One of Machine Learning UB’s missions is to create opportunities for the people to work on real world data (Mongolian data). Under this mission, we collaborate with Mongolia’s leading data-driven companies and organize data competitions on Kaggle platform. Organizing data competitions have the following significance.
Support open data in Mongolia
Experience real word data analysis and data based solutions
Improve interpersonal skills and learn to work in an online space
Challenge yourself with data driven decision making
Showcase your skillset, experience and knowledge
Learn from the industry experts
Ulaanbaatar city’s suburbs mostly consist of housing called yurt or ger. Ger is a traditional Mongolian housing that has round wooden structure with white outer covers which makes it distinctly identifiable from space. Ger districts largely contribute to the air pollution of the city and there have been attempts to count the number of gers around Ulaanbaatar city but the exact number still varies and undetermined. Therefore, using AI and satellite images, we can count the number of gers with less time and resources.
Mongolian is considered a low resource language and machine trainable open data is scarce. To promote Mongolian language open data, MLUB has collaborated with National University of Mongolian and Pocket app to organize the first natural language processing (NLP) competition. The task is to build machine learning or deep learning models to predict the synsets of the selected words.
The competition is organized among the Deep Learning UB Summer School 2021 participants. MLUB collaborated with DLUB and presented a computer vision problem for the participants to further their learning experience. The data competition problem was to predict the names of 25 different dishes from around the world.
A car plate number recognition system is commonly used in many companies in Mongolia, however, the most Mongolian organizations use systems offered by foreign companies. Therefore, to raise the awareness and promote the skillset of Mongolian engineers, MLUB put forward a computer vision task to predict Mongolian car plate number.
Ulaanbaatar is one of the most polluted cities in the world. As an effort to reduce air pollution, the government has installed 13 air quality measuring stations around the city and has been using traditional methods to tackle the problem. MLUB collaborated with the UNDP, UNICEF and PIN to offer data based solutions by utilizing the data from the stations and showcase how modern methods can assist to solve the problem.
As a part of “Data Nomads: AI & DS Conference”, the 2-stage competition was organized. The lack of public transportation quality and traffic congestion are one of many problems that Ulaanbaatar city has. In the second stage, the competition task was to predict the number of people in a bus at a given station. The overload of passengers leads to hours of waiting and uncertainty. Therefore, it’s important to offer information about the bus overload, so that the people can make better decisions and plan their lives.
As a part of Data Nomads: AI & DS Conference, the 2-stage competition was organized. The lack of public transportation quality and traffic congestion are one of many problems that Ulaanbaatar city has. In the first stage, the competition task was to predict the arrival time of a bus at a given station. Due to lack of technology adoption in the public transportation of Mongolia, the passengers sometimes wait hours for their bus to arrive. Therefore, a smart system that can tell a wait time of a bus will benefit the citizens greatly.