Master student internship in machine learning for smart water alerts

 (via EURAXESS)
Centrum Wiskunde en Informatica (CWI)
Amsterdam, Netherlands
Position Type: 
Internship
Organization Type: 
University/Academia/Research/Think tank
Experience Level: 
Not Specified
Degree Required: 
Bachelor's (Or Equivalent)

EXPIRED

Please note: this job post has expired! To the best of our knowledge, this job is no longer available and this page remains here for archival purposes only.

Project Slimme Water Waker (SWW) in cooperation with Seita B.V.

Background
Sustainable use of drinking water is essential to society. Water resources are under increasing pressure from growing demand and climate change. Dutch water utilities are looking to raise awareness and to find potential ways of saving drinking water. By rolling out decentralised sensors (smart meters), they aim to be able to track patterns in consumption data of drinking water to discover unusual events such as leaks. Unlocking this valuable information is a slow and labour-intensive process, whereas an early-warning system would require real-time automated and predictive analytics.

Task description
This project is exploring an innovative decision support system that combines automated analytics, predictive analytics and user feedback. We want to enrich an existing monitoring platform with the ability to provide labelled history reports and alerts to business with smart water meters. The aim of this project is to help consumers of drinking water to identify problems, such as leaks and inefficient consumption, to save water. The project is a collaboration with the CWI spin-off Seita B.V. in Amsterdam.

The internship investigates data-driven recognition of unusual patterns, and might be applicable as content for a Master’s thesis. The intern will closely collaborate with members of the research group and Seita.

The internship will work towards a machine learning model that detects and classifies unusual patterns in time series data. The work combines both unsupervised and supervised machine learning techniques that should be implemented in Python and tested in simulation.

The candidate’s work will provide input to a toolbox that supports real-time automated and predictive analytis in a purely data-driven fashion. In particular, the toolbox should label unusual patterns and be able to process user feedback on the validity of those labels. To this end, the intern will actively consult with Seita to gather necessary input for simulation scenarios.

The research will contribute to the research group's ambition to valorize fundamental principles that find applications in practical intelligent systems. The project is embedded in a research group with experience and ongoing complementary projects on topics such as multi-agent learning, anytime online planning and control, smart energy systems, etc., and successful applicants will join a diverse team of motivated junior and senior researchers. Group’s alumni have moved on to excellent positions in academia and industry, making it a cornerstone in their career path.