Machine Learning, Deep Learning and Applications

Abstract

This mini-course will present and briefly comment on some basic Machine Learning concepts related to the types of learning they develop, which can be: Supervised Learning and Unsupervised Learning. Within Supervised Learning we find the following types of Neural Networks: Artificial, Convolutional and Recurrent. As for Unsupervised Learning, we have Self-Organizing Maps, Boltz Machines, Autoenco-ders and Generative Adversarial Networks. Supervised Learning has some applications such as classification and regression, computer vision, time series analysis, among others; in Unsupervised Learning, we can find applications in feature detection and clustering, recommendation systems, dimensionality reduction, image generation, etc. It will comment on the history and theory of Artificial Neural Networks and exemplify how they work by calculating the weights of the input and output layers, which are real numbers that represent the learning of the Artificial Neural Network. Finally, an application of Convolutional Neural Networks will be made to  classification of cats and dogs using Python software and/or Google Colebe online. Deep Learning are Neural Networks with more than two hidden layers. An activity will be developed in which participants will calculate the weights of a one-layer Artificial Neural Network called Feed Forward. The mini-course will be offered over two days, with the same content on each day and the number of places will be between 40 and 50 participants per day, with each day lasting 4 hours. The target audience will be academics from Business Administration, Accounting, Mathematics, academics from other courses, the external community and teachers who are interested in the subject.