Muberra Ozmen

PhD candidate at McGill

McGill University

Biography

Hi!

I am Muberra from Turkey, living in Canada since 2019. I am doing PhD at McGill Computer Networks Lab under supervision of Prof. Mark Coates. My research intersects (relational) graph representation learning and multi-label classification. I develop new techniques for modelling label dependencies in multi-label classification problems by exploiting expressive power of graphs.

Download my resumé.

Interests

Machine learning on graphs
Multi-label classification
Probabilistic graphical models

Education

PhD in Computer Science

McGill University
MRes in School of Management, 2017

University College London
MSc in Operations Research, 2016

Middle East Technical University
BSc in Industrial Engineering, 2014

Middle East Technical University

Programming skills

R

Python

+PyTorch +Tensorflow

Matlab

Experience

Teaching Asistant

McGill University

Jan 2019 – Present Montreal, Canada

ECSE 316 - Signals and Networks

Data Scientist

Segmentify

Jan 2018 – Dec 2018 Istanbul, Turkey

Within the scope of the project that I worked on, by monitoring the current progress of the literature on techniques developed for estimation of Customer Purchasing Behaviour and Lifetime Value in database marketing, we aimed to provide a comprehensive, yet practical, framework for customer segmentation to be able to bring related advancements into our own practice.

Teaching and Research Assistant

Middle East Technical University

Jun 2014 – Aug 2016 Ankara, Turkey

Data Mining - Analysis and Optimization Methods In Finance - Stochastic Models in Operations Research - Decision Analysis

Undergraduate Internships & Volunteer Work

Young Guru Academy, IsBank, BOSCH, Mediterranean Games

Sep 2009 – Jun 2014

I was a member of YGA for three years and worked as the project leader of “Blind Leaders” project in which weekly sessions were organized in primary school for the blind. I performed an internship at IsBank to collect data for my team graduation project “Cash Management System Design for a Major Bank in Turkey” which received several awards. I have conducted project “Analysing the Work Design at Coating Processes of BOSCH Diesel Systems” during a long term internship at the company, which involved using simple statistical models for estimation of system parameters, and performing process optimization accordingly.

Featured Publications

Multi-relation Message Passing for Multi-label Text Classification

an encoder-decoder type architecture to model multiple bi-directional relationships between labels for multi-label classification problem

Microwave Radar for Breast Health Monitoring: System Performance Protocol

a system performance protocol for routine use in clinical trials to ensure proper system functioning for breast cancer detection

Interactive Evolutionary Approaches to Multi-objective Feature Selection

interactive evolutionary algorithms for multi-objective feature selection problems that aim to converge the most preferred solution by guiding the search toward the regions that consists of appealing solutions for the DM