The University of Cyprus's MSc Artificial Intelligence is part of the Master programmes in Artificial Intelligence 4 Careers in Europe (MAI4CAREU). One of Master's programme's courses, MAI612 - Machine Learning is split up into several lectures. Taught by Vassilis Vassiliades, PhD, this part of the course is divided in three lectures, nineth, tenth, and eleventh respectively, both focusing on Kernel-based Methods. Lectures 9, 10 and 11 of the MAI612 - Machine Learning coursewithin the MAI4CAREU Master in AI, explore the challenge of mimicking the human brain using algorithms - Neural Networks. 

Lecture on Neural Networks 1 - Modelling learning outcomes

Modelling will first review content on Kernel methods. With this lecture you will learn:

• What an artificial neuron is
• The perceptron model, its learning algorithm and limitations
• How to construct nonlinear decision boundaries by combining perceptrons
• Different activation functions
• What a feedforward neural network (NN) model is and how to represent it
• How to implement a NN for fast computation
• Why NNs compute their own features

Lecture on Neural Networks 2 - Training learning outcomes

With this second lecture you will understand: 

• The cost function of neural network models for regression and classification
• How to train neural networks using backpropagation
• How to implement backpropagation efficiently
• Stochastic gradient descent with momentum
• Advanced optimization methods for training neural networks
• How to improve the performance of neural networks using early stopping, hyperparameter tuning and ensembles
• Evolving neural networks

In particular, the course presents a Step by Step guide for Online Updating of the Backpropagation Algorithm. 

Lecture on Neural Networks 3 - Introduction to Deep Learning learning outcomes

The final lecture on Neural Networks begins by examining what deep learning is and why use it. It is divided into four parts: Image Data, Sequential Data, Recurrent Neural Networks, and Text Data. In this lecture you will learn about:

• What deep learning is and why use it
• Convolutional networks for image data
• Handling sequential data
• Recurrent neural networks
• Backpropagation through time
• The vanishing and exploding gradient problem
• Echo State Networks
• Long short-term memory (LSTM) networks
• Word embeddings for handling text