MORE | Management of Real-time Energy Data - Second MORE webinar: Incremental machine learning models for high-frequency big data
With the lowering costs of sensors, high-volume and high-velocity data are increasingly being generated and analyzed, especially in IoT domains like energy and smart homes. Consequently, applications that require accurate short-term forecasts and predictions are also steadily increasing. This increasing amount of data coupled with frequent changes in patterns, requires the utilization of incremental models and smart data management for machine learning models to ensure accurate predictions on real-time data.
This talk will broadly cover three components of machine learning models for high frequency, high volume data:
Description of SAIL (Streams And Incremental Learning) library for dynamic AutoML of incremental models.
Prequential Model Selection using Performance Gradient-based Saliency Maps for explainable AutoML.
Machine learning models on compressed data.
Dr. Seshu Tirupathi is a Research Scientist in IBM Research Dublin since 2014. His interests include incremental machine learning algorithms, big data, cloud computing and scientific computing. Over the last 8 years, Seshu has made significant contributions to multiple domains like RES, water networks and electricity networks through fundamental research as well as consulting projects. Prior to joining IBM, Seshu was a member of the Scientific Computing group at Brown University, where he earned his PhD in Applied Mathematics. He also holds an M.Eng. in Mechanical Engineering from Cornell University and a B.Tech. degree in Aerospace Engineering from Indian Institute of Technology Kanpur.
Date: Friday 9 June, 11.00 EET
25 May 2023