Hossein Azizpour

Postdoctoral Researcher at Science for Life Laboratory

Google Scholar
Science for Life Laboratory, Gamma 6, Solna, Sweden
+46 (8) 790 6353

Note. I will start as an assistant professor in machine learning at KTH from January 2018 and will have a few master thesis projects available in deep learning and computer vision. If you are enthusiastic about deep learning, have experience with its frameworks and have good grades, feel free to contact me.

Bio. I am currently working in Dr. Kevin Smith group at Science for Life Laboratory as a postdoctoral researcher. My research is at the moment focused on two exciting areas. First and foremost, during my postdoc, I have been getting familiar to many influential medical and biomedical applications of machine learning. Thanks to the increasing availability of digitized medical data, vast interest from MDs and specialists, and maturity of many recognition algorithms, I believe it is the right time and place for the application of the machine learning tools towards a better and healthier life. Particularly, in collaboration with MD specialist researchers from Karolinska Institutet, I have been conducting research using deep learning for diagnosis and prognosis of breast cancer using the mammography and histopathological images. A second important aspect of my research lies in the general understanding and improvement of deep learning methods.

I have done my Ph.D. studies! with the honor of doing research under the supervision of Prof. Stefan Carlsson. During some periods in 2010 and 2011 I was lucky to have the opportunity of working with Dr. Ivan Laptev at INRIA-WILLOW project located in Paris!. I've also spent 3 months in Visual Geometry Group (VGG) at University of Oxford in 2013 and another 3 months at Google Research in 2016. I've got a M.Sc. degree in System, Control and Robotics from KTH School of Electrial Engineering and B.Sc. from Amirkabir University of Technology (Tehran Polytechnic) in Software Engineering!. Every now and then I've been asked to act as a reviewer! for International Journal of Computer Vision (IJCV), IEEE transactions on Pattern Recognition and Machine Intelligence (TPAMI), IEEE Robotics and Automation Letters (RA-L), Transactions on Image Processing (TIP), Transactions on Neural Networks (TNNLS), CVPR, ICCV, ECCV, ACCV, BMVC, and MICCAI.

I was born in Mashhad, Iran. My hobbies are travelling, discussion!, watching movies, reading books, and taking programming contests.


I am generally interested in machine learning and computer vision. Currently, the focus of my research is on deep learning, understanding its learning capabilities as well as interpreting its predictions. As an interesting and unique application field, I am currently working on using deep learning in the exciting field of medical image analysis.
In the past I have worked on methods for feature selection and sharing for various visual recognition tasks. In particular, the focus of my Ph.D. was on learning representation using latent SVM formulation as well as deep networks.


The Preimage of Rectifier Network Activities
Stefan Carlsson Hossein Azizpour, Ali Sharif Razavian, Josephine Sullivan, Kevin Smith,
in International Conference on Learning Representation (ICLR) Workshop Track 2017   
Materials: PDF
Factors of Transferability for a Generic ConvNet Representation
Hossein Azizpour, Ali Sharif Razavian, Josephine Sullivan, Atsuto Maki, Stefan Carlsson
in IEEE Transactions on Pattern Analysis and Machine Intelligence (PAMI) 2016   
Materials: PDF
Spotlight the Negatives: A Generalized Discriminative Latent Model
Hossein Azizpour, Mostafa Arefiyan, Sobhan Naderi Parizi, Stefan Carlsson
in BMVC 2015, Swansea, UK   
Materials: PDF Poster PDF
From Generic to Specific Deep Representation for Visual Recognition [Best Paper Award]
Hossein Azizpour, Ali Sharif Razavian, Josephine Sullivan, Atsuto Maki, Stefan Carlsson
in CVPRW 2015, DeepVision workshop, Boston, US   
Materials: PDF Poster PDF
Persistent Evidence of Local Image Properties in Generic ConvNets
Ali Sharif Razavian, Hossein Azizpour, Josephine Sullivan, Atsuto Maki, Carl Henrik Ek, Stefan Carlsson
in SCIA 2015, Copenhagen, Denmark   
Materials: PDF
Self-tuned Visual Subclass Learning with Shared Samples An Incremental Approach
Hossein Azizpour, Stefan Carlsson   
arXiv pre-print

Materials: PDF
CNN Features off-the-shelf: An Astounding Baseline for Recognition [Best Paper Runner-up Award]
Ali S Razavian Hossein Azizpour, Josephine Sullivan, Stefan Carlsson   
in CVPRW 2014, DeepVision workshop, Columbus, US   
Materials: PDF   Project Page     arXiv Tech Report  
Multi-view Body Part Recognition with Random Forests   [Best Industry Paper Award]
Vahid Kazemi, Magnus Burenius, Hossein Azizpour, Josephine Sullivan,   
in BMVC 2013, Bristol, UK
Materials:   Project and Data   PDF
Object Detection Using Strongly-Supervised Deformable Part Models
Hossein Azizpour, Ivan Laptev.   
in ECCV 2012, Florence, Italy
Materials:   Project and Data   Code    PDF
Mixture Component Identification and Learning for Visual Recognition
Omid Aghazadeh, Hossein Azizpour, Josephine Sullivan, Stefan Carlsson.
in ECCV 2012, Florence, Italy
Materials: PDF

Courses and Lectures

  • Search Engines and Information Retrieval Systems, DD2476, KTH, 2017, as 1-time Guest Lecturer on image retrieval
  • Deep Learning in Data Science, DD2424, KTH, 2017, as 1-time Guest Lecturer on generative adversarial networks
  • Image Based Recognition and Classification, DD2427, KTH, 2017, as Course Responsible and Teacher
  • Bigger Advanced individual course in computer science, DD2464, KTH, 2016, as Project Supervisor
  • Machine Learning Seminars, Linkoping University, 2015, as Invited Speaker on deep learning
  • Computer Vision Reading Group, 2011-2015, as Organizer, main presenter
  • Training for ACM International Collegiate Programming Contest, Azad Parand University, 2008, as Course Responsible
  • Design and Implementation of Algorithms, Amirkabir University, 2007, as Lecturer

Selected Awards and Honors

  • ranked 1st for the position of assistant professor in machine learning, KTH, 2017
  • top-10 highest-cited computer vision paper on arXiv in last 5 years, [Google Scholar] 2017
  • Invited to give a keynote at the Swedish Deep Learning Symposium (SSDL) 2017
  • "Strongly Exceeded Expectation" ranking at Google Research Mountain View, 2016
  • Best Paper Award from DeepVision workshop in CVPR 2015
  • Google Travel Grant for CVPR 2015 DeepVision workshop
  • Best Paper Runner-Up Award from DeepVision workshop in CVPR 2014
  • Best Industry Paper Prize from BMVC 2013
  • Outstanding reviewers' award for CVPR 2013
  • Granted Scholarship at KTH for an amount of 50000 SEK in 2010
  • Two times ACM ICPC World Finalist 2007 in Tokyo, Japan and 2008 in Banff, Canada

Institutes visited for research and collaboration

Some activities and hobbies

WARNING: The information published herein (except for the codes) is for the personal use of the reader. Copyright and all rights therein are retained by authors or by other copyright holders. Making copies of the information in full or any portion for purposes other than own use is a violation of copyright law.