- Oggetto:
Mathematical foundations of deep learning and applications
- Oggetto:
Mathematical foundations of deep learning and applications
- Oggetto:
Academic year 2017/2018
- Course ID
- DP17
- Teacher
- Prof. Rossella Cancelliere (Titolare del corso)
- Year
- 1° anno 2° anno 3° anno
- Teaching period
- Ciclo di incontri
- Type
- A scelta dello studente
- Course disciplinary sector (SSD)
- INF/01 - informatica
- Delivery
- Tradizionale
- Language
- Inglese
- Type of examination
- Relazione finale
- Oggetto:
Sommario del corso
- Oggetto:
Program
The course is an introduction to the basics of Deep Learning, which is the Flagship Technology today in Data Science and Artificial Intelligence. As a subdomain of machine learning, Deep Learning has been at the core of impressive algorithmic and operational progresses over the last 5 years. This family of methods has positioned itself in a few years as the state of the art in domains like vision, natural language processing, games, language translation, dialogue, multi-sensors analysis, etc.
Major actors (GAFAs, BATs, etc) have recruited large research teams dedicated to this specific field (e.g. Google Brain in the US, Google Deepmind in the UK well known for its program AlphaGo, Facebook AI lab. in the US and in France, Baidu institute of Deep Learning in China, etc).
Understanding the basics of this domain is today mandatory for students and researchers in Data Science and Artificial Intelligence. The objective of the course is to provide such an introduction.Introduction to the basic concepts of machine learning, formal framework.
The birth of neural networks: the Perceptron and Adaline models. Basic statistical predictive models: linear regression and logistic regression.
Optimization basics: gradient and stochastic gradient methods.
Multilayer Perceptrons. Generalization properties and complexity control.
Introduction to deep learning: auto-associators and Convolutional Neural Networks.
Dealing with sequences: Recurrent Neural Networks.
Unsupervised learning: generative models, Generative Adversarial Networks and Variational Auto-Encoders.
Applications in the domains of vision, natural language processing, complex signal analysis.
Suggested readings and bibliography
- Oggetto:
Deep Learning - Goodfellow, Bengio, Courville - The MIT press
- Oggetto:
Class schedule
Lessons: dal 05/06/2018 to 26/06/2018
Notes: Le lezioni avranno luogo nei giorni: 5,6,7/6/2018, dalle 10 alle 12 in aula F. 12/6, dalle 10 alle 12 in aula F.
20,21,22,25,26/6/2018, dalle 10 alle 12 in aula D.
26/6/2018, dalle 10 alle 12 in sala riunioni.- Oggetto:
Note
Il corso sarà condiviso col Prof. Patrick Gallinari (Sorbonne Université, Parigi). Sarà pertanto prevalentemente in lingua inglese.
- Oggetto:
