On June 12, the Great Mathematical Workshop (GMW) started. GMW is a complex of two weeks of intensive work on various projects, organized by the Mathematical Center in Akademgorodok in Novosibirsk. This year the event is also organized by the Higher School of Economics, the Regional Scientific and Educational Mathematical Center of Tomsk State University and the Euler International Mathematical Institute.
In its framework students and scientists develop projects in such fields as medicine, urbanism, high technology, agriculture and many others. The projects are based on inquiries of business and scientific customers who turn to the workshop to find new solutions for their problems. Twenty-nine projects from various customers are presented at the workshop this year.
During the workshop, participants work in groups on their tasks and share experiences and results with colleagues in general plenary sessions and common lectures. This year, all projects were divided into six clusters on similar topics: machine learning, algebra and logic, geometry and topology, optimization, modeling, mathematics and life. In addition to teamwork, GMW hosts specialized lectures for each cluster.
, director of the Mathematical Center in Akademgorodok, explains the idea of the workshop:
The American Mathematical Society classifies the hierarchy of mathematical knowledge into three levels, ranging from global topics to narrow subject areas. The higher we climb the tree of knowledge, the more difficult it is even for professionals to understand what's happening. It is even more difficult for people who seek to use math in other areas of science. We want to make mathematics closer to people.
The main task of the workshop is to solve the problem of communication between mathematicians and specialists from other fields and thereby make mathematics accessible for solving problems from different fields.
Each project is a whole separate planet, whose inhabitants speak their own language. The interaction of different scientific worlds with each other, according to the organizers, will accelerate not only the implementation of the project, but also the change of perception of their subject areas.
Evgeny Vdovin adds:
We tried to divide projects into clusters in accordance with their topics and work tools. That way they have common points of communication, which they can use to they can start talking to each other. Now we have come to the conclusion that forming clusters on a thematic basis may not have been the best idea, but we already have new ideas for further workshops.
Projects of the workshop
The main task of urbanists is to solve the problems of the urban environment, interacting with residents of nearby areas. The goal of the project is to automate the process of collecting requests for changes in the urban environment — thus make it clearer. The project curator Anna Avdyushina is assembling the team for the second time. Last time, on the First Workshop of the Mathematical center in Akademgorodok, the team tried to find the boundaries of buildings on satellite images to simplify the process of creating maps. This year, the team focused on creating a map of disadvantaged areas of Moscow using the most common names of streets, shops and organizations that are found on the map of the outskirts of the city. All the collected information can be easily used at the research level to analyze any city in Russia. The participants conducted a survey, according to the results of which they found out that pubs, cafes, bars, microloan offices and hairdressers with female names are not associated with well-being among residents.
Machine extraction of meaning from text and its application in the tourism business
The goal of the project is to create a chatbot that can partially replace a consultant in the tourism business. The algorithm will have to extract information from the text, analyze and offer the client vacation options that match his wishes and budget. In the future, the bot will be able to work with the client's objections and offer alternative options. The selection will take into account hotel ratings and personal circumstances: the presence of children, time and type of rest, diet. In the first week of GMW, participants tried classic machine learning. In the future, they plan to use NLP technologies and turn to modern machine learning technologies.
Hunting Space Accelerators: Finding Ultra-High Energy Photons Using Machine Learning
The project, the participants of which analyze the particles formed as a result of the explosions of stars. Merging neutron stars and supernova explosions generate a lot of energy. Particles that carry ultra-high energy are called gamma quanta. Scientists are not yet sure how to find them. Machine learning should help learn how to distinguish these particles from atmospheric protons using 500 million records of the KASCADE air shower detector. An air shower is a stream of particles arising from the decay of cosmic ray photons from their collision with nitrogen and oxygen nuclei in the Earth's atmosphere. The results of the experiment will help physicists better understand the nature of the origin of cosmic radiation and identify the sources of gamma quanta in the celestial sphere.
OverGrower Digital Farms
The goal of the project is to create a simulation of a tomato in an automated hydroponic plant to facilitate testing of environmental conditions on it. The development will help predict the behavior of the plant depending on what happened to it. According to the participants, with a fairly large database, it will be possible to use the formula to build a digital model of many other living organisms. In the future, the project participants will write a neuro-assistant program that will help agronomists take into account environmental parameters.
Automation of recognition of a defect in printing on a 3D printer
Among the participants were applicants from the Engineering School of Mechanics and Mathematics of NSU, who could take part in this project as entrance examinations. The most common mistakes in 3D printing are parts detachment from the support table, violation of the anchor point and plastic feed. To avoid this, they came up with a system for alerting the operator about an error and a stop in printing. The work program was created by the end of the first week of the workshop, but it automatically highlighted all white plastic models as defective. Participants plan to add a color correction layer to the program and move on to a different algorithm.
We talked with the participants and curators of the projects about how the first week of the workshop went and what their impressions were.
, curator of the Digital Urbanism project, says:
This time [in comparison with the last workshop] we talked more, discussed and built the team's work. Everyone worked in their own directions and ideas. Now in the chat we are discussing how we will work on the intermodule. This year we have a very strong team.
, organizer, comments:
One of the most important tasks of the workshop is to teach people to speak about their projects clearly, but not primitively. Do not lose the depth of the content, but at the same time remain understandable for people who do not work at this depth. This is not an easy task. We continue to analyze which tools and activities help and which hinder building a dialogue.
, expert of the project “Analysis of medical images”, reflects:
I am pleased that the students were interested in the topic. A potent amount of work has been done, the topic has proven its relevance and interest. When you share knowledge with people outside your field, fresh associations arise. Working at the intersection of sciences is always interesting.
, curator of the project “Automation of recognition of printing defects on a 3D printer”, shares her impressions:
All participants were very active and asked questions. After the first plenary session, we even adjusted the work plan. So it seems to me that the bridge between the projects has finally begun to form.
The Seminar of the Impossible
As part of the workshop, the "Seminar of the Impossible"
was held. Participants discussed the future of mathematical education in school and university curricula. One of the organizers, Timur Nasybullov
, shared his impressions:
It seems to me that at least ten more such discussions are needed in order to come to understandable results. I am a mathematician, I write articles, I work in a field that is difficult to explain, and I understand that no more than a hundred people on earth can understand it. Studying mathematics allows me to better teach it to students; if I, as a scientist, do not discover anything important, one of them will be able to do it.