle an image according to samples