George Panagopoulos

George Panagopoulos

Machine Learning Scientist

About Me

Education and Work Experience

I am a postdoctoral research scientist on machine learning at the University of Luxembourg , under the supervision of Prof. Jun Pang. I've spent one year working on operations research with Dr. Nikolaos Liakopoulos in the Algorithms and Optimization lab of Amazon Transportation Services. I received my Ph.D. in machine learning for graphs from the Data Science and Mining Team of the Computer Science Laboratory of École Polytechnique, supervised by Prof. Michalis Vazirgiannis and Assistant Prof. Fragkiskos Malliaros . Previously, I was a teaching and research assistant at the University of Houston, where I obtained my M.Sc. in computer science advised by Prof. Ioannis Pavlidis. Before that, I was working as a research associate for 2 years in the Institute of Informatics and Telecommunications of NCSR Demokritos. I hold a B.Sc. in Informatics and Telematics from Harokopio University of Athens, where I completed my diploma thesis with Assistant Prof. Iraklis Varlamis .

Scientific Interests

My interests lie in relational machine learning, causal inference, and information retrieval. More specifically I am researching machine learning methods for causal inference in large-scale heterogeneous data coming from biomedical or commercial settings. I am also exploring active and online learning methods for experimentation in clinical trials and e-commerce applications. In the recent past, I have worked on combinatorial optimization, forecasting, classification, and epidemic-spreading problems. In terms of applications, I have worked with a variety of data like social, logistics, epidemic, molecular, product, protein, coauthorship, word, and brain graphs. Finally, I am also experienced with NLP in long/short texts and digital signal processing for statistical inference from wearable sensors or neural signals.

Resume

Education

January 2019 - March 2022

PhD in Computer Science

Ecole Polytechnique

Advisor: Dr. Michalis Vazirgiannis, Dr. Fragkiskos Malliaros

August 2016 - May 2018

Masters in Computer Science

University of Houston

Advisor: Dr. Ioannis Pavlidis

September 2010 - July 2014

Bachelor in Informatics and Telematics

Harokopio University of Athens

Advisor: Dr. Iraklis Varlamis

Experience

Descember 2022--June 2023

Applied Scientist

Amazon
April 2022--October 2022

Applied Scientist Intern

Amazon
June 2018--March 2022

Research Engineer

École Polytechnique
August 2017--May 2018

Teaching Assistant

University of Houston

Statistical Methods in Research

Software Engineering

August 2016 - May 2017

Research Assistant

September 2014 - July 2016

Research Associate

NCSR Demokritos, Software Knowledge and Engineering Lab
June 2013 - September 2013

Software Engineering Intern

NCSR Demokritos

Research

Journals

C Kosma, G Nikolentzos, G Panagopoulosu>, JM Steyaert, M Vazirgiannis

Neural Ordinary Differential Equations for Modeling Epidemic Spreading

Transactions on Machine Learning Research,2023

Short Description: A graph neural ordinary differential equation model to predict the Susceptible-Infectious-Recovered epidemic model. [pdf] [code]

G. Panagopoulos, F. Malliaros and M. Vazirgiannis

Multi-task Learning for Influence Estimation and Maximization

IEEE Transactions On Knowledge and Data Engineering (2020) Impact Factor: 4.935

Short Description: A multi-task influence learnng model and an algorithm that uses the learnt representations to perform efficient influence maximization. [code] [video]

G. Panagopoulos and I. Pavlidis

Forecasting Markers of Habitual Driving Behaviors Associated With Crash Risk

IEEE Transactions On Intelligent Transportation Systems (2019) Impact Factor: 4.051

Short Description: An extreme gradient boosting model that takes as input the physiological signals of a driver and vehicle indications and provides short-term predictions of distracted or aggressive driving. [code] [presentation]

I. Pavlidis, D. Zavlin, A. Khatri, A. Wesley, G. Panagopoulos and A. Echo

Absence of Stressful Conditions Accelerates Dexterous Skill Acquisition in Surgery

Nature Scientific Reports (2019) Impact Factor: 4.122

Short Description: Analysis of an experiment that dealt with the skill acquisition of surgigal trainees, using statistical inference and visualization of physiological recordings and questionairs. [data]

G. Panagopoulos, G. Tsatsaronis and I. Varlamis

Detecting Rising Stars in Dynamic Collaborative Networks

Elsevier Journal of Informetrics (2017) Impact Factor: 3.879

Short Description: Identify young scientists with high potential, using social network analysis and unsupervised machine learning. [code]

Conferences

G Panagopoulos, N Tziortziotis, F Malliaros, M Vazirgiannis

Maximizing Influence with Graph Neural Networks

International Conference on Social Networks Analysis and Mining (ASONAM),2023

Short Description: A graph neural network to predict the Independent Cascade model and a submodular algorithm based on its representations to perform influence maximization. [pdf] [code]

B. Rozemberczki, P. Scherer, Y. He, G. Panagopoulos, M. Astefanoaei, O. Kiss, F. Beres, N. Collignon, R. Sarkar

PyTorch Geometric Temporal: Spatiotemporal Signal Processing with Neural Machine Learning Models Best Resource Paper

Conference on Information and Knowledge Management (CIKM resource track),2021

Short Description: A library with temporal graph neural network methods and temporal graph datasets. [pdf] [code]

G Nikolentzos, G Panagopoulos, M Vazirgiannis.

An Empirical Study of the Expressiveness of Graph Kernels and Graph Neural Networks

International Conference on Artificial Neural Networks (ICANN),2021

Short Description: An experimental study to quantify the accuracy of graph kernels and graph neural networks for approximating graph similarity using a well-known similarity function for evaluation. [pdf]

G. Panagopoulos, G. Nikoletzos, M. Vazirgiannis.

Transfer Graph Neural Networks for Pandemic Forecasting

AAAI International Conference on Artificial Intelligence (AAAI),2021

Short Description: A graph neural network that uses temporal mobility networks and the recent history of the epidemic to predict the number of COVID-19 cases per day in NUTS3 geographical regions of different EU countries. [pdf] [code]

G. Panagopoulos, F. Malliaros, M. Vazirgiannis.

Influence Maximization using Influence and Susceptibility Embeddings Best paper nomination

AAAI International Conference on Web and Social Media (ICWSM), 2020

Short Description: An influence maximization method based on influence representation learning from diffusion cascades. [pdf] [code] [video]

G. Panagopoulos, C. Xypolopoulos, K. Skianis, C. Giatsidis, J. Tang, M. Vazirgiannis.

Scientometrics for Success and Influence in the Microsoft Academic Graph

International Conference on Complex Networks and Their Applications (Complex Networks), 2019

Short Description: An online application with visualizations on scientific success, using field-based h-index and large scale d-core decomposition on the MAG. [code] [app] [presentation]

G. Panagopoulos, F. Malliaros, M. Vazirgiannis

DiffuGreedy: An Influence Maximization Algorithm based on Diffusion Cascades

International Conference on Complex Networks and Their Applications (Complex Networks), 2018

Short Description: We propose a new algorithm to perform influence maximization utilizing diffusion cascades that have taken place over the network. [pdf][code]

G. Panagopoulos

Multi-Task Learning for Commercial Brain Computer Interfaces

IEEE BioInformatics and BioEngineering (BIBE), 2017

Short Description: Apply and compare conventional and multi-task machine learning algorithms used in Brain Computer Interface literature in two open datasets of mental monitoring experiments which utilized Neurosky EEG device. [pdf] [code] [presentation] [poster]

P. Karampiperis, A. Koukourikos G. Panagopoulos

From Computational Creativity Metrics to the Principal Components of Human Creativity

Knowledge, Information and Creativity Support Systems (KICSS), 2014

Short Description: Model the creativity of children playing serious games in school, with natural language processing and semantic analysis. [pdf] [code]

A Koukourikos, P Karampiperis, G. Panagopoulos

Creative Stories: A Storytelling Game fostering Creativity

Cognition and Exploratory Learning in Digital Age (CELDA), 2014

Short Description: A serious game using computational semantic lateral thinking techniques to motivate children write more creative essays. [pdf] [code] [presentation]

Workshops

G. Panagopoulos, H. Jalalzai

Graph Neural Networks with Extreme Nodes Discrimination

Deep Learning on Graphs: Methods and Applications, KDD 2020

Short Description: An experimental analysis on the use of extreme value theory in semi-supervised classification using graph neural networks. [pdf] [code]

P. Boniol, G. Panagopoulos, C. Xypolopoulos, R. El Hamdani, D. Restrepo Amariles, M. Vazirgiannis

Performance in the Courtroom: Automated Processing and Visualization of Appeal Court Decisions in France

Workshop on Natural Legal Language Processing, KDD 2020

Short Description: Text mining and analysis of entities in legal documents of French appeal court decisions. [pdf]

G. Panagopoulos, C. Palmer

A Specialized Interactive Data Application for EEG Based Sleep Studies

Workshop on Assistive Technologies for Decision making in Healthcare, PETRA 2017

Short Description: A web data application using signal processing techniques and visualizations to assist sleep EEG analysis in a study with anxious children. [pdf] [code] [presentation]

G. Panagopoulos, P. Karampiperis, A. Koukourikos, S. Konstantinidis

Creativity Profiling Server: Modelling the Principal Components of Human Creativity over Texts

Workshop on Deep Content Analytics Techniques for Personalized and Intelligent Services, UMAP 2015

Short Description: A server modeling a user's creativity based on textual exhibits, employing computational creativity metrics and unsupervised learning. [pdf] [code] [presentation]

Posters

G. Panagopoulos

Network Inference from Neural Activation Time Series: A comparative review

International Conference on Network Science, 2018

Short Description: Apply and compare different network inference techniques to Kaggle connectomics small dataset. [pdf] [code] [poster]

Resources

Courses

Textbooks

  • Kevin Murphy, "Machine Learning: a Probabilistic Perspective", MIT press
  • Jon Kleinberg and Éva Tardos, "Algorithm Design", Pearson Education
  • Simon Haykin, "Neural Networks, A Comprehensive Foundation", Pearson
  • Deepayan Chakrabarti and Christos Faloutsos, "Graph Mining: Laws, Tools, and Case Studies", Morgan & Claypool Publishers
  • Albert-László Barabási and Pósfai Márton, "Network science", Cambridge university press

Books

  • Roger Penrose, "The Emperor's New Mind", Oxford University Press
  • Marvin Minski, "The Society of Mind", Simon & Schuster
  • Martin Davis, "Engines of Logic: Mathematicians and the Origin of the Computer", Norton
  • Doxiadis Apostolos and Christos Papadimitriou, "LOGICOMIX: an epic search for truth", Bloomsbury Publishing

Friends and Colleagues

Blog