Course on Machine Learning from Mail.Ru Mail

    September 27 begins a machine learning course from Mail.Ru Mail . Classes will be held twice a week at the Mail.Ru Group office for three months. Registration is open to students of Moscow universities.

    During the course, Mail.Ru Mail and Antispam specialists will talk about ML technologies, which are used to make Mail an even more convenient and modern product. Under the cat details about the course: format, program, authors and prospects for graduates.


    The course consists of 20 classes : 18 lectures and 2 exams (intermediate and final). Each lesson contains theoretical and practical parts; in the course of the practical part, students will develop several classifiers, thoroughly work with the text and implement basic algorithms, the understanding of which is necessary in the work.

    Participants will also receive 5 practical homework. According to the authors, “the complexity will be chosen so that it would be necessary to break the head over the task, but not to the detriment of university studies”.


    1. Regression from one variable and several variables. Retraining and ways to deal with it.
    2. Logistic regression. Binary and multi-class classification. KNN classifier.
    3. Support vector machine.
    4. Work with the text: preprocessing and vector representation, classification tasks.
    5. Thematic modeling: pLSI, LDA.
    6. Vector text view: word2vec, fastText.
    7. Algorithms for reducing the dimension: PCA, LSH.
    8. EM-algorithm, k-means and s-means algorithm.
    9. Hierarchical clustering algorithms. Clustering evaluation metrics.
    10. Decision trees Tree ensembles: random forest.
    11. Gradient booster over trees: xgboost.
    12. AB testing.
    13. Interval quality assessment of classifiers in production.

    Classes will be held twice a week (Tuesday, Thursday) from 18:00 to 21:00 at the Mail.Ru Group office from September 27 to December 11.

    The course is designed to explain in more detail how machine learning is used in real-world tasks. Unlike most ML courses, this one focuses on practice, and not on the academic component. The main goal of the authors is that after passing the course, students can solve typical practical problems.

    Internship after training

    Graduates will be offered paid internships in the machine learning commands of the Mail and Antispam, where ML methods are used to solve such problems as filtering unwanted traffic, determining hacks, separating important letters from unimportant ones, classifying the meaning of letters, etc. Video about the Mail team.

    Trainees work on combat missions, their only difference from experienced colleagues is the ability to build a flexible schedule, working 20 hours a week. You can get an internship only after the completion of educational projects of the Mail.Ru Group.

    How to enter the course

    The course is designed for undergraduates and graduate students of technical universities , studying in the mathematical or physical-technical direction.

    To get to the course, you must register by link until 10:00 am (MSK) on September 22 and go through online testing any time from 10:00 on September 22 to 10:00 on September 24. The link to the test will come to the email address specified during registration, and will be available within two days. Classes will begin on September 27th .

    Required minimum for admission:

    • knowledge of linear algebra and probability theory;
    • ability to program in Python;
    • familiarity with Numpy and Sklearn;
    • plus will be the knowledge of optimization methods.

    Materials for preparation:

    1. Books: Zorich “Mathematical Analysis”, Yu.V. Prokhorov, L.S. Ponomarenko "Lectures on probability theory and mathematical statistics", Borovkov A.A. "Theory of Probability", Gmurman V.E. "Theory of Probability and Mathematical Statistics";
    2. course of lectures "Neural networks in ML" on the Tehnostrim training channel ;
    3. useful info on python .

    The first lesson will be held next week. Register for a course, go through training and join the ML industry!

    Also popular now: