Teaching
Students
-
Postdoc 2021-23
-
Postdoc 2021-23
Timothée Mathieu, Inria Lille
Topic: Robust Bandits under Stochastic Corruptions
-
PhD 2024-
Udvas Das, Inria Lille
Topic: Impacts of Constraints on Bandits and RL
-
PhD 2022-
Ayoub Azara, Inria Lille
Topic: Auditing ML Algorithms
-
PhD 2021-
Achraf Azize, Inria Lille
Topic: Learning to Play the Privacy Game
-
Intern 2023
Priyank Agrawal, Columbia University
Topic: Regret Minimisation under Linear Constraints
-
Intern 2021-22
Bishwamittra Ghosh, National University of Singapore
Topic: Auditing and Explaining Bias of ML Algorithms
-
Masters 2023
Thomas Michel, Ecole Polytechnique, France
Topic: Risk-sensitive Reinforcement Learning for Forest Management
-
Masters 2022-2023
Udvas Das, RKMVERI, India
Topic: Impacts of Constraints on Adversarial Bandits
-
Masters 2021-2022
Pratik Karmakar, RKMVERI, India
Topic: Adaptive Distributionally-Equivalent and Max-Information Extraction Attack
-
Masters 2022
William Tioulong, École Polytechnique
Topic: Algorithmic Auditing of Bias
-
Masters 2022
Xabier Fabian Roldan Figueredo, PSL University
Topic: Asynchronous Hyperparameter Tuning with Bandits
-
Masters 2021
Cyriaque Rousselot, University of Lille, France
Topic: Online Information Inference Attack with Reinforcement Learning
-
Masters 2021
Maxime Boutton, University of Lille, France
Topic: Measuring Bias in Outcome of ML Algorithms
-
Masters 2020
Igor Ryazanov, Chalmers University of Technology, Sweden
Topic: Deep Learning for Deep Water (in collaboration with Maritime Studies and Maritime Environmental Sciences)
-
Masters 2020
Sushma Tungal and Pragya Singh, Chalmers University of Technology, Sweden
Topic: Preconditioning of Batteries using Machine Learning (in collaboration with Volvo Cars)
Courses
-
Fall-Spring 2023-24
Sequential Decision Making
Lecturer, M2 Masters of Data Science, Université de Lille, and Centrale Lille
Fundamentals of online learning, bandits, and reinforcement learning with an exposition of both theory and practice. Details are available in this webpage.
-
Spring 2020-22
Anonymization and Privacy (Anonymisation, confidentialité)
Lecturer, M2 Masters IASD (AI and Data Science), PSL University: Paris-Dauphine, École normale supérieure, and Mines ParisTech
The course covers the following topics: Basics of data privacy and anonymization, Measuring anonymity: k-anonymity, l-diversity, m-closeness, A probabilistic framework: differential privacy, Differentially private mechanisms, Privacy-preserving Machine Learning, Deployment of differential privacy, Federated learning, Privacy attacks, Processing data anonymously: homomorphic encryption. Check this link for the course materials.
-
Spring 2021-24
Advanced Machine learning & Decision Theory: Lectures on Responsible ML
Lecturer, Specialisation on Data Science and AI, Centrale Lille
A lecture series on responsible AI: privacy, robustness, and fairness.
-
Fall & Spring 2021-24
Research Reading Group
Lecturer, M2 Masters of Data Science, Université de Lille, and Centrale Lille
Training for reading, writing, presenting, and reviewing paper for developing an ethical research career.
-
Spring 2020
FDAT070: Decision Making Under Uncertainty and Reinforcement Learning
Instructor, Department of Computer Science and Engineering, Chalmers University of Technology
This course gives an introduction to decision theory and reinforcement learning, focusing on theory and algorithms. reinforcement learning problems. Each module lasts for about a week, and includes a short lecture of basic concepts, an overview of proofs and some hands-on work. Assessment is through assignments and a mini project.
Pedagogical Courses
-
Autumn 2020
CLS905: Supervising Research Students
Diploma in Teaching and Learning in Higher Education, Chalmers University of Technology, Sweden
The course addresses supervision of doctoral students’ research and the study and learning process. It also highlights the close link between supervision and education in the skills of teaching and learning. The course clarifies how a supervisor may meet a PhD student as an individual by taking an inclusive approach and, with that student, build a relationship based on mutual understanding of postgraduate education in general and the student’s research project in particular. The course provides plenty of opportunity for discussion and reflection on how to start up and follow the doctoral supervision process in different phases.
-
Spring 2020
CIU965: Diversity and Inclusion for Learning in Higher Education
Diploma in Teaching and Learning in Higher Education, Chalmers University of Technology, Sweden
This course uses a non-essentialist, experience-driven approach which is discourse-based, theory-referenced and interaction-orientated. This means that, before a session, participants will be expected to read relevant literature and complete tasks to facilitate discussion. Participants’ own experience in these discussions is seen as crucial in building up an understanding of this area. The focus will be on diversity in terms of culture, gender and special needs. Discussions will focus on key issues in these areas and models for interpreting them, so as to produce as inclusive an education as possible.
-
Autumn 2019
CIU950: University Teaching and Learning
Diploma in Teaching and Learning in Higher Education, Chalmers University of Technology, Sweden
The course focuses on the recent worldwide paradigm shift in higher education from teaching to learning. Today’s university lecturers are expected to be more of a “guide on the side” than a “sage on the stage”, more of a facilitator of learning than a presenter of information. Three fundamental questions therefore form the backbone of the course:
- What do we want our students to learn, and how can we make the goals clear to our them?
- How do we arrange teaching so that our students have the greatest chance of learning?
- How can we find out whether they have learnt what we wanted them to learn?
An important idea tied to these questions is constructive alignment, which means that different aspects of a teaching and learning situation should be brought together in a mutually reinforcing way that supports student learning.
Reading Group
-
Spring 2022-2023
Causal Reinforcement Learning
This is a reading group consituted by researchers in CausalXRL project for studying connections between causality and reinforcement learning.
-
Fall 2015-16
Measure Theory
This is an informal reading group made by a few graduate students of School of Computing. The study began from simple curiosity to know the grammar behind probability theory that we often use in machine learning and artificial inteligence literature. Though we have started with the puritarian classic of Halmos but we have moved to the simple and practical introduction by Tao in search of a link between reality and measure theory.
References:
Measure theory. Halmos, Paul R. Vol. 18. Springer, 2013.
An introduction to measure theory. Tao, Terence, ed. Vol. 126. American Mathematical Soc., 2011.
(Ancient) Courses
-
Fall 2017
CS6234: Advanced Algorithms
Teaching Assistant, Department of Computer Science, School of Computing, National University of Singapore
This module is aimed at graduate students who are doing or intend to do advanced research in algorithms design and analysis in all areas of computer science. The module covers advanced material on combinatorial algorithms, with emphasis on efficient algorithms, and explores their use in a variety of application areas. Topics covered include, but are not restricted to, linear programming, graph matching and network congestion, approximation algorithms, randomized algorithms, online algorithms, and learning algorithms. It is a seminar-based module that will expose students to the state-of-the-art and also the current research in these areas.
-
Spring 2017
CS3230: Design and Analysis of Algorithms
Teaching Assistant, Department of Computer Science, School of Computing, National University of Singapore
This module introduces different techniques of designing and analysing algorithms. Students will learn about the framework for algorithm analysis, for example, lower bound arguments, average case analysis, and the theory of NP-completeness. In addition, students are exposed to various algorithm design paradigms. The module serves two purposes: to improve the students' ability to design algorithms in different areas, and to prepare students for the study of more advanced algorithms. The module covers lower and upper bounds, recurrences, basic algorithm paradigms (such as prune-and-search, dynamic programming, branch-and-bound, graph traversal, and randomised approaches), amortized analysis, NP-completeness, and some selected advanced topics.
-
Spring 2017
CS1010E: Programming Methodology
Teaching Assistant, Department of Computer Science, School of Computing, National University of Singapore
This module introduces the fundamental concepts of problem solving by computing and programming using an imperative programming language. It is the first and foremost introductory course to computing and the first part of a three-part series on programming and problem solving by computing, which includes CS1020 and CS2010. Topics include problem solving by computing, writing pseudo-codes, problem formulation and problem solving, program development, coding, testing and debugging, fundamental programming constructs (variables, types, expressions, assignments, functions, control structures, etc.), fundamental data structures: arrays, strings and structures, simple file processing, and basic recursion.
-
Fall 2016-17
CS1231: Discrete Structures
Teaching Assistant, Department of Computer Science, School of Computing, National University of Singapore
This module introduces mathematical tools required in the study of computer science. Topics covered in this course include:
1) Logic and proof techniques: propositions, conditionals, quantifications;
2) Relations and Functions: Equivalence relations and partitions, partially ordered sets, well-ordering principle, function equality, Boolean/identity/inverse functions, Bijection;
3) Mathematical formulationo of data models (linear model, trees, graphs);
4) Counting and Combinatoric: Pigeonhole Principle, Inclusion-Exclusion Principle, Number of relations on a set, number of injections from one finit set to another, Diagonalisation proof: An infinite countable set has an uncountable power set; Algorithmic proof: An infinite set has a countably infinite subset, subsets of countable sets are countable.
-
Fall 2015-16
CS3230: Design and Analysis of Algorithms
Teaching Assistant, Department of Computer Science, School of Computing, National University of Singapore
This module introduces different techniques of designing and analysing algorithms. Students will learn about the framework for algorithm analysis, for example, lower bound arguments, average case analysis, and the theory of NP-completeness. In addition, students are exposed to various algorithm design paradigms. The module serves two purposes: to improve the students' ability to design algorithms in different areas, and to prepare students for the study of more advanced algorithms. The module covers lower and upper bounds, recurrences, basic algorithm paradigms (such as prune-and-search, dynamic programming, branch-and-bound, graph traversal, and randomised approaches), amortized analysis, NP-completeness, and some selected advanced topics.
-
Spring 2014-15
CS1020E: Data Structures and Algorithms I
Teaching Assistant, Department of Computer Science, School of Computing, National University of Singapore
This module is the second part of a three-part series on introductory programming and problem solving by computing. It continues the introduction that begins in CS1010, and emphasises objectoriented programming with application to simple data structures. Topics include object-oriented problem modeling with objects, classes and methods, object-oriented problem formulation and solving, data structure implementation strageties, abstraction and encapsulation of data structures, object-oriented programming constructs, APIs and class libraries, exception handling, lists, linked lists, stacks, queues, hash tables and their algorithmic design, sorting and searching methods, recursive algorithms, and Big-O notation. This module is appropriate for Faculty of Engineering students too.
-
Spring 2014-15
DSC5211C: Quantitative risk management
Teaching Assistant, Department of Decision Sciences, Business School, National University of Singapore
The aim of this course is to provide an introduction to the probability and statistical methods used by financial institutions and supply chain managers to model market, credit and operational risk. Topics addressed include loss distributions, multivariate models, dependence and copulas, extreme value theory, risk measures, risk aggregation, risk allocation and supply chain risk management.