Antal, BálintBod, Gergely2014-01-162014-01-1620132014-01-16http://hdl.handle.net/2437/178637Self-taught learning is a new framework in the domain of machine learning. It has the potential that by using unsupervised learning strategies to automatically learn and extract information and consequently create a succint representation of the given input. I have presented the theoretical underpinnings of the idea and then proceeded to show a possible implementation of the algorithm in the MATLAB programming language. I have also included my experimental results obtained during my research.39en-USunsupervised learningmachine learningneural networkautoencoderDEENK Témalista::Informatika::Alkalmazott matematikaDEENK Témalista::Informatika