Artificial Intelligence

Course in Bachelor’s program (3rd year, series E)

Teachers

Lect. Șerban Mihalache

Course Description

This course provides a fundamental perspective on the field of machine learning: the K‑means Model (KMM), the K-nearest Neighbors algorithm (KNN), Gaussian Mixture Models (GMM), Decision Trees (DT), Random Forests (RF), Support Vector Machines (SVM), and fundamental artificial neural networks, particularly Fully‑connected Neural Networks (FCNN). Additionally, an introduction to the field of deep learning is offered: Convolutional Neural Networks (CNN), Residual Neural Networks (ResNet), and advanced training techniques – regularization, dropout, batch normalization, etc.
– The lectures offer an introduction to the specific problems in the field and addresses the three fundamental paradigms (supervised learning, unsupervised learning, and reinforcement learning), providing a comparative analysis of these and the types of applications that are natively suited to each. The most important machine learning techniques and methods are presented, tailored for clustering, supervised classification, and regression problems (KMM, KNN, GMM, DT, RF, SVM, FCNN, CNN, and ResNet).
– The laboratory begins with a comprehensive introduction to using the Python programming language and the essential packages used in the field of machine learning (numpy, scipy, pandas, matplotlib, etc.). The remaining sessions cover the practical implementation of the main machine learning models studied in the lectures, using the scikit-learn and Keras/TensorFlow Python packages, and their use for various clustering, supervised classification, and regression applications.

Contents

Course

  • “Introduction” – Brief history. State of the art. Remarkable results. Ethical aspects and concerns
  • “Fundamental concepts. Paradigms” – Definitions. Machine learning paradigms (supervised, unsupervised, reinforcement) and their comparative analysis. Experimental methodologies and training and testing techniques. Dataset curation principles
  • “The K-means Model (KMM). The K-nearest Neighbors algorithm (KNN)” – Theory. Principles of operation. Advantages and limitations. Examples
  • “Gaussian Mixture Models (GMM)” – Theory. Principles of operation. Advantages and limitations. Examples
  • “Decision Trees (DT). Random Forests (RF)” – Theory. Principles of operation. Advantages and limitations. Examples
  • “Support Vector Machines (SVM)” – Theory. Principles of operation. Advantages and limitations. Examples
  • “Fully-connected Neural Networks (FCNN)” – Theory. Fundamental principles and principles of operation. Advantages and limitations. Examples
  • “Introduction to Deep Learning” – Convolutional Neural Networks (CNN), Residual Neural Networks (ResNet). Advanced training techniques

Laboratory

  • Introduction to the Python programming language and additional packages (numpy, scipy, pandas, matplotlib, etc.)
  • Clustering applications: the K-means Model (KMM), Gaussian Mixture Models (GMM). Classification applications: the K-nearest Neighbors algorithm (KNN)
  • Classification applications: Decision Trees (DT), Random Forests (RF)
  • Classification applications: Support Vector Machines (SVM)
  • Classification and regression applications: Fully-connected Neural Networks (FCNN)
  • Final colloquium

Grading

Laboratory evaluation (continuous): 20%
Laboratory evaluation (colloquium): 20%
Course midterm test (written): 10%
Course final exam (written): 50%