PhD Research Seminar: Machine Learning for Indoor Scene Classification and Forecasting Energy Consumption

First talk: Indoor Scene Classification and Recognition with Machine Learning   
Speaker: Ismail Kayali, third-year PhD student, Faculty of Computer Science

The growth of multi-million-item datasets have allowed machine learning algorithms to give near-human semantic classification of visual patterns such as objects and scenes. Indoor scene classification becomes incrementally harder when progressing from pixels to objects and then scenes. With its high coverage and high diversity of models, we will be able to implement a system of visual context to control the progress on currently intractable optical recognition difficulties. Such difficulties could involve determining the movements occurring in a given scene, detecting mutable objects or human behaviors for a particular area.

Second talk: Forecasting Energy Consumption Based on Automated Machine Learning     
Speaker: Konstantin Danilov, third-year PhD student, HSE MIEM

An automated machine learning pipeline will be considered in relation to the problem of forecasting energy consumption. Special attention will be paid to the feature engineering step. We will present a comparison of the main automated feature engineering methods.