Course

Deep Neural Networks (ELE680)

In this course, you will be introduced to the foundations of deep learning, the most effective and common network structures and how to build, train and evaluate deep neural networks for different applications.


Dette er emnebeskrivelsen for studieåret 2023-2024. Merk at det kan komme endringer.

See course description and exam/assesment information for this semester (2024-2025)

Semesters

Fakta

Emnekode

ELE680

Vekting (stp)

5

Semester undervisningsstart

Autumn

Undervisningsspråk

English

Antall semestre

1

Vurderingssemester

Autumn

Timeplan

Vis timeplan

Litteratur

Pensumlisten finner du i Leganto

Content

In this course, you will be introduced to the foundations of deep learning, the most effective and common network structures and their applications and how to build, train and evaluate deep neural networks for different applications. This includes:

  • Neurons, layers, back propagation, optimizers, loss functions, hyperparameters
  • Unsupervised, supervised and semi-supervised learning approaches.
  • Transfer learning
  • Multilayer Perceptron Network (MPN)
  • Convolutional Neural Network (CNN)
  • Time Series analysis
  • Reccurent Neural Network (RNN) and Long Short-Term Memory networks (LSTMs)
  • Autoencoders

  • Natural language processing (NLP)

    • Natural Language Understanding (NLU) and Embeddings.
  • Image classification and Object detection
  • Video Activity recognition
  • Deep Reinforcement Learning
  • Deep learning in image reconstruction and medical imaging.

Learning outcome

At the end of this course the student should have a fundamental understanding of methods and deep neural network structures commonly used in deep learning. The student should also be able to build, train and evaluate models for one or more practical deep learning problems.

Forkunnskapskrav

Ingen

Anbefalte forkunnskaper

Machine Learning (ELE520)

Exam

Form of assessment Weight Duration Marks Aid Exam system Withdrawal deadline Exam date
Project assessement 1/1 Letter grades — — —


The assigned project is carried out in groups of two students. Exceptionally, it can be one or three students per group. The report describes and documents work in the project. The report is made in collaboration with all the participants in the group and all participants will get the same grade. An oral presentation of the project is a mandatory part of the project.

There is no resit exam in this course. A new project report must be submitted the next time the course is taught.

Vilkår for å gå opp til eksamen/vurdering

2 assignments
2 of 2 assignments need to be approved by course instructor within the specified deadlines.

Fagperson(er)

Course teacher:

Trygve Christian Eftestøl

Course teacher:

Kjersti Engan

Head of Department:

Tom Ryen

Course teacher:

Øyvind Meinich-Bache

Course coordinator:

Øyvind Meinich-Bache

Course teacher:

Vinay Jayarama Setty

Method of work

The course has a duration of approximately 8 weeks and will be completed mid October. Lectures will be held the first 5 weeks. The students are expected to spend additional 6-8 hours a week on self-study and assignments.

The project will be carried out in the last three weeks of the course and it is expected that each student spend about 15 hours per week.

Åpent for

Data Science - Master of Science Degree Programme Computational Engineering - Master of Science Degree Programme Computer Science - Master of Science Degree Programme Computer Science - Master of Science Degree Programme, Part-Time Cybernetics and Applied AI - Master of Science Degree Programme
Exchange programme at Faculty of Science and Technology

Emneevaluering

There must be an early dialogue between the course supervisor, the student union representative and the students. The purpose is feedback from the students for changes and adjustments in the course for the current semester.In addition, a digital subject evaluation must be carried out at least every three years. Its purpose is to gather the students experiences with the course.

Litteratur

Book Deep learning Goodfellow, Ian, Bengio, Yoshua; Courville, Aaron, Cambridge, Mass., MIT Press, XXII, 775 s., cop. 2016, isbn:978-0-262-03561-3, For most topics it includes more details than required by the course. Relevant chapters and sub-chapters: 5.1-5.3, 5.7-5.11, 6.1-6,6, 7.4, 7.6, 7.8, 7.12, 8.1-8.5, 9.1-9.11, 10.1-10.12, 11.1-11.4, 11.6, 12.1-12.5, 14.1-14.9, 15.2. Article Robust Object Tracking with Online Multiple Instance Learning Boris Babenko et al., IEEE Transactions on Pattern Analysis and Machine Intelligence, 8, 33, 2011, Weak and Strong labels Article Not-so-supervised: A survey of semi-supervised, multi-instance, and transfer learning in medical image analysis Veronika Cheplygina et al., Medical Image Analysis, 54, 2019, 280-296, Weak and Strong labels https://doi.org/10.1016/j.media.2019.03.009View online Article A Framework for Multiple -Instance Learning Oded Maron et al., Weak and Strong labels Article Artificial intelligence in computational pathology – challenges and future directions Sandra Morales et al., Digital Signal Processing, 119, 2021, Weak and Strong labels https://doi.org/10.1016/j.dsp.2021.103196View online Article Weakly Supervised Object Localization with Multi-Fold Multiple Instance Learning Ramazan Gokberk Cinbis et al., IEEE Transactions on Pattern Analysis and Machine Intelligence, 1, 39, 2017, Weak and Strong labels Article A brief introduction to weakly supervised learning Zhi-Hua Zhou et al., National Science Review, 5, 2018, Weak and Strong labels https://academic.oup.com/nsr/article/5/1/44/4093912https://doi.org/10.1093/nsr/nwx106View online Article Quo Vadis, Action Recognition? A New Model and the Kinetics Dataset Joao Carreira et al., Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2017, Video activity recognition https://openaccess.thecvf.com/content_cvpr_2017/html/Carreira_Quo_Vadis_Action_CVPR_2017_paper.htmlView online Article A deep learning reconstruction framework for X-ray computed tomography with incomplete data Jianbing Dong et al., PLoS One, 11, 14, 2019, Image Reconstruction and Medical Imaging https://doi.org/10.1371/journal.pone.0224426View online Article Deep learning based image reconstruction algorithm for limited-angle translational computed tomography Jiaxi Wang et al., PLoS One, 1, 15, 2020, Image Reconstruction and Medical Imaging Article Deep learning for biomedical image reconstruction: a survey Hanene Ben Yedder et al., The Artificial Intelligence Review, 1, 54, 2021, Image Reconstruction and Medical Imaging

Kontakt

Course teacher:

Trygve Christian Eftestøl

Course teacher:

Kjersti Engan

Head of Department:

Tom Ryen

Course teacher:

Øyvind Meinich-Bache

Course coordinator:

Øyvind Meinich-Bache

Course teacher:

Vinay Jayarama Setty

Tilbys av

Faculty of Science and Technology

Department of Electrical Engineering and Computer Science

The course description is retrieved from FS (Felles studentsystem). Version 1