ProgramA majority (i.e. all except those happening at Polytechnique) Data AI courses are 24h, count 2.5 ECTS, and are validated by labs, presentations and/or exams. A course is validated when one obtain an average of, at least, 10/20.
M1 yearDuring the M1 year, a student CAN:
- validate up to two research projects awarded 5 ECTS each
- do an 2-3 months internship
- choose up to 10 ECTS outside of DATA AI courses with at most 5 ECTS in non-math/CS topics
- validate a certain number of DATA AI courses.
- validate two research projects OR one internship and at least one research project.
- obtain a total of 60 ECTS.
- Note that these constraints require you to follow a large number of Data AI courses (at least 13, more depending on your choice of project, internship, non Data AI courses, etc.).
M2 yearTo validate the M2 year, a student must accomplish the following:
- fulfill all the Data AI mandatory requirements (see below)
- acquire a total of at least 25 ECTS in Data AI courses and at least 30 ECTS in courses (including the Data AI and mandatory courses)
- validate the M2 internship for 30 ECTS
Data AI mandatory requirementsStudents must validate at least one course for each of the following groups, before the end of the M2 year:
- Group Ethics:
- TPT-DATAAI951 - AI Ethics (Maxwell Winston, Sophie Chabridon, Ada Diaconescu, Fabian Suchanek)
- Group Softskills:
- TPT-DATAAI941 Softskills seminar - Softskills seminar (M2 only) (Fabian Suchanek)
- Group Databases:
- X-INF553 - Database management systems (Ioana Manolescu)
- TPT-SD202 - Databases (Louis Jachiet, Antoine Amarilli)
- TPT-SD202D - Databases (slot D) (Louis Jachiet, Antoine Amarilli)
- Group Big Data Systems:
- TPT-DATAAI922 - Big Data Processing (Louis Jachiet)
- X-INF583 - Systems for Big Data (Angelos Anadiotis / Yanlei Diao)
- TPT-DATAAI921 - Architectures for Big Data (Ioana Manolescu)
- TSP-CSC5003-1 - Big data infrastructures (Bruno Defude)
- Group Logics:
- TPT-IA301 - Logics and Symbolic AI (Isabelle Bloch)
- TPT-SD206 - Logic & Knowledge representation (Jean-Louis Dessalles)
- Group Machine Learning:
- TPT-DATAAI901 - Machine Learning: Shallow & Deep Learning (Mounim El Yacoubi)
- X-INF554 - Machine & Deep Learning Introduction (M. Vazirgiannis)
- Group Data AI basics:
- TPT-DATAAI900 - Data AI basics (Angelos Anadiotis, Tiphaine Viard, Louis Jachiet, Fabian Suchanek, Jean-Louis Dessalles)
Research projects (mandatory for M1 students)
Research projects are a way for students to have a first contact with research, they correspond to a short internship of roughly 10 days worth of work but scattered throughout a semester.
The goal of each research project should be to make a contribution to research in the broad sense, which includes re-implementing well known techniques, benchmarking different approaches, etc.
Research projects are for M1 students. M1 students have to validate two research projects. Theses two projects can be with the same adviser but the second project will be approved only after the first one is finished.
Selecting a topicStudents are encouraged to propose topic and find someone to advise them but the Data AI program will provide a list of research projects topics. In both cases, the student should find a research project adviser. Once the student and the adviser both agree on making a research project together and on a topic, the advisor should send an email to the master team containing the name of the student, the project description. Once the research project has been approved by the Data AI team, it can start.
Doing the research projectThe student should spend 10 days in a lab working with his/her adviser on the selected research topic. The work of students can include reading articles, writing code or ideas, making experiments, etc. At the end of the research project the student should write a short document summarizing its contribution.
After the research project
At the end of the year, the Data AI team will organize a "poster session". Each student should prepare a poster that will be printed by the Data AI team. M1 students need to validate two research projects, the poster can dedicated to one or both projects.
Timeline for 2020 S1 research projects
Students should have a research project by Nov 1. (this means finding an advisor and having him send an email to the Data AI team by this date).
The project represents 10 full days of work, typically spread out at a pace of about one day per week. First semester projects must be completed by February 3, 2021 (note that, for the year 2020-2021 M1 students can do their two research projects in semester 2). Keep in mind that there are two weeks of break over Christmas, so you need to start the project by Nov. 9 at the latest.
Time frame for M1The master’s M1 program can optionally include an internship of at least 2 months and at most 3 months. The internship should take place in the second half of the study year. The internship should be defended before the 15th of August.
Time frame for M2The master’s M2 program includes an internship of at least 5 months and at most 6 months. The internship should take place in the second half of the study year. The internship should be completed during that study year. This means that the internship should start in the month following the exams of the first period (around February) and by 1st of June at the latest, so as to be completed by the 1st of October.
Finding an internship
The internship can take place either in a company or in a research lab, in France or abroad. Students who aim to do a PhD are encouraged to do their internship in a research lab, so that the internship could potentially lead to a PhD.
The student is responsible for finding an internship. The program coordinators (Goran Frehse and Louis Jachiet) will send internship offers via the mailing list. The reference institute (Télécom Paris) usually also organizes internship forums. Students who wish to do an internship in a research lab are also encouraged to contact the lecturers directly to see if internship positions are available.
The Internship Agreement
- The student makes sure she/he has a proper internship offer. The offer should be a document that contains at least the company name and address, the location and duration of the internship, the required skills/qualifications of the student, the title of the internship, and at least one paragraph about the expected work. The document should be in English or French (translations provided by the student are OK, if accompanied by the original document).
- The student makes sure that the topic of the internship falls broadly into the thematic scope of the master's program, and that the internship will be primarily on a single project (rather than helping out with different small tasks) of research nature. If in doubt, the student checks with the program coordinators before applying to the internship.
- The student applies to the internship, and, when accepted, proceeds with the following steps.
- The student fills out the form in the Synapse system. All tabs should be filled. If the form does not appear in Synapse, please contact Danielle (and put the program coordinators in Cc).
- The student finds a scientific advisor from Télécom Paris, who is different from the advisor of the internship. To this purpose the student should contact Data AI lecturers with knowledge on the topic of their internship. If the student fails to find a scientific advisor s/he can contact the coordinators for help. If the lecturer agrees to be the scientific advisor, the student informs the internship coordinator (currently Danielle Deloy) of the choice. This shall happen via an email that includes the name, the email address, the phone number, the professional address, and the institute of the scientific advisor, with the advisor in CC.
- The student sends the internship offer (see Point 1) by email to the program coordinators with the name of the scientific advisor.
- The host institution of the internship designates an internship advisor, who supervises the student during the internship. This person is typically an employee of the company where the internship takes place, or the researcher with whom the student wants to work in case of a research internship.
- The student fills out the internship agreement (a form with the title “CONVENTION DE STAGE”). The “Etablissement d’inscription administrative” is the university in which the student is registered (e.g., Paris-Sud or Télécom Paris). The “Organisme d’accueil” is the host institution. The “Stagiaire” is the student. The “enseignant référent” is the scientific advisor. The “tuteur” is the internship advisor. The program coordinators do not appear in this agreement.
- The student prints 6 (in words: six) copies of the internship agreement, and brings them to the internship coordinator for signing.
Doing the internshipThe student does the internship under the guidance of the internship advisor. The scientific advisor does not intervene, consult, collaborate, or co-organize. She or he mediates in case of disagreement between the student and the internship advisor. Before the end of the internship the student MUST contact the scientific advisor for the organization of the defense.
Defending the internship
The internship finishes with a report and an oral defense. Both have to be in English.
The internship report has usually 30-60 pages for M2 internships and 20 pages for M1 internships. The student sends the report to the scientific advisor around 1 week before the defense. The report should clarify the following points: the general context, the problem studied, your contribution, the details of the contributions, the arguments supporting their validity or the interest of your contribution. These different topics could be (but don't have to be) the sections of your report. They might not apply to your specific case.
The defense consists of a talk by the student of 20 minutes, followed by questions by the advisors. The defense is attended by the scientific advisor and other people if desired. The presence of the internship advisor is desirable. If the internship advisor cannot come, she or he shares feedback about the internship with the scientific advisor. In exceptional cases, the defense can take place via video-conference. The defense should take place not more than 3 weeks before the end of the internship, and not more than 1 month after the end of the internship. The defense has to happen during the current year of study. The student is in charge of organizing the defense. The defense takes place at the reference institute (Télécom Paris) or at the institute of the scientific advisor. The scientific advisor will help book a room.
The internship is graded by the scientific advisor in coordination with the internship advisor, by taking into account the quality of the work, the report, and the defense. Following the defense, the scientific advisor fills in the Internship Evaluation form on Synapses (for Télécom Paris lecturers) or sends her/his assessment to the internship coordinator by email. In addition, the internship advisor will be solicited separately for feedback by the internship coordinator.
The internship is validated if the grade of the internship is at least 10/20.