Advanced Machine Learning


Educational goals

The aim of this course is to give the student a spherical view of the area of machine learning by studying the most important models and learning methods with and without supervision. Moreover, elements of learning theory are offered so that the student can gain insight into the efficiency of the models, their capabilities and their limitations. With the successful completion of the course the student will be able to:

  • Know a wide range of machine learning methods and their application areas
  • Understand the types of problems solved by these methods
  • Analyze a problem that requires the use of machine learning and apply the appropriate method to it
  • Compose solutions in complicated problems by combining machine learning methods
  • Evaluate, using appropriate tools, the performance of a machine learning model or system

Course Contents

Supervised Learning

  • Multilayer Neural Networks, methods and training issues
  • Deep Learning, Deep Belief Networks, Deep autoencoders, Convolutional Neural Networks
  • Probabilistic Bayesian models, Gaussian mixture models, the Expectation-Maximization (EM) algorithm
  • Combining  models, Bagging, Boosting, mixtures of experts
  • Recurrent Neural Networks, Time Delay Neural Networks, training using Backpropagation Through Time, LSTM model, GRU model
  • Bayesian networks, graphical inference models, directed and undirected graphs, Hidden Markov Models

Unsupervised Learning

  • Principal Component Analysis (PCA), Factor Analysis

Reinforcement Learning

  • The armed-bandit problem, Markovian Decision Processes, Dynamic Programming, Monte Carlo methods

Application examples

Teaching Methods - Evaluation

Teaching Method
  • Face to face lectures
  • Optional programming exercises
Use of ICT means
  • Use of the e-learning platform
Teaching Organization
Activity Semester workload
Individual study and analysis of literature128
Total 180
Students evaluation

Final written exam using combination of multiple choice questions, short questions and problem solving questions.
Optional programming exercises

Recommended Bibliography

Recommended Bibliography through "Eudoxus"
  1. Κωνσταντίνος Διαμαντάρας, Δημήτρης Μπότσης, "Μηχανική Μάθηση", Εκδόσεις Κλειδάριθμος ΕΠΕ, Έκδοση: 1η/2019, ISBN: 978-960-461-995-5, Κωδικός Βιβλίου στον Εύδοξο: 86198212
  2. Κωνσταντίνος Διαμαντάρας, "Τεχνητά Νευρωνικά Δίκτυα", Εκδόσεις Κλειδάριθμος ΕΠΕ, Έκδοση: 1η/2007, ISBN: 978-960-461-080-8, Κωδικός Βιβλίου στον Εύδοξο: 13908
Complementary international bibliography
  1. Goodfellow Ian, Bengio Yoshua and Courville Aaron, "Deep Learning", MIT Press, 2016,
  2. Theodoridis, Sergios, "Machine learning: a Bayesian and optimization perspective", Academic Press, 2015.
  3. Bishop, Christopher M., "Pattern recognition and machine learning", Springer, 2006.
Scientific journals
  1. Neurocomputing, Elsevier
  2. IEEE Transactions on Neural Networks and Learning Systems,
  3. Pattern Recognition, Elsevier,
  4. IEEE Transactions on Pattern Analysis and Machine Intelligence