Admission requirements
Assumed prior knowledge
Elementary calculus and linear algebra; basics of probability theory and statistics. Fluency in Python; basic commands of Linux.
Description
The course provides an introduction to key concepts, algorithms and architectures for neural networks, with strong emphasis on Deep Learning and its applications. It covers the following topics:
Part One: Classical Neural Networks
Basics of Machine Learning and Pattern Recognition
Linear models: Linear Separability, Perceptron, Convergence Theorem, Cover's Theorem, Multi-Class Perceptron
Multi-layer Perceptron
Algorithms for training Deep Networks: Stochastic Gradient Descent and its variants; Backpropagation
Alternative activation and loss functions; Regularization, Dropout, Batch Normalization
Introduction to GPU-computing, Keras, TensorFlow
Part Two: DeepLearning
Convolutional Networks; key architectures and applications
Recurrent Neural Networks; LSTM and GRU Networks; Word Embeddings
Autoencoders
GAN Networks
Deep Learning for Reinforcement Learning
During the course several state-of-the-art applications of Deep Learning to image recognition, computer vision, language modeling, game playing programs, etc., will be discussed. The course consists of weekly lectures, three programming assignments (in Python, TensorFlow, Keras) and the final written exam.
Course objectives
The objective of this course are:
to provide a general introduction to the field of deep neural networks and their applications
to develop practical skills for designing and training neural networks for tasks like image classification, speech recognition, modeling sequential data, game playing
to master a popular system for developing deep networks: TensorFlow with Keras
Timetable
The most recent timetable can be found at the Computer Science (MSc) student website.
Mode of instruction
Lectures
Computer Labs
Practical Assignments
Course load
Total hours of study 6 EC course: 168 hrs.
Assessment method
The final grade will be the weighted average of grades for:
programming assignments (60%)
written exam (40%)
The teacher will inform the students how the inspection of and follow-up discussion of the exams will take place.
Reading list
- Aurélien Géron, Hands-On Machine Learning with Scikit-Learn, Keras, and TensorFlow: Concepts, Tools, and Techniques to Build Intelligent Systems (Second Edition)
Registration
You have to sign up for courses and exams (including retakes) in uSis. Check this link for information about how to register for courses.
Contact
Lecturer: dr. Wojtek Kowalczyk