WELCOME TO MY PORTFOLIO


About Me

Hi there, I am Omar Ikne, a Master-Level Data Science Student at at Centrale Lille - University of Lille.
I’m interested in Machine Learning, Deep Learning and Artificial Intelligence.
I’m looking to collaborate on Machine Learning projects.

Some Fields of Interest

  • Machine Learning
  • Mathematics
  • Artificial Intelligence
  • Deep Learning

I am looking for a 6 months internship in Machine Learning starting in March 2022

Work Experience & Academic Projects

Here are some of the professional experiences I had during my studies:

Data Science Intern

Apr 2021 – Jul 2021

Automatic Construction of Biological Interaction Models within Marine Ecosystems using LFIT (Learning From Interpretation Transition) framework.

  • Data pre-processing: Data cleaning, Discretization of features of interest, Data augmentation.
  • Rules improvement using data augmentation and Pareto front.
  • Building an influence graph from the improved rules.
  • Publication of an extended abstract & poster session at an international conference.
Abstract

The study of the dynamics of systems is of growing interest in many scientific fields, including biology. Marine ecology models are used to study and anticipate population variations of plankton and marine microalgae species. These variations can have an impact on ecological niches, the economy or the climate. The understanding of these systems is therefore particularly interesting, but their complexity makes it difficult to create new models. LFIT (Learning From Interpretation Transition) is a machine learning framework that aims at learning the dynamics of a system by observing its state transitions. Learning these dynamics is equivalent to building a causal model of the system, i.e., a representation of what can happen in a system and why it happens. LFIT provides explainable predictions in the form of logical rules that can be processed to extract global influences between plankton species.
Keywords: Marine ecosystems . Biological networks . Learning Dynamics . Inductive logic programming . Dynamical systems . Interaction graphs .


Centre de Recherche en Informatique, Signal et Automatique de Lille, Lille, France.

Computer Science Intern

Nov 2019 – Jan 2020

Design and implementation of the computational part for an intravenous infusion simulation application using Matlab.

Center for Innovation and Technological Testing, Valenciennes, France.

Industrial Engineering Intern

Jun 2019 – Aug 2019

A two-month internship at the printing company La Manufacture d'Histoires Deux-Ponts.

Manufacture d'Histoires Deux-Ponts, Grenoble, France.

Research Project

Oct 2021 - Now

Head pose correction for facial recognition using Generative Adversarial Networks (GANs).

  • Identify new solutions proposed in the literature to address this scientific challenge.
  • Explore solutions based on generative adversarial networks (GANs).
  • Implement a new solution inspired by recent works in the literature and former works of the research team.
Abstract

Facial expressions are part of our daily lives. They are useful in various situations, such as the detection of suspicious or dangerous behaviors in public places, medical diagnosis assistance, as well as the analysis of marketing preferences or the reinforcement of interactive exchanges. Our objective is to find and develop a solution to make recognition solutions efficient in the presence of head movement by going through a face normalization phase while minimizing the geometric deformations induced in the normalization process. This would allow to come back to an ideal case of analysis to recognize expressions (i.e. static and frontal face) whatever the pose and the movement of the face.
Keywords: image processing . facial expression . deep learning (generative adversarial network).


IMT Nord Europe, Lille, France.

Research Project

Oct 2020 - Mar 2021

Harmonization of brain MR signal using Machine Learning models (Random Forest).

  • Data pre-processing.
  • Dataset construction using the ComBat tool.
  • Building the Neuroharmony tool from scratch to harmonize brain MRIs.
Abstract

Neuroimaging is no exception to the growing involvement in the application of machine learning, especially with the increasing availability of neuroimaging datasets. But one of the main limitations of these datasets is their small size. To overcome this problem, data are collected from different sites and centers. However, this introduces heterogeneity in the data due to the variability of the scanners and protocols followed during the acquisition process. There are conventional approaches to harmonize multi-scanner datasets such as Combat but this type of approach requires a representative sample for each scanner included in the dataset. Therefore, they are not suitable for machine learning models that aim to harmonize data from unknown scanners or sites. To overcome this problem, the Neuroharmony tool was introduced to harmonize single images from unknown scanners using a set of image quality measures accessible from the images and without requiring a single representative sample.
Keywords: Harmonization . ComBat . Neuroharmony . MRI . Multi-site.


CHU (Centre Hospitalier Universitaire) Lille, Lille, France.

Databases Project

Nov 2019 – Jan 2020

Implementation of an algorithm to import an XML document into an SQL relational database, and translating the largest possible fragment of XPath to equivalent SQL queries using Python.

Overview  The main steps of this project are:
  1. Proposing a relational encoding scheme for a given XML document.
  2. Implementation of the functionality of importing an XML document into a relational database using SAX API.
  3. Developing of a scheme of translating largest possible fragment of XPath to equivalent SQL queries evaluated over the encoded instance.
  4. Proposing an automated testing approach to verify that the system is working correctly.
  5. Developing a protocol for experimental comparison of the querying (time) performance of the system and an existing XML query system.

Univerity of Lille, Lille, France.

Personal Mini-Projects

2020

The implementation of some classic video games and algorithms using Python.

  • Sudoku
  • Game of life
  • Maze Generator
  • Snake

Code

Background

Here is my educational background:

Master's Degree

Sep 2020 - Now

1st and 2nd year of Master in Data Science at Centrale Lille - University of Lille

Some Interesting Courses

  • Deep Learning
  • Machine Learning
  • Natural Language Processing
  • Numerical Analysis & Optimization
  • Bayesian Machine Learning
  • Distributed and Parallel Computing/Programming
  • Privacy Preserving Machine Learning
  • Signal Processing
  • Kernel Machines
  • Reinforcement Learning

Univerity of Lille, Lille, France

Engineering Student

Sep 2018 - Sep 2019

1st year engineering student at Grenoble-INP PAGORA: International School of Paper, Printed Communication and Biomaterials.

Grenoble-INP Pagora, Grenoble, France.

CPGE

Sep 2016 - Jun 2018

2-year intensive university-level studies in Maths, Physics and Chemistry in preparation to enter highly-competitive French engineering schools.

Some Interesting Courses

  • Mathematics
  • Physics
  • Chemistry
  • Computer Science
  • Literature

CPGE Ibn Ghazi, Rabat, Morocco.

Baccalaureate Degree

Sep 2015 - Jun 2016

Baccalaureate degree at Boumalne Dades High School, Serie: Sciences, Speciality: Mathematics.

Boumalne Dades High School, Boumalne Dades, Morocco.

Publications & Conferences

Here are some of my publications and conferences:

Extended Abstarct

2021

IJCLR 2021 (1st International Joint Conference on Learning and Reasoning)

Title : Automatic Modeling of Dynamical Interactions Within Marine Ecosystems
Authors : Omar Iken, Maxime Folschette and Tony Ribeiro
Track : ILP (Inductive Logic Programming)

Abstract

article

Marine ecology models are used to study and anticipate population variations of plankton and microalgae species. These variations can have an impact on ecological niches, the economy or the climate. Our objective is the automation of the creation of such models. Learning From Interpretation Transition (LFIT) is a framework that aims at learning the dynamics of a system by observing its state transitions. LFIT provides explainable predictions in the form of logical rules. In this paper, we introduce a method that allows to extract an influence graph from a LFIT model. We also propose an heuristic to improve the model against noise in the data.
Keywords: logical modeling · dynamic systems · heuristics · interaction graph.



Poster Session

2021

IJCLR 2021 (1st International Joint Conference on Learning and Reasoning)

Title : Automatic Modeling of Dynamical Interactions Within Marine Ecosystems
Authors : Omar Iken, Maxime Folschette and Tony Ribeiro
Track : ILP (Inductive Logic Programming)

Abstract

poster

Marine ecology models are used to study and anticipate population variations of plankton and microalgae species. These variations can have an impact on ecological niches, the economy or the climate. Our objective is the automation of the creation of such models. Learning From Interpretation Transition (LFIT) is a framework that aims at learning the dynamics of a system by observing its state transitions. LFIT provides explainable predictions in the form of logical rules. In this paper, we introduce a method that allows to extract an influence graph from a LFIT model. We also propose an heuristic to improve the model against noise in the data.
Keywords: logical modeling · dynamic systems · heuristics · interaction graph.

Skills & Languages

Here are some of my IT skills and the languages I speak:

    Programming Languages

    • PythonPython

    • R R

    • MatlabMatlab

    • JavascriptJS


    Data Analysis & Machine Learning

    • PyTorchtorch

    • Keraskeras

    • Sklearnsklearn

    • Pandaspandas

    • Numpynumpy

    • Matplotlibplt


    Development & Tools

    • Gitgit

    • Jupyterjupyter

    • Latexlatex

    • Officeoffice


    Databases

    • SQLsql

    • Cyphercypher

    Operating Systems

    • Linuxlinux

    • Windows windows



    Graphical User Interface

    • Pyqt5Pyqt5

    • PygamePygame


Spoken Languages

  • English(ABC) : Advanced level
  • French(ABC) : Advanced level
  • Arabic(‫أب ت‬) : Proficiency level
  • Tamazight(ⴰⵣⵛ) : Native tongue

.

Get in Touch

Feel free to contact me !