Full course description
Conventional machine learning techniques were limited in processing data in their raw forms and a lot of domain experts were required in transforming raw data into meaningful features or representations. Deep Learning techniques have revolutionized many application domains ranging from auditory to vision signal processing. In this course we will study various concepts in deep architectures using both artificial neural networks as well as kernel based models. Several deep learning models such as convolutional neural networks, auto-encoders, generative adversarial networks and their variants among other state-of-the-art models will be covered in depth. We will further study different types of deep architectures used for domain adaptation problems where one is encountered with heterogeneous datasets as well as multi-modal datasets. The regularization and optimization methods used in deep learning framework will be discussed. Tensorflow, an open-source machine learning platform, will be introduced. In this course we will also study deep kernel based models and their connections to artificial neural network based models. This course will be equipped with a practical component, and students are expected to write their own deep learning code and test its performance on various problems. In addition they are strongly encouraged to participate in mini-projects (in a group or individual) targeting a conference paper.
Advanced Concepts in Machine Learning
Papers published in top international conferences and journals in machine learning field.