CliqueLib

“A Machine Learning Library aiming to provide reliable ensemble classifiers”

Aliraza Punjani (amp2280) Stanislav Karel Peceny (skp2140) Srinidhi Srinivasa Raghavan (ss5145)

CliqueLib is a C++ machine learning library implementing ensemble methods. Ensemble methods use several machine learning algorithms in order to achieve better prediction than could be gained by using any of the individual algorithms. Hence, the library would provide implementation for classification algorithms like logistic regression, SVMs, or KNN classifiers. These algorithms could be either used as standalone algorithms or combined to form an ensemble classifier. C++ libraries like MultiBoost and MLPack provide support for just Adaboost without allowing the usage of other classifiers for boosting. Our library aims to provide a spectrum of ensembles for the user to choose one that’s suitable for their requirements.