Telecommunication Engineering

First part: Introduction to Computational Intelligence: Soft Computing; Pattern Recognition; Machine Learning. Complex systems. Generalization capability. Ampliative and non-ampliative logic inferences. Deduction and Induction. Induction over metric spaces. Artificial Intelligence. Turing test. Necessary ingredients for intelligent behavior. Oriented and non-oriented processes. Process sampling. Classification problem definition. Label space. Dynamical processes. Definitions of classification, function approximation and clustering problems. Design chain in pattern recognition and machine learning. Representation and preproccessing functions. Feature generation and selection. k-clustering problems and free-clustering problems. Distance functions. Minkowsky distances. Analogy as a fundamental inference in decision theory. k-NN Algorithm. Data driven modelling systems. Statistic and affine normalization. Cluster representatives: difference between metric and non-metric spaces. Points to sets dissimilarities. Fundamental data types in machine learning. Correlation and covariance matrices. Mahalanobis Distance. K-NN algorithm for function approximation. k-means algorithm. MinSod Representative. K-medoids algorithm. Learning as optimization. Optimization problems. Gradient, Hessian. Linear functions and quadratic forms. Necessary and sufficient condition for local and global minima. Numerical optimization methods. Gradient descent. Newton method. Linear regression and Least Squares Estimator (LSE) algorithm. The perceptron. Multi-layer perceptrons: the xor case. Multi-layer perceptrons: the case of connected and non-connected decision regions. Multi-layer perceptron for function approximation. Error back propagation. Evolutionary optimization: Genetic Algorithms. Introduction to fuzzy systems. Fuzzy inference systems. Fuzzy reasoning. Modus ponens. Generalized modus ponens. Sugeno and Mamdani rules. ANFIS networks. ANFIS networks training by gradient descent and by clustering techniques. Second part: Setup and configuration of an Anaconda environment for machine learning application prototyping. Visual Studio code and GitHub. Usage of integrated development environments and Jupyter notebooks for interactive experimentation in Python. Core data structures in Python and their application in classical machine learning. Introduction to essential Python libraries for data analysis and scientific computing. Overfitting problem. Structural complexity measures. Occam's razor criterion. N-fold cross validation. Performance measures in classification and function approximation. Support Vector Machines and supervised learning problems. Hard Margin SVMs. Soft-Margin SVMs. Kernelized SVM. Working in Non-Metric Spaces. Multiclass SVMs. A Gentle Introduction to Deep Learning. Training Neural Network in practices. Neural Network Architectures. Convolutional Neural Networks. Learning sequences and Recurrent Neural Network. Hybrid architectures. Practice with Python Notebook on Linear and Logistic regression. MLE e gradient descend solutions. Introduction to Deep Learning Python libraries. Tensor Flow and PyTorch. Hardware for Deep Learning. Automatic differentiation. Practice on Python Notebooks. Introduction to time series forecasting. Taxonomy, definitions and main goals. Component of a Time Series and main statistical properties. Traditional Models and Deep Learning models for forecasting. Practice on energy time series forecasting. Computational Intelligence for Energy Management. Introduction, objectives and main drives. Smart Grids and Renewable Energy Communities. Energy Management Systems. Practice on Genetic Algorithms and the optimization of an Energy Management Systems based on a Fuzzy Inference System. Integrating the forecasting module in the loop. Single MG power flows optimization. Renewable Energy Community (REC) optimization.

Elementary notions of Geometry, Algebra, Differential Calculus, Signal Theory, Information Theory, Informatics, Digital Signal Processing.

Kruse, R., Borgelt, C., Braune, C., Mostaghim, S., & Steinbrecher, M. (2016). Computational intelligence: a methodological introduction. Springer. Lecture notes and slides available at teacher's site

The course is organized as a series of lectures and case study illustrations.

It is strongly recommended to attend classroom lessons.

The final exam consists in the evaluation of a homework. The topic of the homework is usually agreed with the teacher.

The course is organized as a series of lectures and case study illustrations.