ADS Sponsored PPP Research Assistant Projects
Four proposals for joint Research Assistant projects have been approved. This is the latest initiative from ADS to encourage collaboration between academia and industry on new and innovative research areas. The goal being to lead to larger, longer-term research collaborations.
The ADS “Research Coupons” were announced at the ADS Highlight Event in December 2019. The coupons are intended to encourage new collaborations between academia and industry. ADS provides €5000, for a research assistant to work in an academic-industry partnership. The remaining amount to employ the research assistant will be covered by the industry partner.
Four proposals for the ADS Research Coupons have been approved:
1. Commercial Waste in the City – Gemeente Amsterdam, SKIAlabs and HvA
Small companies and organisations in the City of Amsterdam disposing of their waste have two options to do so: contracting a private waste collector or making use of the household waste system. Many of these small companies and organisations choose the latter. In exchange for a municipal fixed fee (the so called “Reinigingsrecht”) they can make use of household waste facilities for a maximum of nine bags of garbage per week.
What is the impact of Reinigingsrecht in the household and commercial waste system?
It is presently unknown what the exact impact of commercial waste disposal via the system of Reinigingsrecht is on both the household waste system and the commercial waste system. This inhibits effective policy making in order to improve efficiency and environmental impact and prevent abuse.
The research assistant will closely collaborate with a group of researchers from the HvA and SKIAlabs to work on data collection and modelling of commercial waste collected by private waste collectors.
2. Large-scale Battery Simulation System – GigaStorage and HvA
This project will focus on data science in the context of the largest battery system that will be built in the Netherlands this summer. This battery (NEC) has a capacity of 12 MW and will be placed at Engie’s Windnet farm. The goal of the project is to retrieve different datasets from the batteries itself, the windmills and the transformer (Tennet) and financial data.
3. eXplainable AI in Healthcare – myTomorrows and VU Amsterdam
The inability of AI systems to explain their decisions affects their credibility and trustworthiness, leaving AI-based methods taking decisions for life-critical problems such as healthcare problematic. This calls for a new paradigm of intelligent systems that can deliver understandable results along with transparent, reliable explanations.
This research will study how to use Knowledge Graphs (KGs) to explain recommendations to a human-users. The goal is to use knowledge representation and reasoning as a framework to enable intelligent systems to explain their decisions in a way that humans can understand.
This will reduce the workload of medical experts; improve patients’ lives, with recommendations being provided faster; and improve trust toward AI tools, as explanations will be generated from trusted sources, in a more understandable and transparent way.
4. Applying Techniques for Object Detection – VodafoneZiggo and HvA
When using machine learning techniques, such as Convolutional Neural Networks, large amounts of labelled training data are needed to train the models. From an object classification problem, some methods using meta-learning, in the form of few-shot learning, have already helped to tackle this issue.
The research assistant will help with the research and development of a Deep Neural Network to combine meta-learning with object detection. The main goal is to develop a system that can detect appliances in customers’ fuse-boxes and street cabinets of VodafoneZiggo using fewer images.
The project will use a state-of-the art technique for object detection YOLO – You Only Look Once – and combine it with meta-learning, using methods such as MAML or Reptile.
Please contact us if you would like more information about any of these projects