Skip to Main Content (Press Enter)

Logo UNIBG
  • ×
  • Home
  • Degrees
  • Courses
  • People
  • Outputs
  • Organizations
  • Third Mission
  • Projects
  • Expertise & Skills

UNI-FIND
Logo UNIBG

|

UNI-FIND

unibg.it
  • ×
  • Home
  • Degrees
  • Courses
  • People
  • Outputs
  • Organizations
  • Third Mission
  • Projects
  • Expertise & Skills
  1. Courses

DEEP LEARNING - 38113

courses
ID:
38113
Dettaglio:
SSD: Data Processing Systems Duration: 48 CFU: 6
Located in:
DALMINE
Url:
Course Details:
COMPUTER SCIENCE AND ENGINEERING - 38-270/PERCORSO COMUNE Year: 1
COMPUTER SCIENCE AND ENGINEERING - 38-270/PERCORSO COMUNE Year: 2
Year:
2025
  • Overview
  • Syllabus
  • Degrees
  • People
  • Other

Overview

Date/time interval

Primo Semestre (15/09/2025 - 20/12/2025)

Syllabus

Course Objectives

  1. Understand the foundational principles of deep learning, including optimization, neural network training, and generalization.
  2. Design and implement advanced neural network architectures such as CNNs, RNNs, and transformers.
  3. Analyze and apply generative models and unsupervised representation learning techniques.
  4. Identify and address challenges in deep learning, including data bias, distribution shifts, robustness, and domain adaptation.
  5. Apply deep learning methods to real-world problems in computer vision, natural language processing, audio and speech processing, and reinforcement learning.
  6. Critically assess and apply transfer learning, semi-supervised learning, and few-shot learning techniques.
  7. Evaluate the ethical and social implications of deep learning systems, with attention to fairness, bias, and interpretability.
  8. Develop the skills to complete a deep learning project in order to move from problem formulation to model deployment.

Course Prerequisites

This course is designed for students who have some prior exposure to machine learning and are comfortable with basic mathematics and programming. You should have:

  • A general understanding of core concepts in linear algebra, calculus, and probability at a basic level.
  • Some experience with machine learning, such as training classifiers or working with common algorithms. Completion of an introductory ML course (or equivalent experience) is helpful but not mandatory.
  • Basic Python programming skills, including familiarity with libraries like NumPy and Pandas. Prior use of PyTorch or TensorFlow is a plus but not required.



Teaching Methods

The course will use slides and Google Colab for hands-on coding practice. All the materials will be made available to the students before the lectures.


Note that all the lecture materials and the mode of lecture delivery will be in English.


Assessment Methods

Student learning will be verified through a combination of regular homework assignments and a final written exam.


Contents

  • Introduction to Machine Learning and Deep Learning: the first part covers the foundations of learning, including optimization techniques, gradient-based algorithms, generalization, and the training of neural networks.
  • Advanced Neural Networks: convolutional neural networks (CNNs) for image data, recurrent neural networks (RNNs) for sequential modeling, and transformers, which form the basis for modern large-scale models.
  • Generative Models and Unsupervised Representation Learning: generative models include Variational Autoencoders (VAEs), Generative Adversarial Networks (GANs), and diffusion models. As well as unsupervised and self-supervised representation learning methods (e.g., DINO, SimCLR).
  • Challenges in Learning-Based Models: dataset bias, distribution shift, robustness, and generalization, transfer learning, semi-supervised learning, and adaptation techniques that help models perform well across tasks and domains.
  • Applications of Deep Learning:
  • Computer Vision: image classification, object detection, segmentation, image captioning, vision-language models, and text-to-image generation.
  • Natural Language Processing: embeddings, tokenization, large language models like BERT and GPT, fine-tuning, prompting, and machine translation.
  • Audio and speech: audio features, classification, ASR, and models like Wav2Vec and Whisper.
  • Deep Reinforcement Learning: basic RL concepts and deep learning-based policy models.
  • Responsible AI: interpretability, fairness, bias, and ethical considerations in the deployment of deep learning systems.

Online Resources

  • E-learning
  • Leganto - Reading lists

Degrees

Degrees

COMPUTER SCIENCE AND ENGINEERING - 38-270 
Master's Degree
2 years
No Results Found

People

People

ROY Subhankar
AREA MIN. 09 - Ingegneria industriale e dell'informazione
Gruppo 09/IINF-05 - SISTEMI DI ELABORAZIONE DELLE INFORMAZIONI
Settore IINF-05/A - Sistemi di elaborazione delle informazioni
Ricercatori Legge 240/10 - t.det.
No Results Found

Other

Main module

DEEP LEARNING
  • Use of cookies

Powered by VIVO | Designed by Cineca | 26.4.0.0