Ahmed MAZARI
Deep Learning Researcher - Geometric Deep Learning for Computational Fluid Dynamics (CFD) at Extrality- Claim this Profile
Click to upgrade to our gold package
for the full feature experience.
-
Kabyle Native or bilingual proficiency
-
English Full professional proficiency
-
French Native or bilingual proficiency
-
Berber Native or bilingual proficiency
-
German Elementary proficiency
-
Arabic Native or bilingual proficiency
Topline Score
Bio
Credentials
-
Introduction to Programming with MATLAB Score : 94.3 %
Coursera Verified CertificatesSep, 2015- Nov, 2024 -
Introduction to Mathematical Philosophy Score : 83.3 %
Coursera.org - Ludwig-Maximilians-Universität München (LMU)Jul, 2015- Nov, 2024 -
Synapses, Neurons and Brains score : 90.9%
Coursera Verified CertificatesJul, 2015- Nov, 2024 -
Computational Neuroscience score : 94.2 %
Coursera Verified CertificatesJun, 2015- Nov, 2024 -
Søren Kierkegaard - Subjectivity, Irony and the Crisis of Modernity score : 82%
Coursera Verified CertificatesMay, 2015- Nov, 2024 -
Philosophy and the Sciences Score: 83.1%
Coursera Verified CertificatesJan, 2015- Nov, 2024 -
Introduction to Philosophy score : 88,6%
Coursera Verified CertificatesNov, 2014- Nov, 2024 -
Learning How to Learn: Powerful mental tools to help you master tough subjects
Coursera Verified CertificatesNov, 2014- Nov, 2024 -
SciWrite Writing in the Sciences score: 77%
Stanford UniversityNov, 2014- Nov, 2024 -
Statistical Inference score : 97,2%
Coursera Verified CertificatesAug, 2014- Nov, 2024 -
R Programming
CourseraJul, 2014- Nov, 2024 -
The Data Scientist’s Toolbox
CourseraJul, 2014- Nov, 2024 -
Machine Learning
CourseraMay, 2014- Nov, 2024 -
Artificial Intelligence Planning
CourseraMar, 2014- Nov, 2024 -
Aléatoire : une Introduction aux Probabilités
CourseraFeb, 2014- Nov, 2024 -
Artificial Intelligence for Robotics
UdacityOct, 2013- Nov, 2024 -
Intro to Artificial Intelligence
UdacityApr, 2013- Nov, 2024
Experience
-
Extrality
-
France
-
Software Development
-
1 - 100 Employee
-
Deep Learning Researcher - Geometric Deep Learning for Computational Fluid Dynamics (CFD)
-
Oct 2020 - Present
Geometric Deep Learning, Mesh / Graph Neural Networks, large scale graph mesh datasets,Equivariant Neural Networks, Statistical Signal Processing, Fluid Mechanics, Turbulence Modelling, Navier-Stokes equations, Partial Differential Equations (PDEs), Multi-Scale Representations, Neural Operators, Hybrid Simulation, Uncertainty Quantification, Geometry Morphing Geometric Deep Learning, Mesh / Graph Neural Networks, large scale graph mesh datasets,Equivariant Neural Networks, Statistical Signal Processing, Fluid Mechanics, Turbulence Modelling, Navier-Stokes equations, Partial Differential Equations (PDEs), Multi-Scale Representations, Neural Operators, Hybrid Simulation, Uncertainty Quantification, Geometry Morphing
-
-
-
LIP6 - Laboratoire d'Informatique Sorbonne Université/CNRS
-
France
-
Research Services
-
1 - 100 Employee
-
PHD Student. Graph Convolutional Neural Networks and Multiple Kernel Learning for Action Recognition
-
Oct 2017 - Sep 2020
Deep Video Representations, Multiple Aggregation Learning, Hierarchical Pooling, Graph Construction, Pooling and Convolution on Graphs, Geometric Deep Learning, Statistical Graph Signal Processing. Three years scholarship funded by EDITE ( Ecole Doctorale Informatique, Télécommunications et Electronique ) of Paris Laboratory : LIP6-CNRS Team : MLIA Google Scholar : https://scholar.google.com/citations?user=KkPlX-EAAAAJ&hl=fr&authuser=2&citsig=AMD79opMGG0eGsf3z2kEWD8o-KS9uAxRlA Show less
-
-
-
Conciliator - Dhatim
-
France
-
Software Development
-
1 - 100 Employee
-
Convolutional Recurrent Neural Network & Markov Models for Optical Character Recognition
-
Mar 2017 - Sep 2017
The resurgence of deep neural networks have lead to major breakthroughs and successes mainly in computer vision and natural language processing. In the context of optical character recognition (OCR), there is a considerable growth of document images (e.g scanned invoices) which makes their manual annotation and analysis out of reach. Thus, there is a need to come up with a reliable automatic solution to be able to annotate and extract detailed information in a reasonable time. Scanned documents are usually noisy and their resolution is variable. When the quality of document is poor some characters are not well extracted. The presence of noise and blur make the extraction more challenging. In this internship, we evaluated OCR engines and benchmarked OCR algorithms : Markov models, recurrent and deep neural networks. The existing system that Dhatim uses is based on bi-gram model. It shows poor performance : up to about 45% accuracy. We then proposed an end-to-end trainable neural network which is convolutional recurrent neural network (CRNN). It predicts sequence labels without any pre-segmented inputs or post-process outputs. This deep neural network consists of : convolutional neural network (CNN) which takes the input images, 2 bidirectional long short term memory (Bi-LSTM) and a connectionist temporal classification (CTC) layer. We obtained 95.69% accuracy at a sequence level prediction and 98.41% at a character level prediction. Key words : Convolutional Neural Network, bidirectional Recurrent Neural Network, Connectionist Temporal Classification, Transfer Learning, Markov models Show less
-
-
-
Inria
-
France
-
Research Services
-
700 & Above Employee
-
Research Internship on Deep Boltzmann Machine
-
Jun 2016 - Sep 2016
Initiation to research : Toward a hierarchical restricted boltzmann machine. Key words : contrastive divergence, gibbs sampling, deep learning, hierarchical models https://tao.lri.fr/tiki-index.php Initiation to research : Toward a hierarchical restricted boltzmann machine. Key words : contrastive divergence, gibbs sampling, deep learning, hierarchical models https://tao.lri.fr/tiki-index.php
-
-
-
Applied mathematics laboratory university of Bejaia-Algeria
-
University of Bejaia-Algeria
-
An optimal indulgent consensus protocol for asynchronous distributed systems
-
Jan 2015 - Jun 2015
This master thesis in computer science by research in the area of distributed systems and networking presents works and recent results obtained in the domain of asynchronous distributed systems based upon consensus problem. The study aims to propose an optimal indulgent consensus protocol with efficient communication complexity. An indulgent algorithm is a distributed algorithm that tolerates asynchronous periods when failure detectors are unreliable. In our work, we introduce a new indulgent consensus protocol that optimally solves the problem of consensus. In order to achieve efficiency in term of the number of exchanged messages and rounds needed, the developed protocol utilizes (t + 1) coordinators to send messages in each round. The aforementioned protocol guaranties that at least one process decides at (t + 2) round with a complexity in number of exchanged messages of O(n.t) and O(t) rounds at most (t + 3). t is the maximum number of failures that a system can tolerates. However, the quest for an optimal indulgent algorithm regarding "early-deciding" and "early-stopping" is still open. We conclude with the following open problem : Can we find an optimal indulgent consensus algorithm which takes into consideration the above mentioned constraints with O(n.t) exchanged messages and O(t) rounds ? Key words : Failure detectors, asynchronous distributed systems, computational complexity theory, synchronization, fault tolerance, consensus, parallel and distributed computing, routing protocols Show less
-
-
Education
-
2017 - 2020
Pierre and Marie Curie University
Doctor of Philosophy - PhD, Graph Convolutional Neural Networks and Multiple Kernel Learning for Action Recognition in Videos -
2016 - 2017
Paris-Sud University (Paris XI)
Master’s Degree, Artificial Intelligence -
2015 - 2016
Université Paris Descartes
Master’s Degree, Artificial Intelligence and machine learning for data science -
2010 - 2015
University of Bejaia - Algeria
Master’s Degree, Distributed computing and networking
Community
