Process

• Students must register for the courses IFT2015 and IFT2125 in the 1st semester of their studies. If one of the two courses is not offered, the inscription will be postponed to the next  semester in which the course is offered. If the student has already taken one of the courses (e.g. during his undergraduate studies at DIRO), he/she must inform the director of the predoc committee
  
  
• PhD students taking these courses as part of their predoctoral examination are not required to submit homeworks (though they are encouraged to do so). The evaluation will only be based on the midterm and final exams of the course. The professor will give a literal grade to each of the students based on the overall results of all students enrolled in the course..

Success Criteria

• The student must obtain a grade greater than or equal to B+ in both courses.
  
  
• If the student gets a grade below B+ in one of the two courses, he/she must retake the exams of the course during the next semester in which it is offered. No interruption of studies will be possible for a student who failed their first attempt at the exam.
  
  
• If the student gets a grade below B+ the second time he/she takes the exam, he/she will be excluded from the PhD program.

Exemption for exceptional students

• A student who has already taken IFT2015 or IFT2125 during his undergraduate or masters studies and obtained a grade greater than or equal to A- is exempted from taking the exams for this course.
  
  
• A student who passed a course equivalent to IFT2015 or IFT2125 in a university/school other than UdeM can be exempted from taking the corresponding exams. The student must provide evidence that the syllabus of the course he passed is equivalent to one of IFT2015 or IFT2125 and that he/she obtained a grade greater than or equal to A-. The decision of whether this exemption is granted or not is taken by the predoctoral examination committee in coordination with the head of the DIRO graduate studies committee.

Process

The validation of the student’s knowledge of their specialty domain is evaluated through a written exam offered during the Winter and Fall semesters. It is the responsibility of the student’s supervisor and the head of the lab to facilitate and accelerate the completion of this part of the predoctoral examination.

To be allowed to write this exam forming part 2 of his/her predoctoral requirements, the student must have completed part 1 of the predoc (or have been exempted from part 1) and must have taken the two graduate courses required in his/her PhD program.

Calendar for Winter 2023

Success Criteria

Students must obtain a grade greater than  or equal to B in order to pass this part of the predoctoral examination. If a student obtains a lower grade, he/she must take the exam in the next semester in which  it is offered. No interruption of studies will be possible for a student who failed their first attempt at the exam. If the student gets a grade below B the second time he/she takes the exam,he/she will be excluded from the PhD program.

Topics

The areas of study for the specialty exam correspond to the different DIRO labs.

• Informatique :
  

  • Biologie informatique et théorique (LBIT)
  • Génie logiciel (GEODES)
  • Informatique théorique et quantique (LITQ)
  • Imagerie
    • Infographie (LIGUM)
    • Traitement d'images
    • Vision 3D (VISION)

  • Intelligence artificielle
    • Algorithmes d'apprentissage (MILA)
    • Linguistique informatique (RALI)
    • Robotique

  • Neuroinformatique
  • Parallélisme (LTP)
  • Réseaux de communication (LRC)

• Recherche opérationnelle :
  
  • Optimisation déterministe
  • Modèles stochastiques

La matière à couvrir pour chaque domaine de l'examen de spécialité est la suivante :

LBIT (Miklós Csűrös, Nadia El-Mabrouk, Sylvie Hamel, François Major)

L'examen en biologie computationnelle est constitué de 2 parties portant sur 2 thèmes différents. L'étudiant doit choisir 2 thèmes parmi les 4 proposés ci-dessous et étudier en profondeur les articles donnés pour ces thèmes. L'étudiant doit communiquer son choix au comité. L'examen sera composé de 2 questionnaires portants sur les 2 thèmes choisis par l'étudiant.

• Thème I. Génomique comparative (proposé par Sylvie Hamel)
   
  
  1. Anne Bergeron, Julia Mixtacki et Jens Stoye, "A new linear time algorithm to compute the genomic distance via the double cut and join distance", Journal of Theoretical Computer Science, 410(51), 2009.
     http://drops.dagstuhl.de/opus/volltexte/2010/2689/pdf/10231.StoyeJens.Paper.2689.pdf
      
  2. Aïda Ouangraoua et Anne Bergeron, "Parking functions, labeled trees and DCJ sorting scenarios", RECOMB-CG 2009, LNCS 5817, pp 24-35, 2009.
     https://arxiv.org/abs/0903.2499
      

• Thème II. Phylogénie (proposé par Nadia El-Mabrouk)
   
  
  1. Guy Perrière et Céline Brochier-Armanet, "Concepts et méthodes en phylogénie moléculaire", Springer-Verlag, France, 2010
      

• Thème III. Bio-informatique de la régulation génique par les microARN (proposé par François Major)
   
  
  1. Hafner et al., "Transcriptome-wide identification of RNA-binding protein and microRNA target sites by PAR-CLIP". Cell 141, 129-141 (2010).
      
  2. Selbach et al., "Widespread changes in protein synthesis induced by microRNAs". Nature455, 58-63 (2008).
      

• Thème IV. Next-generation sequencing, genome assembly (proposé par Miklós Csűrös)
   
  
  1. Jay Shendure et Erez Lieberman Aiden, "The expanding scope of DNA sequencing", Nature Biotechnology 30:1084-1094 (2012).
      
  2. Niranjan Nagarajan et Mihai Pop, "Sequence assembly demystified", Nature Reviews Genetics, 14:157-167 (2013).

GEODES (Houari Sahraoui, Michalis Famelis, Eugene Syriani)

L'examen en génie logiciel est constitué de 3 parties, chacune contenant des questions sur un article.

Les articles sont :

1. D. Poshyvanyk, Y. G. Gueheneuc, A. Marcus, G. Antoniol and V. Rajlich, "Feature Location Using Probabilistic Ranking of Methods Based on Execution Scenarios and Information Retrieval," in IEEE Transactions on Software Engineering, vol. 33, no. 6, pp. 420-432, June 2007.http://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=4181710ieeexplore.ieee.org/stamp/stamp.jsp


2. Paige Rodeghero, Siyuan Jiang, Ameer Armaly, and Collin McMillan. 2017. Detecting user story information in developer-client conversations to generate extractive summaries. In Proceedings of the 39th International Conference on Software Engineering (ICSE '17). IEEE Press, Piscataway, NJ, USA, 49-59. DOI: doi.org/10.1109/ICSE.2017.13


3. Egea, Marina, and Carolina Dania. "SQL-PL 4OCL: an automatic code generator from OCL to SQL procedural language." Software & Systems Modeling (2017): 1-23.https://doi.org/10.1007/s10270-017-0597-6doi.org/10.1007/s10270-017-0597-6

LITQ (Gilles Brassard)

Les sujets seront tirés des 4 livres suivants :

1. John E. Hopcroft, Rajeev Motwani et Jeffrey D. Ullman, Introduction to Automata Theory, Languages, and Computation, 2e édition, Addison-Wesley, 2001.
   • Chapitres 1, 2, 3, 4, 5 (sauf 5.3), 6 (sauf 6.4), 7, 8, 9 (sauf 9.4), 10, 11 (sauf 11.5).
     
     ATTENTION : Ces numéros de chapitre correspondent uniquement à la 2de édition.
     
     

2. Robert W. Floyd et Richard Beigel, The Language of Machines : An Introduction to Computability and Formal Languages, W.H. Freeman & Company, 1994.
   • Chapitre 8 (sauf 8.3).
     
     ATTENTION : Seule la traduction française est disponible à la bibliothèque (Le langage des machines) et celle-ci est parfaitement adéquate.
     
     

3. Gilles Brassard et Paul Bratley, Fundamentals of Algorithmics, Prentice Hall, 1996.
   • Chapitre 12.
     
     

4. Michael Sipser, Introduction to the Theory of Computation, 2e édition, Thomson, 2006.
   • Chapitres 7 et 8, puis sections 9.3 et 10.4

Imagerie (LIGUM / Traitement d'Images / VISION) (Pierre Poulin)

L'examen en imagerie est constitué de deux (2) parties, chacune provenant de questions sur un article. L'étudiant doit donc choisir deux (2) articles parmi ceux offerts par les professeurs en imagerie (Mikhail Bessmeltsev, Max Mignotte, Pierre Poulin, Sébastien Roy, Neil Stewart, Bernhard Thomaszewski). L'étudiant doit en aviser le responsable (Pierre Poulin) le plus tôt possible afin d'assembler son questionnaire. Lors de l'examen, une version imprimée de chaque article choisi sera fournie à l'étudiant.

Intelligence artificielle

L'étudiant devra choisir une des 3 options possibles sur laquelle son examen de spécialité portera :

1. Apprentissage
2. Traitement de la langue naturelle
3. Robotique

APPRENTISSAGE (Guillaume Rabusseau)

1. Deep Learning Book http://www.deeplearningbook.org
   Partie I et II au complet.
   Partie III: sections 13.1, 13.2, 13.5; 14.1--6; 15.2--5; 16 et 17 en entier.
   
   
2. Pattern Recognition and Machine Learning, Christopher Bishop, Springer, 2006
   Sections: 1.5, 4.1--3, 6.1--3, 7.1, 9.1--4
3. All topics covered in the syllabi of IFT6390 and IFT6135

TRAITEMENT DE LA LANGUE NATURELLE (Philippe Langlais)

• Daniel Bikel et Imed Zitouni, "Multilingual Natural Language Processing Applications: From Theory to Practice", IBM Press, 2012.
  chap. 1: Structure d’un mot; chap. 2: Structure d’un document; chap. 3: Parsing; chap. 5: Modèles de langue; chap. 10: Traduction automatique; chap .11: Recherche d'information; chap. 12: Résumé automatique.

ROBOTIQUE (Glen Berseth and Liam Paull)

• The combined material from the courses IFT6757 and IFT6095

LTP (Marc Feeley, Stefan Monnier)

1. Andrew W. Appel, "Modern compiler implementation in C/ML/Java", Cambridge University press, deuxième édition, 2002. Chapitres 1 - 17.
    
2. Benjamin C. Pierce, "Types and Programming Languages", The MIT Press, 2002 Chapitres 1-14 (types simples), 22-25 (Système F).

Neuroinformatics

The neuroinformatics exam consists in 2 parts covering 2 topics. The student must select 2 out of the 4 topics below and perform an in-depth study of the papers or books listed for each topic selected. The student must indicate his/her choice of topics at the time of registering for the exam. The exam will consist of 2 questionnaires covering the 2 topics selected :

• Thème I. Modèles computationnels du cerveau et imagerie par résonance magnétique fonctionnelle (responsable : Pierre Bellec)
   
  
  1. R. A. Poldrack, J. A. Mumford and T. E. Nichols. Handbook of functional MRI analysis. www.fmri-data-analysis.org
      
  2. D. Schwartz, M. Toneva, L. Whebe. Inducing brain-relevant bias in natural language processing models. arxiv.org/pdf/1911.03268.pdf
      

• Thème II. Neuroscience computationnelle, modélisation mathématique et dynamique de réseaux (responsable : Guillaume Lajoie)
   
  
  1. CHAPITRES DE LIVRE: Theoretical Neuroscience: Computational and Mathematical Modeling of Neural Systems, P. Dayan & L.F. Abbott, MIT Press.
      
  • Chapitre 5:  Model Neurons I: Neuroelectronics
  • Chapitre 7: Network models
     
  2. ARTICLE: van Vreeswijk, C., & Sompolinsky, H. (1996). Chaos in Neuronal Networks with Balanced Excitatory and Inhibitory Activity. Science (New York, NY), 274(5293), 1724–1726. doi.org/10.1126/science.274.5293.1724
      

• Thème III. Apprentissage machine, modélisation et traitement des signaux cérébraux en MEG/EEG (responsable : Karim Jerbi)
   
  
  1. Partie 1: Magnetoencephalography: Methods and signal processing - Review Article: Baillet S "Magnetoencephalography for brain electrophysiology and imaging", 2017, Nature Neuroscience
     www.ucalgary.ca/i3t/files/i3t/baillet_2017.pdf
      
  2. Partie 2: Leveraging ANNs for MEG/EEG research:
      
  • MEG:  Zubarev et al. “Adaptive neural network classifier for decoding MEG signals”, 2019, Neuroimage research.aalto.fi/files/34091186/1_s2.0_S1053811919303544_main_1.pdf
  • EEG: Yannick Roy et al. “Deep learning-based electroencephalography analysis: a systematic review”  2019 J. Neural Eng. http://faubertlab.com/wp-content/uploads/2020/01/Roy_2019_J._Neural_Eng._16_051001.pdf

   

• Thème IV.  Continual Learning in Nonstationary Environments (Modèles et algorithmes d'IA inspirés des neurosciences)  (responsable : Irina Rish)
   
  
  1. Parisi, G.I., Kemker, R., Part, J.L., Kanan, C. and Wermter, S., 2019. Continual lifelong learning with neural networks: A review. Neural Networks. 
  2. Learning Representations from EEG with Deep Recurrent-Convolutional Neural Networks. Pouya Bashivan, Irina Rish, Mohammed Yeasin, Noel Codella. ICLR 2016 : International Conference on Learning Representations 2016
      

RÉSEAUX DE COMMUNICATION
A. Hafid : IFT 6320

1. William Stallings, "Data and Computer Communications", 7th edition. 
   Tous les chapitres sauf : 10.5, 11, 19.3 et 19.4
    
2. Andrew Tanenbaum, "Computer networks", 3rd edition. 
   Chapitres 1 et 5 seulement.

RECHERCHE OPÉRATIONNELLE

Le candidat a le choix entre 2 examens différents selon que ses champs d'intérêt relèvent de l'optimisation déterministe ou des modèles stochastiques. Chaque examen comporte une partie portant sur le domaine complémentaire (pondération de 20 %). Bien que les matières sujettes à examen soient décrites en détail ci-dessous, les candidats sont encouragés à communiquer avec les responsables des 2 orientations pour toute information supplémentaire.

ORIENTATION OPTIMISATION DÉTERMINISTE
Jean-Yves Potvin

La matière correspond grosso modo à celle du cours IFT6575.

1. Programmation linéaire : Formulation, méthode géométrique, algorithme du simplexe, dualité, problème de flots à coût minimum.
    
2. Programmation non linéaire : Optimisation sans contraintes : fonction à une seule variable, direction de descente, fonction à plusieurs variables; optimisation avec contraintes et conditions d'optimalité; dualité lagrangienne.
    
3. Programmation linéaire en nombres entiers : Formulation, méthode de coupes, méthode d'énumération implicite (branch and bound), relaxation lagrangienne.
    
4. Modèles stochastiques : Rappel de probabilités et de statistiques, chaîne de Markov, processus de décisions markoviens, chaîne de Markov continu, files d'attente.
    
5. Simulation

La matière à couvrir se trouve sur le site web http://www.iro.umontreal.ca/~potvin sous le lien "Matière pour l'examen général de synthèse (prédoc)".

D'autres documents intéressants se trouvent plus bas, dans la liste des références en recherche opérationnelle.

ORIENTATION MODÈLES STOCHASTIQUES
Pierre L'Ecuyer

1. Modèles stochastiques : ([10], tout sauf les sections étoilées).
    
2. Simulation : (chapitres 1 à 6 de [5]).
    
3. Programmation dynamique : (chapitre 21 de [4]).
    
4. Programmation linéaire : Algorithme du simplexe, dualité (chapitre 2 de [7] et les sections correspondantes de [3]).
    
5. Programmation linéaire en nombres entiers : Modélisation, branch-and-bound (chapitres 1 et 7 de [11]).

Références en recherche opérationnelle :

[1] M.S. Bazaraa, H.D. Sherali et C.M. Shetty, « Nonlinear Programming », 2e edition, Wiley, 1993.

[2] M.S. Bazaraa, J.J. Jarvis et H.D. Sherali, « Linear Programming and Network Flows », 3e edition, Wiley, 2005.

[3] V. Chvátal, « Linear Programming », Freeman, 1983.

[4] F.S. Hillier et G.J. Lieberman, « Introduction to Operations Research », 8e edition, 2005, McGraw Hill.

[5] P. L'Ecuyer, « Stochastic Discrete-Event Simulation », support du cours IFT6561, disponible auprès du professeur.

[6] D.G. Luenberger et Y. Ye, « Linear and Nonlinear Programming », 3e edition, Springer, 2008.

[7] P. Marcotte, Support du cours IFT1575.

[8] P. Marcotte, Support du cours IFT3515.

[9] P. Marcotte, Support du cours IFT3655.

[10] H.M. Taylor et S. Karlin, « An Introduction To Stochastic Modelling », 3e edition, Academic Press, 1998.

[11] L.A. Wolsey, « Integer Programming », Wiley, 1998.

[12] S.M. Ross, « Probability Models », 5e edition, Academic Press, 1993.

Other details

1. The exams from previous years can be found in the library.
2. Students must specify their area of specialization when they register for the exam.
3. The examination is a closed book exam (unless otherwise indicated).
4. The exams are corrected by the professor(s) who wrote them. A literal grade will be attributed to each exam. The decision of whether each student passed or failed the exam is taken by the predoc committee.
5. No interruption of studies will be possible for a student who failed their first attempt at the exam.
6. Graded exams can be consulted by the students according to the usual procedures.

This last part of the examination is an oral presentation of the student’s research subject. In addition to the oral presentation, the student must submit a written manuscript describing the research subject, with a substantive bibliography, and describing the different research directions envisioned for the PhD.

Presentation subject presentation

This part of the predoctoral examination aims to verify that the students have identified their thesis subject, that they are familiar with the field of research and that he/she already has promising ideas to successfully complete the PhD  program within a reasonable period of time. The jury does not see the presentation as a contract to be fulfilled, but rather as the presentation of avenues of research which might lead to the advancement of computer science.

This part of the exam is individual and a jury is formed for each student. The jury is made up of the research director(s) and two other professors (the president and a member). The examination consists of an oral presentation and a written report which must be given to the members of the jury at least one week before the scheduled date of the presentation.

If, upon reading the document, the jury does not believe that the candidate is ready to make their presentation, they can either delay the date of the presentation and request a new version of the report, or decide that the student has failed.

It is the student, with the agreement of his research director, who requests the formation of the jury by the graduate studies committee. It is the president of the jury who organizes the presentation, to which all the members of the Department are invited.

The content of the report and of the presentation are determined by the student in collaboration with their supervisor. However, as a suggestion, we give here relevant points that may be included in the report.

Report

The student must, in a report of about 30 to 40 pages, present their research topic. It is important that the report demonstrates the student’s knowledge of the field, with a comprehensive bibliographic review identifying the relevance of the research topic and establishing relations with the chosen subject. In short, the report should convince the jury that the subject is interesting and promising and that the candidate is capable of obtaining tangible results within a reasonable timeframe. Indeed, the report should also contain a realistic timeline.

Oral Presentation

Lasting approximately 45 minutes, this presentation takes up the important points of the report by emphasizing the precise statement of the thesis subject and by establishing the necessary relations between the main works in the field. Out of respect for listeners who are not members of the jury, the presentation should not imply a prior reading of the report, which serves primarily to support the assertions made during the oral presentation.

At the end of the presentation, the members of the jury decide whether or not the student has passed this 3rd part of the predoctoral examination.