candidacy exam
Event details
Date | 01.03.2018 |
Hour | 14:00 › 16:00 |
Speaker | Rabeeh Karimi |
Location | |
Category | Conferences - Seminars |
EDIC candidacy exam
Exam president: Prof. Pascal Frossard
Thesis advisor: Prof. Volkan Cevher
Co-examiner: Prof. Alexandre Alahi
Abstract
Abstract—Nowadays deep learning methods have achieved
state-of-the-art performance in many fields, including signal
processing. Such data-driven methods are learning the signal
structure from large training data, neglecting conventional signal
recovery approaches. It is worth investigating how infusing
conventional signal processing methods into deep learning models
could benefit current models with the hope to solve a wider range
of problems, and increasing the performance and accuracy of
deep learning architectures. Towards this goal, we explain briefly
how to jointly train sampling pattern and reconstruction network
for signal recovery task given compressive measurements.
Background papers
Learning-Based Compressive Subsampling, by Luca Baldassarre et all.
Learning to invert: Signal recovery via deep convolutional networks, by Ali. Mousavi and Richard G.Baraniuk
https://arxiv.org/pdf/1701.03891.pdf
Learning representations from EGG with deep recurrent convolutional neural networks, by
Pouya Bashivan et all.
Exam president: Prof. Pascal Frossard
Thesis advisor: Prof. Volkan Cevher
Co-examiner: Prof. Alexandre Alahi
Abstract
Abstract—Nowadays deep learning methods have achieved
state-of-the-art performance in many fields, including signal
processing. Such data-driven methods are learning the signal
structure from large training data, neglecting conventional signal
recovery approaches. It is worth investigating how infusing
conventional signal processing methods into deep learning models
could benefit current models with the hope to solve a wider range
of problems, and increasing the performance and accuracy of
deep learning architectures. Towards this goal, we explain briefly
how to jointly train sampling pattern and reconstruction network
for signal recovery task given compressive measurements.
Background papers
Learning-Based Compressive Subsampling, by Luca Baldassarre et all.
Learning to invert: Signal recovery via deep convolutional networks, by Ali. Mousavi and Richard G.Baraniuk
https://arxiv.org/pdf/1701.03891.pdf
Learning representations from EGG with deep recurrent convolutional neural networks, by
Pouya Bashivan et all.
Practical information
- General public
- Free
Contact
- EDIC - [email protected]