Summer school on
Deep Learning and
Bayesian Methods

August 2018, Moscow, Russia


Bayesian Methods Research Group

Summer School Scope & Goals

At Deep|Bayes summer school, we will discuss how Bayesian Methods can be combined with Deep Learning and lead to better results in machine learning applications. Recent research has proven that the use of Bayesian approach can be beneficial in various ways. School participants will learn methods and techniques that are crucial for understanding current research in machine learning. They will also have an opportunity to draw connections between Bayesian Methods and Reinforcements Learning, learn modern stochastic optimization methods and regularization techniques for neural networks. Lectures will be followed by practical sessions.

Audience

Our summer school is intended for:

  • Undergraduate students (preferably those who completed at least 2 years of study) and graduate students with strong mathematical background and decent knowledge of machine learning including deep learning
  • Researchers and industry professionals working in machine learning or related areas, willing to expand their knowledge and skills

Participation prerequisites

  1. Strong knowledge of machine learning and familiarity with deep learning.
  2. Mathematics: proficiency in linear algebra and probability theory is highly desirable.
  3. Programming: Python and NumPy.
  4. English will be the language of Deep|Bayes 2018 summer school, so participants are expected to be comfortable with technical English.

What will I learn at the Deep|Bayes school?

  • Why are Bayesian methods useful (in machine learning and everyday life)? What is randomness, indeed?
  • Frameworks and libraries for neural networks. How to code a neural network in 10 minutes?
  • Latent Variable Models. How to train a model able to discover patterns unknown before training?
  • Scalable probabilistic models. Why can it be useful to turn a probabilistic inference problem into an optimization problem?
  • Why do neural networks need an attention mechanism? How to generate a caption for a picture?
  • The connection between Reinforcement Learning and Bayesian methods. How to train stochastic computational graphs?
  • Automatic Dropout rate tuning. Do neural networks overfit? (spoiler: well, yes)
  • Stochastic optimization. How to optimize a function faster than computing its value at one point?

Our goal is to show that using the Bayesian approach in deep neural networks can expand the range of their applicability and improve their performance. Although there are many different problem settings in machine learning, probabilistic inference for Bayesian networks can be performed in a similar way for most of them. Do you want to learn more about these methods? Come to our summer school!

Preliminary dates

The following is a very preliminary list of dates based on our experience with the previous run of the school

  1. Early February: Applications open
  2. Early April: Applications close
  3. Late May - Early June: Acceptance notification
  4. Late August: Summer School

FAQ

  • How do I apply?

    The most up-to-date information about applying and registration is available here. If you have any additional questions, please, email us at info@deepbayes.ru

  • What is the language of the school?

    The working language of the school is English (unlike the previous run, which was in Russian). All the lectures and practical sessions will be held in English, so the participants will need sufficient proficiency in technical English.

  • What is the participation fee?

    The fee is 1000 rubles for university students (if enrolled by August 2018), 10 000 rubles for academic researchers and 30 000 rubles for participants from industry.

  • If I pay the fee, does it automatically mean I can participate?

    No, candidates will have to complete the application form. The participants will be selected on a competitive basis. Only approved candidates will be asked to proceed with paying the fee. The selection process is necessary because we have a limited number of spots available, and we also want to make sure that our participants have a suitable background to understand the material and have a useful experience at Deep|Bayes School.

  • What is the selection process like?

    Candidates fill an application form, write a short essay and complete a few small assignments to demonstrate their knowledge of machine learning and programming skills. The selection committee will carefully review all the applications and will proceed with the strongest candidates based on their technical skills and motivation.

  • How competitive is the admission?

    Last year we received more than 300 applications for 100 slots (however, for this year the number of available slots may be different).

  • Will meals be provided?

    We will provide coffee breaks and lunches.

  • I do not know much about machine learning, but I would like to learn more. Should I participate in Deep|Bayes school?

    Unfortunately, no. Our summer school is intended for candidates who have strong knowledge of machine learning and are familiar with neural networks. Such level of participants enables us to delve into to technical details and quickly proceed to advanced topics while being confident that the material is accessible for everyone. For learning machine learning fundamentals, we encourage you to check out numerous resources available on the web.

  • What will I get from participating in the School?

    You will learn modern techniques in deep learning and discover benefits of Bayesian approach for neural networks. Topics discussed during the School will help you understand modern research papers. And, of course, the School provides an excellent opportunity to meet like-minded people and form new professional connections with speakers, tutors and fellow school participants.

  • I live outside Moscow. Can I still participate?

    Yes, if you get selected. We will provide a limited number of travel grants partially covering a trip to and from Moscow and accommodation for the time of the summer school.

  • What about lectures live streaming?

    We won’t have live broadcasting, but we will upload video lectures during and after the school.

  • I have more questions!

    Please, email us at info@deepbayes.ru.

Speakers and Lecturers

Most of our lecturers and tutors are members of the Bayesian Methods Research Group and researchers from the world’s leading research centers. Many school lecturers have published at top international machine learning conferences such as NIPS, ICML, ICCV, CVPR, ICLR, AISTATS, and others. Today, the Bayesian Methods Research group is one of the leading machine learning research groups in Russia. Members of the group have developed a range of university courses in Bayesian Methods, Deep Learning, Optimization and Probabilistic Graphical Models and have substantial teaching experience. Our research group is actively collaborating with such companies as Samsung Electronics, Yandex, SberTech, Kaspersky, Schlumberger, and others.

How to Apply

Application will open soon. Please, stay tuned! If you wish to receive a notification about application start and other important announcements, please leave your contact information in the form.

Selection of participants is highly competitive. Last year applicants had to complete a small programming assignment and write a detailed review on a machine learning research paper of their choice

Partners

We welcome financial support and sponsorship from any industry or academic organization. Funding is needed for:
  • Meals and coffee-break snacks
  • Travel grants for students outside Moscow
  • Swag
  • A networking event
Have you any questions regarding the Deep|Bayes School support, please, email us at info@deepbayes.ru.

Contact

If you have any further questions about our summer school, or you wish to become our partner, please send an email to info@deepbayes.ru.

You can also follow us on Twitter.