Providing basic elements, concepts and fundamental tools, based on computational methods, to represent learning, knowledge and reasoning in conditions of uncertainty. Designing and developing software tools based on machine learning techniques to solve data analysis, knowledge discovery and decision support problems in the presence of uncertain or incomplete information.
teacher profile teaching materials
Classical AI System, Machine Learning Definition, Classical Approach and Applications.
Fundamentals of Machine Learning, Types of Learning, Training Methods,
Generalization Methods. Data Issues, Insufficient Data, Non-Representative
Data, Poor Quality Data. Machine Learning Issues Related to
Modeling, Model Selection, Complexity and Hyper-Parameter Setting.
Supervised Learning for Regression Problems. Linear Models, Mean
Square Error, Learning as MSE Minimization. Polynomial Regression.
Overfitting and Underfitting. Hyper-Parameter Optimization, Validation Set. Examples of
Programming in Matlab and Python Language.
Supervised Learning for Classification Problems. Binary Classification,
Logistic Regression. Metrics for binary classification problems, accuracy,
precision and recall, confusion matrix, F1-score, ROC and AuC. Separating hyperplanes
for binary classification problems, theoretical aspects and definition of hyperplane, concept of
margin, support vectors and linear non-separability. Programming examples in
Matlab and Python language.
Fundamental algorithms for supervised learning. Support vector
machines, separating hyperplanes with soft-margin constraints, kernel tricks and linearity.
Machine learning decision trees, choice of attributes and values, information
entropy. Ensemble learning, parallel models, random forest.
Sequential ensemble models and boosting. Model selection methods, validation set,
error decomposition and balancing, bias-variance trade-off.
Fundamental algorithms for unsupervised learning.
Clustering algorithms. K-means and optimal value of k, clustering applications, supervised approach,
semisupervised, with clustering. DBSCAN, practical rules advantages and disadvantages of
DBSCAN compared to k.means. Algorithms for dimensionality reduction, PCA, example
of using PCA for dimensionality reduction.
Learning with artificial neural networks. The perceptron. Multi-layer perceptron networks
(MLP). Clustering with neural networks. Learning vector quantization (LVQ). The
manifold learning problem and SOM networks. Deep learning with neural networks. Fundamental principles.
Convolutional neural networks (CNN). Notes on advanced Deep-Learning architectures. . Convolutional
neural networks, convolution layer, activation layer, pooling layer. Examples of
programming in Matlab and Python language.
Analysis, selection and transformation of data. Image analysis, decomposition in YUV/YCbCr color spaces. Time/frequency data analysis, Fourier transform (outline),
spectrograms. Programming examples in Matlab and Python.
Natural language processing. Transfer learning and model architectures for text
classification, question-answering, machine translation and text
generation.
Book: G. Barone, “Machine Learning and Artificial Intelligence: methodologies for the development of automatic systems”, Dario Flaccovio Editore, 217 pp.
Programme
Artificial Intelligence and Machine Learning Fundamentals. Types of Artificial Intelligence,Classical AI System, Machine Learning Definition, Classical Approach and Applications.
Fundamentals of Machine Learning, Types of Learning, Training Methods,
Generalization Methods. Data Issues, Insufficient Data, Non-Representative
Data, Poor Quality Data. Machine Learning Issues Related to
Modeling, Model Selection, Complexity and Hyper-Parameter Setting.
Supervised Learning for Regression Problems. Linear Models, Mean
Square Error, Learning as MSE Minimization. Polynomial Regression.
Overfitting and Underfitting. Hyper-Parameter Optimization, Validation Set. Examples of
Programming in Matlab and Python Language.
Supervised Learning for Classification Problems. Binary Classification,
Logistic Regression. Metrics for binary classification problems, accuracy,
precision and recall, confusion matrix, F1-score, ROC and AuC. Separating hyperplanes
for binary classification problems, theoretical aspects and definition of hyperplane, concept of
margin, support vectors and linear non-separability. Programming examples in
Matlab and Python language.
Fundamental algorithms for supervised learning. Support vector
machines, separating hyperplanes with soft-margin constraints, kernel tricks and linearity.
Machine learning decision trees, choice of attributes and values, information
entropy. Ensemble learning, parallel models, random forest.
Sequential ensemble models and boosting. Model selection methods, validation set,
error decomposition and balancing, bias-variance trade-off.
Fundamental algorithms for unsupervised learning.
Clustering algorithms. K-means and optimal value of k, clustering applications, supervised approach,
semisupervised, with clustering. DBSCAN, practical rules advantages and disadvantages of
DBSCAN compared to k.means. Algorithms for dimensionality reduction, PCA, example
of using PCA for dimensionality reduction.
Learning with artificial neural networks. The perceptron. Multi-layer perceptron networks
(MLP). Clustering with neural networks. Learning vector quantization (LVQ). The
manifold learning problem and SOM networks. Deep learning with neural networks. Fundamental principles.
Convolutional neural networks (CNN). Notes on advanced Deep-Learning architectures. . Convolutional
neural networks, convolution layer, activation layer, pooling layer. Examples of
programming in Matlab and Python language.
Analysis, selection and transformation of data. Image analysis, decomposition in YUV/YCbCr color spaces. Time/frequency data analysis, Fourier transform (outline),
spectrograms. Programming examples in Matlab and Python.
Natural language processing. Transfer learning and model architectures for text
classification, question-answering, machine translation and text
generation.
Core Documentation
Lecture notes by the teacher on the University Moodle and MS Teams platformBook: G. Barone, “Machine Learning and Artificial Intelligence: methodologies for the development of automatic systems”, Dario Flaccovio Editore, 217 pp.
Attendance
2 lessons per week with optional attendanceType of evaluation
Written test with multiple answers in moodle Analysis and implementation project of a ML/DL method on datasets of various nature