Harokopio University
School: School of Digital Technology
Department: Informatics and Telematics
Program: Undergraduate Programme

Machine Learning and Applications

Semester: 6 ECTS: 5.0 Elective Erasmus

General

Code: ΕΠ34

Language: Greek

Delivery: Face-to-face

Prerequisites: Computational Mathematics
Discrete Mathematics
Probability Theory
Numerical Analysis
Programming I & II
Object Oriented Programming I & II
Artificial Intelligence

Workload

  • Lectures: 39.0 hours
  • Lab: 13.0 hours
  • Study: 40.0 hours
  • Project: 33.0 hours

Course Content

- Introduction to Machine Learning (ML). Definitions. Recent developments and successes.
- Recap: Types of machine learning problems. Machine learning model generalization. Simple and multiple linear regression. Solution via ordinary least squares and normal equations. Logistic regression and the Perceptron model. Extension to non-linear models.
- Maxmimum likelihood estimation for ML model training.
- Data preparation. Normalization, standardization, one-hot encoding, cyclic encoding.
- Applications and examples in the scikit-learn environment.
- Artificial Neural Networks (ANN) and multi-layer perceptrons.
- ANN training and the backpropagation algorithm.
- Introduction to Pytorch
- Convolutional neural networks for representation learning in signals and images.
- Regularization methods in ML.
- Recurrent Neural Networks (RNNs)
- Applications and examples in image classification.
- Word vector representation learning.
- Applications and examples in text classification.
- Introduction to attention mechanisms
- Transformer models
- Overview of the ChatGPT (OpenAI) and PaLM (Google) architectures

Learning Outcomes

This course aims to familiarize students with modern machine learning techniques and their applications, focusing particularly on deep learning methods with Artificial Neural Networks. Students learn to handle data and apply machine learning techniques to tabular data, signals, images, and texts, using languages and tools like Python, scikit-learn, and PyTorch. Additionally, the course covers theoretical and practical topics such as data preparation, model training and evaluation, and advanced topics like Convolutional and Recurrent Neural Networks, attention mechanisms, and transformer models, including the popular BERT and GPT architectures.

Skills

- Adaptation in new conditions
- Independent work
- Team work
- Decision making
- Promoting free, creative and deductive reasoning