Abstract:

Both complex and dynamic evolving nature of time series data make forecasting among one of the most important and challenging tasks in time series analysis. Typical methods for forecasting are designed to model time-evolving dependencies between data observations. However, it is generally accepted that none of them is universally valid for every application. Therefore, methods for learning heterogeneous ensembles by combining a diverse set of forecasts together appear as a promising solution to tackle this task. Ensemble construction is based on several steps including base models generation, pruning and aggregation. The goal of this talk is to present the most promising R packages/solutions for automating adaptive ensemble models construction for online time series forecasting by reducing the human intervention in the ML-loop

Short Bio:

I am Amal Saadallah and I work as a research assistant in the artificial intelligence group of the TU Dortmund, within B3 subproject (Data Mining on Sensor Data of Automated Processes) of the Collaborative Research Center SFB 876 (Providing Information by Resource-Constrained Data Analysis). My main research work is about adaptive ensemble methods for online learning, time series analysis and forecasting and sensor-simulation data mining. I am also investigating industry 4.0 related applications, more specifically for machining processes. Hence, one of the major research directions in B3 is the combination of machine learning methods and process simulation systems for predictive maintenance and quality control of milling processes.

My list of publications can also give you the big picture of my research activities:

https://www-ai.cs.tu-dortmund.de/PERSONAL/saadallah.html

Click here for more info.