آزمایشگاه پژوهشی یادگیری عمیق دانشگاه تهران

-for-authors.
Submissions must be sent through http://ees.elsevier.com/prletters/ .


Authors have to select the acronym "AGbR4PR" as the article type, from the "Choose Article Type" pull-down menu during the submission process. The maximal length of a paper is 7 pages in the PRLetters layout and may become
8 in the revised version if referees explicitly request significant additions. The submitted papers should not have been previously published or be under consideration for publication elsewhere. If a submission is the extended version of a conference paper, the original work should be explicitly referenced and a description of the changes should be provided.
Also, in this case, PRLetters submission should include at least 30% new material of high relevance (more experiments, proofs of theorems not included in the conference paper, more comparisons with other methods in the literature and so on); the title of the PRLetters paper should be different; the same figures cannot be used, and the parts in common between the conference paper and the extended version cannot be verbatim the same.


Guest Editors


Donatello Conte (Managing Guest Editor)
Computer Science Laboratory of Tours (LIFAT - EA 6300), University of Tours
Email: donatello.conte@univ-tours.fr

Jean-Yves Ramel
Computer Science Laboratory of Tours (LIFAT - EA 6300), University of Tours
Email: jean-yves.ramel@univ-tours.fr


Pasquale Foggia
Dip. di Ingegneria dell’Informazione ed Elettrica e Matematica Applicata, University of Salerno
Email: pfoggia@unisa.it

@ut_deep

Virtual Special Issue on Advances in Graph-based Representations for Pattern Recognition (AGbR4PR) Deadline extended (15/03/2020)



https://www.journals.elsevier.com/pattern-recognition-letters/call-for-paper
s/advances-in-graph-based-representations

Motivations

Graph-based representation and learning/inference algorithms are widely applied to structural pattern recognition, image analysis, machine learning and computer vision. Facing the multitude of scientific problems and the wide applications of graph-based representations, the IAPR TC-15 (Graph-based Representations in Pattern Recognition) promotes a series of workshops called IAPR-TC15 Workshop on Graph-based Representations in Pattern Recognition (GbR) since more than 20 years. This series of workshops has benefitted the community in triggering scientific research and exchanging progresses all along years. The 12th edition of GbR was held in Tours, France, in June 2019, and saw several original contributions linked to the actual strong interest for deep learning and artificial intelligence.


This special issue should aim to report the last advances in theory, methods and applications using graphs for pattern representation and recognition.
The scope ranges from various computing issues like combining machine learning with graphs, graph mining, graph representations of shapes, images and networks, to applications in pattern recognition, computer vision and data mining.


Topics

The topics of the Special Issue include, but are not limited to:
- Graph matching
- Graph-based image segmentation
- Machine Learning / Deep Learning on graphs
- Graph representation of shapes
- Graph-based learning and clustering
- Data mining with graphs
- Graph distance and similarity measures
- Kernel methods for graphs
- Graph embedding
- Belief-propagation methods
- Graph-cuts methods
- Graphs in computational topology and bioinformatics
- Graphs in social network analysis

Important Dates
- Submission period: until March 15, 2020

- Notification to Authors for the 1st reviewing round: April 30 2020
- 1st Revision submission: June 30 2020
- Notification to Authors for the 2nd reviewing round: August 15 2020
- 2nd Revision submission: September 30 2020
- Final Notification to Authors: November 15 2020

Authors can submit exclusively during the submission period and they should select the acronym of the special issue (AGbR4PR) as “article type” when uploading their articles.


Reviewing Process

The review process will follow the standard PR letters scheme that each paper will be reviewed by two referees. The referees will include some TC15 program committee members and other invited referees selected from the EES.


Submission Guidelines


All submissions have to be prepared according to the Guide for Authors as published in the Journal Web Site at
http://www.elsevier.com/journals/pattern-recognition-letters/0167-8655/guide

The Matrix Calculus You Need For Deep Learning

This article is an attempt to explain all the matrix calculus you need in order to understand the training of deep neural networks. We assume no math knowledge beyond what you learned in calculus 1, and provide links to help you refresh the necessary math where needed. Note that you do not need to understand this material before you start learning to train and use deep learning in practice; rather, this material is for those who are already familiar with the basics of neural networks, and wish to deepen their understanding of the underlying math.

https://explained.ai/matrix-calculus/index.html

@irandeeplearning

SinGAN

With SinGAN, you can train a generative model from a single natural image, and then generate random samples form the given image

SinGAN can be also use to a line of image manipulation task

https://arxiv.org/abs/1905.01164

@irandeeplearning