With the aim of sharing some content related to the research, teaching and learning of data science and artificial intelligence, on the afternoon of August 20, 2022, at room B2.108, Faculty of Mathematics and Science organized a Seminar with the theme “Some issues in data science research, teaching and learning”.

Seminar is honored to welcome guests who are Mr. Nguyen An Khuong and Dr. Nguyen Tien Thinh, along with excellent students from the Faculty of Computer Science, University of Science and Technology – Vietnam National University, Ho Chi Minh City. In this seminar, the guests gave interesting talks about their recent research works.

The first topic was  “Reducing noise by deep learning” presented by Nguyen Long Vu, currently an AI research engineer of FPT company.

In recent years, especially during the complicated epidemic situation, activities through online platforms have become more and more popular and play an important role. The quality of online activities is often affected by the surrounding environment of the individuals attending. One of the important factors is noise. To limit this, people often use commercial software or expensive noise-cancelling headphones. However, not everyone can access these products. With the goal of providing a free, convenient solution to improve the quality of online activities, Nguyen Long Vu has researched and successfully built an application that allows filtering noise for online activities. online with many advantages. In his research, the author has detailed studies related to different aspects of sound. At the same time, flexibly use Mathematical tools such as Fourier transform, as well as modern methods in the field of Deep Learning such as LSTM to solve the posed problem.

Another equally interesting topic is “Study on topic model using Latent Dirichlet Allocation (LDA) model on tourism evaluation data” submitted by friends Le Thanh Son and Tran Dinh Vinh Thuy, currently a Master 2 student at the University of Limoges.

In Vietnam as well as many countries around the world, tourism is one of the fields that make an important contribution to the national economy. Identifying tourism trends in order to have appropriate policies to develop tourism is one of the necessary and important needs. To achieve this goal, there are many approaches. One of them is to identify trends based on customer experiences, shares and reviews. Along with the development of media, tourist experiences and reviews can be done online. These reviews are a great and rich source of data. Exploiting and effectively using this data source is one of the important and necessary needs for managers.

The LDA model is a statistical model that allows to discover hidden topics in a corpus of many different documents. This is the most common model used in topic classification due to its fast computation and high accuracy. In their thesis, the authors studied in detail the mathematical basis of the LDA model and applied it on the user review dataset collected from TripAdvisor to identify the hidden topics in it. . Through these topics, the authors conducted analysis to understand tourism trends and found many interesting results.

The seminar attracted the attention of many teachers and students of the Department of Mathematics and Statistics.

——-

B.L.Thuy

News and photos: Faculty of Mathematics – Statistics