Admission to the Department

On this page, you will find information about the admission process to the Intelligent Systems Department at MIPT, including details for individual course attendees, the interview procedure for undergraduates, the master’s program admission process, and research work at the department.

For Individual Course Attendees

  1. Come, listen, and participate.
  2. All information about current courses is available in the Telegram channel.
  3. Grade sheets for the department are processed through the dean’s office—submit them there.
  4. Courses require preparation, therefore:
    • confirm with instructors that you can join the course,
    • during the first few weeks, determine whether you will be able to pass the course.
  5. Find out what is required to pass the course.

Admission Process for 2nd-3rd Year Undergraduates

  1. The student fills out the application form,
  2. Solves one of the trial tasks listed below (fourth-year students present their thesis work),
  3. Forms an opinion about the research conducted by department students (Theses, Research Reports),
  4. On the scheduled date, gives a three-minute presentation about the task and about a topic or project that interested them,
  5. Receives the department’s decision that same evening,
  6. Awaits the dean’s office directive on department assignment.

Trial Tasks Fall 2025 and Earlier

Fall 2025

Task 64

Visualize the Gerchberg–Saxton algorithm for two-dimensional images. Explain the two-dimensional Fourier transform, the source and target spaces, and the convergence within them.

Task 63

Explain methods for the numerical solution of ordinary differential equations with illustrations using real measurement data. It is desirable to include the Adjoint State Method.

Task 62

Explain the Galerkin method using an example with real data. Replace the linear model with a two-layer neural network. Show the difference in the solution. Visualize the data and the model.

Task 61

Explain the finite element method with an example using real data (elasticity equation, Poisson's equation, other partial differential equations of your choice). Visualize different approaches, for linear models and neural networks.

Task 60

Illustrate the dimensionality reduction algorithm in the Galerkin method for solving the Navier-Stokes equation with an illustration using real or synthetic data.

Task 59

Illustrate the dimensionality reduction algorithm on two-dimensional computed tomography data, using a linear model (and a neural network by choice).

Task 58

Explain forecasting a segment of a time series using the method of Singular Spectrum Analysis (Russian version). Analyze the original dimension of the Hankel matrix and its reduced dimension.

Task 57

Explain how higher-order Fourier transforms (with tensor data representation) differ from the one-dimensional Fourier transform. Illustrate using a low-resolution animation (or similar data; with forward and inverse transforms including quality reduction).

Task 56

Compare classical methods for the numerical solution of partial differential equations (elliptic, parabolic, hyperbolic) and neural network approximations of solutions using real data.

Task 55

Given a smartphone IMU time series (accelerometer, gyroscope). The phone is placed on a person's chest. Propose an algorithm for decomposing almost-periodic oscillations (pulse, respiration, random movements).

Task 54

Given two time series, determine if they are causally linked (causal inference) using the Convergent Cross Mapping method. Analyze the dimensions of the spaces.

Tasks from Previous Years

Tasks from previous years can also be used for the presentation.

Tips for Solving Tasks (these are suggestions, not requirements)

List of Topics for Discussion

Compiled in the Topics List section on this page

Selection Results

Master’s Program Admission Procedure

  1. The student fills out the application form,
  2. We gather candidates wishing to join the department upon their request,
  3. We review the application, thesis work, ask about their opinion on topics and student research at the department, and ask questions from the mathematical section of the bachelor’s program (Sections 1, 2, 3 from “Pen and Paper Exercises in Machine Learning” by Michael U. Gutmann)

The purpose of the interview is to understand:

  1. What contribution the student will make to the department’s research,
  2. Whether the student’s qualifications match the level of planned master’s dissertations.

Research Work at the Department

Your research qualifications are your primary goal at the department. The department focuses on the mathematical foundations of machine learning. This means that your work focuses on developing theoretical methods. You obtain theoretical results. Practical applications to illustrate your results in computational experiments may be arbitrary.

Scientific collaboration and exchange of ideas are the foundation of research work. Do not hesitate to write to researchers and tell them about yourself, your results, and your plans. The purpose of your research advisor is to show you the right direction for research, provide a topic, and formulate a problem in general terms. Keep in mind that the advisor works in their own field. How can you learn about their field? Go to scholar.google.com, enter the advisor’s name, and review their work from recent years. Ask your advisor for a research task.

Requirements for a research advisor:

  1. Doctor of Science (D.Sc.) or Candidate of Science (Ph.D.),
  2. Works in the department’s field and maintains contact with the department.

The department cannot assign you a research advisor. You must find one through communication and agreement on joint work. When searching for research advisors, tell them about yourself: your qualifications, achievements, research papers, and plans.

Tip: Look for research advisors in research groups. If you are interested in:

Find members of these research groups. Look at the topics of their recent research papers. Compile a list of co-authors. Review the attached PDF. Write to potential advisors. Important tip: Connect with students and graduate students in your field. Ask them how they organize their work with research advisors.

For quick presentation of initial results, try to participate in student conferences (example).

What You Need to Do

  1. Look at who has been a successful research advisor:
  2. Consult with students and graduate students. Read the guidelines.
  3. Find a research advisor (matching your thinking style and interests).
  4. Ask them to contact the department and approve your research topic.
  5. Begin working with them and present the results during exam week at the beginning of the session.
  6. Reporting format: paper, presentation, code.

Timeline

  1. Third-year students search for research advisors during the spring semester and report this at the research credit.
  2. Fifth-year students have a research advisor at the beginning of the fall semester (we ask about plans).
  3. Graduate students have publications with their research advisor.

Tip: Changing your research advisor and topic at any time is normal—you are developing and priorities change. It is advisable to complete projects so you have something to show. Working with multiple research advisors on different topics is wonderful (if you can manage it). Working in teams is excellent—your personal contribution is what matters.