Course: “Statistical learning: bagging, boosting, SVM, introduction to neural networks”

Within the framework of 2019 International Conference on Applied Probability and Statistics – CAPS 2019 (https://caps2019.viasm.edu.vn/), the Vietnam Institute for Advanced Study in Mathematics (VIASM) organized the mini-course “Statistical learning: bagging, boosting, SVM, introduction to neural networks” on April 1-2, 2019 in Hanoi. The lectures was given by Professor Vincent Lefieux (RTE - Réseau de Transport d'Électricité, France). Professor Vincent Lefieux has been the head of a research group on Data Science at Réseau de Transport d'Électricité (subsidiary of Électricité de France S.A) since 2017. He is a frequent visitors of VIASM in the last 4 years, regularly lecturing, presenting seminars, participating in conferences and meetings at Vietnam Institute for Advanced Study in Mathematics for many years.

The mini-course attracted about 50 participants from various universities, research institutes, and businesses in numerous fields, including Economics, Banking, Healthcare, Aviation, Engineering… Aside from attendees from Hanoi, some other attending individuals were from numerous provinces in the Central and Southern Vietnam such as Hue, Da Nang, Ho Chi Minh city, Quy Nhon,… the Vietnam Institute for Advanced Study in Mathematics provided financial support for 11 attendees from the Central and Southern Vietnam.

Course0104.JPG

Following the series of lectures on Statistical learning implemented since 2018, the Course continues to update knowledge on Statistical Mathematics, which is the theoretical foundation of Ensemble Learning method.

Statistics practitioners often compare Ensemble Learning method to the anecdote “The Six Blind Fortune Tellers and an Elephant”. If the partial analyses of each person in the anecdote are collected and combined, the end result is the precise image of the elephant. In mathematical aspect, ensemble learning is a set of statistical techniques aiming at increasing accuracy and stability of estimation (also known as robustness of the model), decreasing errors (noise, variance, bias).

The lectures focuses on presenting fundamental Statistical techniques in classification problem, namely:

  • Bagging (Boostrap Aggregating) in Random forest models
  • Boosting
  • Support Vector Machine (SVM)
  • Neural Networks.

Further information about the minicourse is  available at at:  http://viasm.edu.vn/hdkh/bbsn.