[Dept. of Computer Science]

Experimentation Project ACS


Title Evaluation of Supervised and Unsupervised Machine Vision Algorithms for hand written Digit Recognition
Student Leonidas Lefakis
Supervisor Marco Wiering
ECTS 7.5
Related Course(s) Learning from Data
Description In this project, two different machine vision algorithms are evaluated. In particular experiments are conducted using Yann LeCun's approach of convolutional networks and
Geoffrey Hinton's approach of using restricted Boltzmann machines (rbm) for training deep encoders.

Both these algorithms are tested on the MNIST dataset that consists of hand written digits. In addition to the algorithms above, both of which use supervised learning for classification,
the performance of a number of unsupervised algorithms inspired by these two different approaches is explored.

Special Note