New machine learning algorithms: A suite of one patent and two publications
Contact
- Boian Alexandrov
rNMF Schematic: A suit of three new methods and algorithms for data exploratory analysis, features extraction, data mining and dimension reduction based on subspace learning via robust Nonnegative Matrix Factorization (NMF) with a custom estimation of the unknown number of factors (features), and integrated Green’s function, time delays, and other physical constraints.
New machine learning algorithms: A suite of one patent and two publications
Unsupervised Machine Learning (ML) methods aim to extract sets of latent (and often previously unknown) features from uncategorized datasets. Nowadays, integration of big multicomponent datasets, powerful computational capabilities, and affordable data storage has resulted in active use of advanced ML algorithms. However, ML tools for efficient and robust extraction of latent features buried in petabytes of multicomponent big datasets are still lacking. Identification of the different manifestations of these latent processes in the data is part of exploratory data science that allows discoveries of new mechanisms and causalities hidden in the datasets.
