Scribd
Upload
82 views2 pages

Neural Networks and Deep Learning: Rationale

This document provides an overview of neural networks and deep learning. It discusses how deep learning uses multilevel data representations in artificial neural networks to learn hierarchical features from data. The course will cover underlying theory, applications in areas like computer vision and natural language processing, and training various deep learning architectures like convolutional and recurrent neural networks. Students will be assessed on assignments applying backpropagation, convolutional networks, and developing large-scale deep learning systems.

Uploaded by

Jorge Guerrero
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
82 views2 pages

Neural Networks and Deep Learning: Rationale

This document provides an overview of neural networks and deep learning. It discusses how deep learning uses multilevel data representations in artificial neural networks to learn hierarchical features from data. The course will cover underlying theory, applications in areas like computer vision and natural language processing, and training various deep learning architectures like convolutional and recurrent neural networks. Students will be assessed on assignments applying backpropagation, convolutional networks, and developing large-scale deep learning systems.

Uploaded by

Jorge Guerrero
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 2

Neural Networks and Deep Learning

Rationale
With the recent boom in artificial intelligence, more specifically, Deep Learning and its underlying
Neural Networks, are essential part of systems that must perform recognition, make decisions and
operate machinery.
Deep Learning in Artificial Neural Networks (ANN) is relevant for Supervised, Unsupervised, and
Reinforcement Learning. This course will provide a thorough examination of the state-of-the-art
and will present the mathematical and algorithmic foundations of Deep Learning in ANN.

Description
Deep Learning concerns multilevel data representation with every level providing hierarchical
explanations of data. These algorithms have proven effective in discovering the structure of data
and the features that most differentiate among classes.
Artificial Neural Networks are created to mimic the human brain. In the process, various
architectures of neurons and connections have been created. Generally, the artificial neurons are
activated through input signals and the connections among them are learned through various
algorithms.
The course will cover the underlying theory, the range of applications, and learning from very
large data sets. Connectionist architectures commonly associated with deep learning, e.g., basic
neural networks, convolutional neural networks and recurrent neural networks will also be
discussed. Methods to train and optimize the architectures and methods to perform effective
inference with them, will be the main focus.

Topic List
Topics may include but are not limited to:
 Deep Learning
o Depth of deep learning
o Overview of methods
o Linear algebra and numerical methods

 Neural Networks
o Feed-forward architecture
o Gradient-based learning

 Deep Neural Networks


o Architecture design
o Computational graphs
o Backpropagation
 Regularization
o Overview
o Multitask learning
o Parameter sharing

 Convolutional Neural Networks


o Operation and algorithm
o Pooling

 Recurrent and Recursive Networks


o Unfolding graphs
o Recurrent neural networks

 Applications and Research


o Large scale systems
o Computer vision
o Natural language processing
o Boltzmann machines

Text Sources
The main textbook for the course is Ian Goodfellow, Yoshua Bengio and Aaron Courville, Deep
Learning, MIT Press, 2016 (available as free html book for reading at
http://www.deeplearningbook.org/) as well as research papers from the field.

Learning Goals
The student must be able to demonstrate a working knowledge of the foundations of deep
learning in neural networks represented by:
 Deep learning backpropagation
 Convolutional neural networks
 Recurrent networks
 Computer vision large scale system processing

Assessment
Assignments and projects will be given to test the students’ abilities in the design and development
of deep learning systems based on neural networks. The students should be able to experiment and
discuss the performance of such systems and evaluate the applicability of the algorithms and neural
network architectures. Evaluation criteria: individual assignments – 50 %, term project – 40%,
participation – 10%.

You might also like