The course will provide fundamentals about theoretical, technical and practical issues in the design and implementation of machine learning systems (classification, clustering, function approximation and prediction problems) based on Computational Intelligence techniques (neural networks, fuzzy logic, evolutionary algorithms), focusing the attention on data and big data analytics that are relevant to atmospheric science and technology applications.
Program in brief
|
INTRODUCTION TO MACHINE LEARNING. Data driven modelling. Soft Computing, Computational Intelligence. Basic data driven modelling problems: pattern recognition, clustering, classification, unsupervised modelling, function approximation, prediction. OPTIMIZATION PROBLEMS. Optimization problem building; Optimality conditions. Linear interpolation and regression. Least Squares algorithms. Orthogonal data transforms. Principal Component Analysis. Numerical optimization algorithms: steepest descent and Newton’s method. ESTIMATION. Non parametric estimation techniques. Parametric estimators: Maximum likelihood, Maximum A Posteriori, Bayesian estimation techniques. Complexity and generalization capability (Occam razor, AIC, MDL, BIC). FUZZY LOGIC. Fuzzy inference principle. Fuzzy Rules. CLUSTERING ALGORITHMS. K‐means, Gaussian mixtures. Clustering performance indexes. Hierarchical clustering. Agent based cluster mining. CLASSIFICATION SYSTEMS. Decision surfaces and discriminant functions. Performance and sensitivity measures. Bayesian classifiers. Classification models based on cluster analysis. Decision trees. DISTRIBUTED LEARNING FOR COMPUTATIONAL INTELLIGENCE APPLICATIONS. neural networks, fuzzy logic, evolutionary algorithms; middleware services and agents; grid computing and cloud computing. DEEP LEARNING COMPUTING. Architectures and applications. LABORATORY. Case studies on multispectral data fusion and image processing, object detection, distributed sensor networks, time series prediction, chaotic systems, Big Data Analytics and atmospheric science and technology applications. |
- Docente: ELIO DI CLAUDIO