Hi, bonjour, yeia

Welcome to my personal web-page. I am Argyris Kalogeratos (Αργύρης Καλογεράτος), I come from Patras, Greece and my background is on Computer Science. My PhD was on the field of Machine Learning (Dept. of Computer Science and Engineering, University of Ioannina, Greece – 2013), under the supervision of Aristidis Likas. This came right after a BSc and MSc on Computer Science at the same Academic Institution.

Currently

Position: I am a permanent researcher at the Centre Giovanni Borelli (ex Centre de Mathématiques et de Leurs Applications – CMLA), Ecole Normale Supérieure Paris-Saclay (ENS Paris-Saclay), University Paris-Saclay, France, and member of the Machine Learning and Massive Data Analysis (MLMDA) research group (head: Nicolas Vayatis).

Research focus: Principal researcher on ML on Graphs.

Research initiatives: I participate to the research activities of the ML Industrial Big Data Chair (C.Borelli team: Charles Truong, Harry Sevi, Argyris Kalogeratos, Mathilde Mougeot, Nicolas Vayatis — past: Julien Audiffren). Among our industrial partners, there are ATOSWordline, SNCF, CEA, Michelin, Bertin Tech, and Banque de France.

Past funding: My research position at CMLA has been funded in the past by the MORANE project (in collaboration with SNCF railway company) and the SODATECH project  (funded by the French state) for which I was the research coordinator and responsible for CMLA’s part. See more details at the “Projects” section.

 

Announcements

Research opportunities for MVA-ers

Here are some of the subjects that the ML group (MLMDA) of Center Borelli offers for MVA Masters program internships:

Research on COVID-19 epicemic

A new web page was posted hosting results from our lab and resources from the research community.

Research opportunities for MVA-ers

Here are some of the subjects that the ML group (MLMDA) of Center Borelli offers for MVA Masters program internships:

Research on COVID-19 epicemic

A new web page was posted hosting results from our lab and resources from the research community.

News

  • TechReport – “Efficient stream-based Max-Min diversification with minimal failure rate” co-authored with Μ. Fekom [pdf]
  • New team memberIt’s a great pleasure that Harry Sevi joined the MLMDA research group. His postdoc position will be funded by the Chair AI and it will be great that we will have the opportunity to work closely together on subjects related to ML and Graphs, as well as on industrial applications.
  • Blog page – Here’s a new blog page with posts around and beyond technical matters that do matter.
  • New internI welcome with pleasure our new intern: Ivan Conjeaud (M1 ENS Economics and Parcour AI) will work on competitive epidemics with application to panic buying, trend diffusion, etc.
  • Workshop – Proud co-organizers of the Digital French-German Summer School with Industry 2020! Several works of our team will be presented in this event!
  • TechReport – “Winning the competition: enhancing counter-contagion in SIS-like epidemic processes” co-authored with S. S. Mannelli [pdf]
  • TechReport – “Model family selection for classification using Neural Decision Trees” co-authored with A. Merida and M. Mougeot [pdf]
  • TechReport – “Offline detection of change-points in the mean for stationary graph signals” co-authored with A. de la Concha and N. Vayatis [pdf]
  • Follow-up – the paper “Dynamic Epidemic Control via Sequential Resource Allocation” co-authored with M. Fekom and N. Vayatis, is the finalization of the previous work published at CDC 2019 and now is under journal review [pdf].
  • ICML 2020 – The paper “Learning the piece-wise constant graph structure of a varying Ising model” co-authored with B. Le Bars, P. Humbert, and N. Vayatis, will appear to the International Conference on Machine Learning 2020 [pdf: a previous version, the final will be out soon].
  • The ONADAP project  – a newly funded research project/collaboration that will try to produce practical solutions for hospital monitoring and decision making during a period such as the recent pandemic. Centre Borelli and the MLMDA team have central role in this project. I will be responsible for the epidemic modeling and automatic decision making part.
  • New interns – I welcome with great pleasure our new interns:
    • Randa Elmrabet Tarmach (M2 MVA) will work on unsupervised statistical learning;
    • Kais Cheikh (M1 ENSTA) will work on budget(ed) learning;
    • Mohamed El Khames BOUMAIZA (M1 ENSTA) will work on change-points detection in network streams;
    • Wendong Liang and Baptiste Loreau (L3 Maths ENS Paris-Saclay) will work on the intersection of topological data analysis and graph theory;
    • Martin Dhaussy (L3 Economics ENS Paris-Saclay) will work graph robustification;
    • Aaron MAMANN (M1)… subject to be specified.
  • New PhD with scholarship – Our PhD project proposal with the title “Towards Interpretable and Versatile Machine Learning” for the candidate Anthea Merida (supervised by M. Mougeot and me) was ranked 1st in the DIM MathInnov of Paris Region PhD 2020.
  • COVID-19 epidemic – A new web page was posted hosting results from our lab and resources from the research community.
  • TechReport – “Learning Laplacian Matrix from Graph Signals with Sparse Spectral Representation”. P. Humbert, B. Le Bars, L. Oudre, A. Kalogeratos, and N. Vayatis, Technical Report, Nov 2019. [pdf]Abstract
    In this paper, we consider the problem of learning a graph structure from multivariate signals, known as graph signals. Such signals are multivariate observations carrying measurements corresponding to the nodes of an unknown graph, which we desire to infer. They are assumed to enjoy a sparse representation in the graph spectral domain, a feature which is known to carry information related to the cluster structure of a graph. The signals are also assumed to behave smoothly with respect to the underlying graph structure. For the graph learning problem, we propose a new optimization program to learn the Laplacian of this graph and provide two algorithms to solve it, called IGL-3SR and FGL-3SR. Based on a 3-steps alternating procedure, both algorithms rely on standard minimization methods – such as manifold gradient descent or linear programming – and have lower complexity compared to state-of-the-art algorithms. While IGL-3SR ensures convergence, FGL-3SR acts as a relaxation and is significantly faster since its alternating process relies on multiple closed-form solutions. Both algorithms are evaluated on synthetic and real data. They are shown to perform as good or better than their competitors in terms of both numerical performance and scalability. Finally, we present a probabilistic interpretation of the optimization program as a Factor Analysis Model. Keywords: Graph learning, graph Laplacian, non-convex optimization, graph signal processing, sparse coding, clustering.
  • Centre Giovanni Borelli – Since the 1st of January 2020, the Centre de Mathématiques et de Leurs Applications (CMLA) of the ENS Paris-Saclay and the research team CognacG of the University Paris Centre have formed the Center Giovanni Borelli (Unité Mixte de Recherche : UMR 9010).
  • TechReport – A new paper was released, as technical report, entitled “Detecting multiple change-points in the time-varying Ising model”. The paper is co-authored with B. Le Bars, P. Humbert, and N. Vayatis, Oct 2019. [pdf][arXiv link] Abstract
    This work focuses on the estimation of change-points in a time-varying Ising graphical model (outputs −1 or 1) evolving in a piecewise constant fashion. The occurring changes alter the graphical structure of the model, a structure which we also estimate in the present paper. For this purpose, we propose a new optimization program consisting in the minimization of a penalized negative conditional log-likelihood. The objective of the penalization is twofold: it imposes the learned graphs to be sparse and, thanks to a fused-type penalty, it enforces them to evolve piecewise constantly. Using few assumptions, we then give a change-point consistency theorem. Up to our knowledge, we are the first to present such theoretical result in the context of time-varying Ising model. Finally, experimental results on several synthetic examples and a real-world dataset demonstrate the empirical performance of our method.
  • ICTAI 2019 – “Optimal Multiple Stopping Rule for Warm-Starting Sequential Selection, co-authored with Mathilde Fekom and Nicolas Vayatis, will appear at the IEEE International Conference on Tools with Artificial Intelligence (ICTAI), to be held at Portland, Oregon, US, on Nov. 4 – 6, 2019. [pdf][arxiv link]Abstract
    This work focuses on the estimation of change-points in a time-varying Ising graphical model (outputs −1 or 1) evolving in a piecewise constant fashion. The occurring changes alter the graphical structure of the model, a structure which we also estimate in the present paper. For this purpose, we propose a new optimization program consisting in the minimization of a penalized negative conditional log-likelihood. The objective of the penalization is twofold: it imposes the learned graphs to be sparse and, thanks to a fused-type penalty, it enforces them to evolve piecewise constantly. Using few assumptions, we then give a change-point consistency theorem. Up to our knowledge, we are the first to present such theoretical result in the context of time-varying Ising model. Finally, experimental results on several synthetic examples and a real-world dataset demonstrate the empirical performance of our method.
  • CDC 2019 – “Sequential Dynamic Resource Allocation for Epidemic Control, co-authored with Mathilde Fekom and Nicolas Vayatis, will appear at the IEEE Conference on Decision and Control 2019 (CDC), to be held at Nice, France, on Dec. 11 – 13, 2019. [pdf][arxiv link]Abstract
    TUnder the Dynamic Resource Allocation (DRA) model, an administrator has the mission to allocate dynamically a limited budget of resources to the nodes of a network in order to reduce a diffusion process (DP) (e.g. an epidemic). The standard DRA assumes that the administrator has constantly full information and instantaneous access to the entire network. Towards bringing such strategies closer to real-life constraints, we first present the Restricted DRA model extension where, at each intervention round, the access is restricted to only a fraction of the network nodes, called sample. Then, inspired by sequential selection problems such as the well-known Secretary Problem, we propose the Sequential DRA (SDRA) model. Our model introduces a sequential aspect in the decision process over the sample of each round, offering a completely new perspective to the dynamic DP control. Finally, we incorporate several sequential selection algorithms to SDRA control strategies and compare their performance in SIS epidemic simulations.
  • TechReport – A new paper was released, as technical report, entitled
    “Revealing posturographic features associated with the risk of falling in patients with Parkinsonian syndromes via machine learning”. The paper is co-authored with I. Bargiotas, M. Limnios, P.–P. Vidal, D. Ricard, and N. Vayatis, July 2018. [pdf][arXiv link] Abstract

    Falling in Parkinsonian syndromes (PS) is associated with postural instability and consists a common cause of disability among PS patients. Current posturographic practices record the body’s center-of-pressure displacement (statokinesigram) while the patient stands on a force platform. Statokinesigrams, after appropriate signal processing, can offer numerous posturographic features, which however challenges the efforts for valid statistics via standard univariate approaches. In this work, we present the ts-AUC, a non-parametric multivariate two-sample test, which we employ to analyze statokinesigram differences among PS patients that are fallers (PSf) and non-fallers (PSNF). We included 123 PS patients who were classified into PSF or PSNF based on clinical assessment and underwent simple Romberg Test (eyes open/eyes closed). We analyzed posturographic features using both multiple testing with p-value adjustment and the ts-AUC. While the ts-AUC showed significant difference between groups (p-value = 0.01), multiple testing did not show any such difference. Interestingly, significant difference between the two groups was found only using the open-eyes protocol. PSF showed significantly increased antero-posterior movements as well as increased posturographic area, compared to PSNF. Our study demonstrates the superiority of the ts-AUC test compared to standard statistical tools in distinguishing PSF and PSNF in the multidimensional feature space. This result highlights more generally the fact that machine learning-based statistical tests can be seen as a natural extension of classical statistical approaches and should be considered, especially when dealing with multifactorial assessments.

See all news...