Corso di laurea magistrale Interateneo in Fisica dei sistemi complessi

Have attended the course of Data Mining: Statistical Modeling and Machine Learning of Data (FIS0180), or having an equivalent knwoledge of Python and machine-learning. English language

The course aims to introduce algorithms and machine learning models for cross-disciplinary applications. The students are expected to have a background of Python and machine learning as the foundations provided by the Data Mining course. This is an advanced course.

The students will be able to analyze and model the data with particular emphasis on the predictive models. The theoretical lectures include examples of programming codes.  

The students are expected to possess appropriate knowledge of the theoretical aspects of the presented machine learning models and being able to design new ones.

• Review of machine learning: relation with statistical learning, types of learning (supervised, unsupervised, reinforcement learning, collaborative filtering)
• Introduction to probabilistic models and learning algorithms: Markov models, hidden Markov models, maximum entropy models, conditional random fields. Graphical models as unifying picture. Introduction to causality.
• Introduction to Neural Networks: feedforward, convolutional, recurrent, attention layers, transformers, ODE-NN. Introduction to Neural Networks: feedforward, convolutional, recurrent, attention layers, transformers, ODE-NN. Generative models: Autoencoders, VAE, GAN.
• Introduction to decision trees and ensemble learning (bagging and boosting)
• Concepts of multitask learning and multi-instance learning.
• Survival analysis and machine learning
• Foundations of interpretable explainable AI, local/global methods, intrisically interpretable, agnostic methods (DPD, ICE, ALE, LIME, SHAP)
• Introduction to Deep Reinforcement Learning
• If time allows:
  • Federated Learning
  • Elements of Quantum Machine Learning
  • Elements of biological-inspired computations for optimization: genetic algorithms, ant colony optimization

Watch out! [New course]: starting from this year the lessons will be held in English.

Lectures at the blackboard, slide projection and practical exercises with a computer.

 

Discussion at the blackboard and assignment of programming projects are proposed during the course.

Exam: the student has to select a paper (e.g. from https://paperswithcode.com/), and he/she has to explain the theoretical bases and the Python code. Then some more questions will be asked on the main course points.

• C.M. Bishop. Pattern Recognition and Machine Learning. Springer, 2006.
• Simon J.D. Prince. Understanding Deep Learning, MIT Press 2023.
• Vasilev Ivan, Advanced Deep Learning with Python: Design and implement advanced next-generation AI solutions using TensorFlow and PyTorch, Packt Publishing (12 dicembre 2019)
• K.P. Machine Learning. A Probabilistic Perspective. MIT Press, 2012.
• T. Hastie, R. Tibshirani, and J. Friedman. The Elements of Statistical Learning. Data Mining, Inference, and Prediction. 2nd edition. Springer, 2009.
• I. Goodfellow, Y. Bengio, A. Courville. Deep Learning. MIT Press, 2016.
  R.S. Sutton and A.G. Barto. Reinforcement Learnin. An Introduction. MIT Press, 2017.
• Andrew NG. Machine Learning Yearning: Technical Strategy for AI Engineers, In the Era of Deep Learning. deeplearning.ai, 2018
• Christoph Molnar, Interpretable Machine Learning, ‎ Independently published (28 febbraio 2022)
• D. Simon. EVOLUTIONARY OPTIMIZATION ALGORITHMS. John Wiley & Sons, 2013.
• Plaat A. Deep Reinforcement Learning, pringer Verlag, Singapore; 1st ed. 2022
• Maria Schuld e Francesco Petruccione, Machine Learning with Quantum Computers, Springer Nature; 2° edizione (18 ottobre 2021)

The course assumes that the students are fluent in Python and have basic knowledge of machine learning.