Mohammad Sabokrou

Machine Learning Researcher

Research Intrests


  • Computer Vision

  • Machine Learning

  • Adverserial Learning

  • Self/unuspervised Learning

  • XAI

  • AI Safety

I am currently a senior researcher with the Institute for Research in Fundamental Sciences(IPM), working at the intersection of machine learning and computer vision. My research interests include machine learning (deep learning and outlier detection) and computer vision (crowded scene analysis and activity recognition).
I am always looking for motivated students to join IPM AI group. If you are interested, please contact me!
  1. News!

  2. One Paper accepted to WACV2021

Publications

Click Here to See the Complete list of publications

Journal
Driver behavior detection and classification using deep convolutional neural networks M Shahverdy, M Fathy, R Berangi, M Sabokrou Expert Systems with Applications 149, 113240
Conference
G2D: Generate to Detect Anomalies M Pourreza, B Mohammadi, M Khaki, S Bouindour, H Snoussi, Mohammad Sabokrou WACV2020
Journal
Deep End-to-End One-Class Classifier M Sabokrou, M Fathy, G Zhao, E Adeli IEEE Transactions on Neural Networks and Learning Systems
Pre-print
Deep-hr: Fast heart rate estimation from face video under realistic conditions M Sabokrou, M Pourreza, X Li, M Fathy, G Zhao arXiv preprint arXiv:2002.04821
Pre-print
AutoIDS: Auto-encoder Based Method for Intrusion Detection System M Gharib, B Mohammadi, SH Dastgerdi, M Sabokrou arXiv preprint arXiv:1911.03306
Conference
End-to-end adversarial learning for intrusion detection in computer networks B Mohammadi, M Sabokrou 2019 IEEE 44th Conference on Local Computer Networks (LCN), 270-273
Conference
Generative Adversarial Irregularity Detection in Mammography Images M Ahmadi, M Sabokrou, M Fathy, R Berangi, E Adeli International Workshop on PRedictive Intelligence In MEdicine, 94-104
Conference
Unsupervised feature ranking and selection based on autoencoders S Sharifipour, H Fayyazi, M Sabokrou, E Adeli ICASSP 2019-2019
Conference
Self-supervised representation learning via neighborhood-relational encoding M Sabokrou, M Khalooei, E Adeli Proceedings of the IEEE International Conference on Computer Vision, 8010-8019
Conference
Sub-word based Persian OCR Using Auto-Encoder Features and Cascade Classifier M Pourreza, R Derakhshan, H Fayyazi, M Sabokrou 2018 9th International Symposium on Telecommunications (IST), 481-485
Conference
Avid: Adversarial visual irregularity detection M Sabokrou, M Pourreza, M Fayyaz, R Entezari, M Fathy, J Gall, E Adeli Asian Conference on Computer Vision, 488-505
Journal
Deep-anomaly: Fully convolutional neural network for fast anomaly detection in crowded scenes M Sabokrou, M Fayyaz, M Fathy, Z Moayed, R Klette Computer Vision and Image Understanding 172, 88-97
Journal
Semantic video segmentation: A review on recent approaches MH Saffar, M Fayyaz, M Sabokrou, M Fathy arXiv preprint arXiv:1806.06172
Journal
Towards principled design of deep convolutional networks: introducing SimpNet SH Hasanpour, M Rouhani, M Fayyaz, M Sabokrou, E Adeli arXiv preprint arXiv:1802.06205
Conference
Adversarially learned one-class classifier for novelty detection M Sabokrou, M Khalooei, M Fathy, E Adeli Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR2018)
Journal
Fast and accurate detection and localization of abnormal behavior in crowded scenes M Sabokrou, M Fathy, Z Moayed, R Klette Machine Vision and Applications 28 (8), 965-985
Journal
Deep-cascade: Cascading 3d deep neural networks for fast anomaly detection and localization in crowded scenes M Sabokrou, M Fayyaz, M Fathy, R Klette IEEE Transactions on Image Processing 26 (4), 1992-2004
Conference
STFCN: spatio-temporal fully convolutional neural network for semantic segmentation of street scenes M Fayyaz, MH Saffar, M Sabokrou, M Fathy, F Huang, R Klette Asian Conference on Computer Vision, 493-509
Conference
STFCN: spatio-temporal FCN for semantic video segmentation M Fayyaz, MH Saffar, M Sabokrou, M Fathy, R Klette, F Huang arXiv preprint arXiv:1608.05971
Journal
Video anomaly detection and localisation based on the sparsity and reconstruction error of auto-encoder M Sabokrou, M Fathy, M Hoseini Electronics Letters 52 (13), 1122-1124
Conference
A novel approach for Finger Vein verification based on self-taught learning M Fayyaz, M Hajizadeh-Saffar, M Sabokrou, M Hoseini, M Fathy 2015 9th Iranian Conference on Machine Vision and Image Processing (MVIP), 88-91
Conference
Online signature verification based on feature representation M Fayyaz, MH Saffar, M Sabokrou, M Hoseini, M Fathy 2015 The International Symposium on Artificial Intelligence and Signal Processing
Conference
Real-time anomaly detection and localization in crowded scenes M Sabokrou, M Fathy, M Hoseini, R Klette Proceedings of the IEEE conference on computer vision and pattern recognition workshop
JOURNAL
Abnormal Event Detection and Localization in a Video based on Similarity Structure M FATHI, M Sabokrou, M Hosseini MODARES JOURNAL OF ELECTRICAL ENGINEERING 14 (3), 16-23
conference
Mobile target tracking in non-overlapping wireless visual sensor Networks using Neural Networks M Sabokrou, M Fathy, M Hosseini ICCKE 2013, 309-314
conference
Solving coverage problem in three-dimensional area using Wireless Visual Sensor Networks M Sabokrou, M Fathy, M Fallhhosseni, M Hoseini Proceedings of 2012 2nd International Conference on Computer Science
conference
An evolvable fuzzy logic system for handoff managementin heterogeneous wireless networks H Fayyazi, M Sabokrou 2012 2nd International eConference on Computer and Knowledge Engineering
conference
Intelligent target tracking in Wireless Visual Sensor Networks M Sabokrou, M Fathy, M Hoseni 2012 2nd International eConference on Computer and Knowledge Engineering
conference
Region-based multi-spectral image segmentation using Evolutionary Strategies M Sabokrou, H Fayyazi, M Hosseini, N Fallahi Proceedings of 2011 International Conference on Computer Science and Network
conference
Solving heterogeneous coverage problem in Wireless Multimedia Sensor Networks in a dynamic environment using Evolutionary Strategies H Fayyazi, M Sabokrou, M Hosseini, A Sabokrou 2011 1st International eConference on Computer and Knowledge Engineering

Top research project details

Adversarially Learned One-Class Classifier for Novelty Detection
M. Sabokrou, M. Khalooei, M. Fathy, E. Adeli

One-class classification (OCC) poses as an essential component in many machine learning and computer vision applications, including novelty, anomaly, and outlier detection systems. With a known definition for a target or normal set of data, one-class classifiers can determine if any given new sample spans within the distribution of the target class. Solving for this task in a general setting is particularly very challenging, due to the high diversity of samples from the target class and the absence of any supervising signal over the novelty (nontarget) concept, which makes designing end-to-end models unattainable.

Self-supervised representation learning via neighborhood-relational encoding

M.Sabokrou, M. Khalooei, M. Fathy, E. Adeli

In this project, we propose a novel self-supervised representation learning by taking advantage of a neighborhoodrelational encoding (NRE) among the training data. Conventional unsupervised learning methods only focused on training deep networks to understand the primitive characteristics of the visual data, mainly to be able to reconstruct the data from a latent space. They often neglected the relation among the samples, which can serve as an important metric for self-supervision. Different from the previous work, NRE aims at preserving the local neighborhood structure on the data manifold. Therefore, it is less sensitive to outliers. We integrate our NRE component with an encoder-decoder structure for learning to represent samples considering their local neighborhood information. Such discriminative and unsupervised representation learning scheme is adaptable to different computer vision tasks due to its independence from intense annotation requirements. We evaluate our proposed method for different tasks, including classification, detection, and segmentation based on the learned latent representations. In addition, we adopt the auto-encoding capability of our proposed method for applications like defense against adversarial example attacks and video anomaly detection. Results confirm the performance of our method is better or at least comparable with the state-of-the-art for each specific application, but with a generic and self-supervised approach

Read more
Email : Sabokro [at] ipm.ir