QUICK START GUIDE
Machine Learning with MATLAB
Train Models in Learner Apps Machine Learning Functions
Import data from Enable PCA for Select among all Export model to Naming Convention
workspace or file feature reduction common algorithms further optimize fit + c(lassification) / r(egression) + model
or integrate with e.g., for SVM classifier m = fitcsvm(X,Y)
other code
Algorithm Model
Decision tree c,r tree
Linear model c,r linear
Support vectors c,r svm
Visually assess
model Gaussian kernel c,r kernel
performance Ensembles (incl. random ensem-
c,r
forest) ble
K-nearest neighbor c knn
Discrim. analysis c discr
Naïve Bayes c nb
Gaussian process r gp
Compare model Shown: Classification Learner (Gen.) Linear regression (g)lm
performance Similar: Regression Learner
Nonlin. regression nlm
Feature Selection Hyperparameter Tuning Deploy
Standalone, Web Apps, Spark
Neighborhood Component Analysis Explore and
Share as standalone, MapReduce, and
Automate identifying the features with change
Apache Spark™ applications; web apps; and
predictive power. parameters
Microsoft ® Excel® add-ins.
fscnca(X labels, 'Lambda',…); in app:
find(mdl.FeatureWeights > 0.01) Integrate with Enterprise IT/OT
Convert into C/C++, Java®, .NET, or Python®
Automated Bayesian Optimization library using MATLAB Compiler SDK™.
Leverage Bayesian model to decide C-Code Generation
which points in the hyperparameter Automatically convert to C/C++ code for
space to try next. Much faster than embedded deployment using MATLAB
grid search. Coder™
mdl = fit…(X,labels, 1. Train model Mdl = fitcsvm(X,Y);
'OptimizeHyperparameters','auto'); saveCompactModel(Mdl,'mySVM');
2.
3. Define entry-point function
function label = predictSVM(x)
m = loadCompactModel('mySVM');
Also available: Matrix factorization label = predict(m,x);
PCA Stepwise regression end
Sparse filtering Reconstruction ICA 4. Generate C code
t-SNE codegen predictSVM –args {X}
Learn more: mathworks.com/machine-learning
mathworks.com
© 2018 The MathWorks, Inc. MATLAB and Simulink are registered trademarks of The MathWorks, Inc. See mathworks.com/trademarks for a list of additional trademarks.
Other product or brand names may be trademarks or registered trademarks of their respective holders.
