Sessions
Per starting time | Per room | Session contents
Per starting time
With one exception, all sessions have a duration of one hour or less
Per room
Alphabetical, with summary
Access to farm data by to the Code-of-Conduct Data Sharing
Tamme van der Wal
Are you also struggling with farm data for your project? The Data Act organises data sharing, emphasising the prevention misuse. The Code of Conduct Data Sharing Agrifood now offers a strong framework for data sharing and (re)use of farm data and is compliant with this Data Act. For research into real farming, and using real farming data, this is an indispensible instrument to do the right things, and to do things right. The workshop will explore the opportunities for scientists and projects, and will show use-cases with the data services platform Farmmaps as a facility that supports this data sharing.
09:30-10:30 in Theatre Cerise
Baby Steps at Twente: How to write a Software Policy for 4344 people
Efe Sözeri
In this session, Efe will share his first 6 months experience as the Software Steward at the University of Twente. With the Policy almost published, a Guideline on the way, and the young researchers knocking the door with license questions, he will talk about the politics of policy making, why you need good lawyers, the C in open science community, and the rewards without recognition.
11:30-12:30 in Kernhem 1
Bus Factor Basics: Protecting Your Project from Single Points of Failure
Alexander van Ittersum, Peter van der Togt
Does one developer on your team have all the knowledge of a crucial piece of software. Is that person retiring soon? Or are your group's servers maintained by only one IT guy or girl who also wants to go on vacation sometimes.
Prevent your bus factor from getting too low and make sure your IT projects are managed and maintained in a professional and effective way. Peter and Alexander talk about what their teams can do so that your IT wizard can also go to bed in peace for once.
15:20-16:20 in Kernhem 4
Challenges in visual insects monitoring
Bas Boom (One-Planet) en Janne Kool (WUR-Agrosystem research)
In the project ' Weet wat er leeft' (Know What Lives), our goal is to automatically detect the insect plagues and their potential predators inhabiting the greenhouse. We have developed several visual methods to monitor both flying insects and crawling mites. These include the use of yellow sticky traps in conjunction with smartphones, hyperspectral cameras, and an automated Berlese trap. A key challenge in the data collection is to transfer real life situations.
13:30-14:30 in Theatre Cerise
Chat GTPeter — Our friendly virtual software engineer
Valentijn Evers, Cristina Huidiu
How will transformers change search and discovery? How can we, as the Library, ensure that our community has access to reliable, ethical and sustainable tools?
Our discovery journey started with a contained experiment focused on building a software engineering support tool based on a small language model that can be deployed on a regular WUR issued machine while still performing at expected levels. Our areas of interest are threefold: the basis of an infrastructure where the model is easily swapable, curbing hallucitations and evaluation and optimisation of the output quality. As we continue to enhance answer structuring, our ultimate goal is to ensure that our connunity has access to tools that are reliable, ethical and sustainable. In this presentation we will focus on our lessons learned to date and how will our results influence future tools and services made available to the community. We've started open source and completely localized. How has your LLM journey been?
Discussion points:
- What would you need to see in order to trust an answer generated by a large language model?
- How are you using LLM's in your day-to-day work?
- What do you see as the main obstacles in large scale adoption LLM-based tools?
15:20-16:20 in Erica
Development of a WUR IOT platform for laboratory equipment
Jasper Koehorst
As the laboratories increasingly collect live data / operational data therefor the need for a unified platform to integrate, monitor, and manage datastreams from diverse laboratory equipment has become critical. Within UNLOCK we co-developed the WUR IoT platform, a modular system designed to connect laboratory devices, manage real-time data collection, and enable remote monitoring.
The platform leverages MQTT for lightweight communication and supports the LabEquipment Adapter Framework (LEAF) to facilitate integration across various protocols and devices. With features such as data transmission, asynchronous processing, and error handling, the platform ensures consistent and efficient device interaction. In this presentation I will explore the architecture, key components, and practical applications of the WUR IoT platform, demonstrating its potential to enhance laboratory workflows and data-driven research.
15:20-16:20 in Kernhem 3
Do you reward your modelers? Models as research output in recognition and rewards for researchers
Jene van der Heide, Theo Jetten, Cheng Liu, Shauna Ni Fhlaithearta
In the new WUR Academic Career Framework (ACF) models are recognised as research output, similar to publications. Unlike publications, models are dynamic, need continued investment beyond the lifetime of a research project and are dependent on the contribution of people in many different types of roles. In this session we will:
(i) Have an open discussion about how we value the work that is needed to maintain WUR's high impact models.
(ii) Introduce the Academic Career Framework in the context of models as a research output.
(iii) Present a series of questions for discussion on the boundary conditions needed for recognising models as a research output.
13:30-14:30 in Kernhem 4
FAIR in action: data and metadata refinement at WUR
Federico Padilla Gonzales, Felicia Wolters, Alessandra Soro
In this session, Felicia Wolters and Federico Padilla Gonzales will present the workflow and the impact of making data and metadata FAIR (Findable, Accessible, Interoperable, and Reusable). Their projects have been granted and supported by the FAIR Data Fund, a funding opportunity issued annually by 4TU.ResearchData. Federico will explain his work on FAIR data: Sharing high-quality spectral data of ~200 emerging chemical risks in the food chain. Felicia will give a talk about generating FAIR data for omics data integration - Mining Plant Specialized Biosynthetic Pathways. The session closes with a discussion panel, which will be hosted by Alessandra Soro (4TU)
09:30-10:30 in Kernhem 3
Getting ready for Digilab TO: What is it?
Ben Schaap (introductie), Bas van Vossen (Deltares), Timo Verwoest (MARIN), Gert-Jan Schotmeijer (Deltares), Roel van Bree (NFI), Paul van Schayck (KNMI).
This session gives an introduction to Digilab TO. Presenters from different Dutch applied research institutes (TNO, Deltares, WUR etc.) highlight challenges and chances related to data infrastructure. Next, key points are discussed where synergy betweens institutes can be expected. General outcomes expected: 1) information on the project Digilab TO, 2) stimulated engagement, building a strong network for applied research.
11:30-12:30 in Kernhem 4
Good Modelling Practice wiki and Model Gallery
Hajo Rijgersberg, Koen Meesters
A Good Modelling Practice wiki was developed. It contains methods recommended by literature as well as your colleagues. It can help to accelerate model development, improve model quality as well as prevent misinterpretation of model results. To gain access to the wiki, send a mail to: models AT wur.nl. The model gallery is a central WUR platform for models, a "shop window" of models, so to say. The web application shows a model on every page, information about the subject, purpose, used methods, precision, scale, authors, etc. The Model Gallery is openly accessible to every WUR employee, using one?s WUR account and password: modelgallery.wurnet.nl (log in with the VPN on outside the campus).
13:30-14:30 in Kernhem 2
How do I publish my research data?
Ignacio Sancho, Koen Bokhorst
In this session we will talk about the importance of publishing your research data. We'll show you where and how you can do this, and we'll give you tips on how to make your data more FAIR.
11:30-12:30 in Kernhem 3
How to efficiently deal with model quality documentation work during the model development process?
Vincent Hin, Thomas Hagenaars, Peter Hobbelen
Within the KB-MAST project, a tentative scheme has been developed to facilitate a most efficient execution of the documentation tasks for model quality, by connecting specific model quality checklist items to specific stages in the model development cycle. In this workshop we will introduce the model quality checklist, define the model development cycle, explain the scheme to connect these two, and ask the audience to contribute to a critical discussion.
13:30-14:30 in Kernhem 1
Hybrid Machine Learning and process-based modelling approaches for climate adaptation strategies
George van Voorn, Hajo Rijgersberg, Xinxin Wang & Cheng Liu and Charlotte Harbers & Xuezhen Guo
Introduction of the theme and developing hybrid ML crop models By George van Voorn Models are essential for forecasting the impacts of climate change on agri-food value chain actors and evaluating the effectiveness of climate adaptation strategies. This session presents hybrid methodologies combining Machine Learning (ML) algorithms with process-based modelling for integrating existing heterogeneous data. We discuss an example of a hybrid differential equation-ML crop model built in Torch (in R / Python) used for capturing genotype-by-environment interactions. FAIR-ML By Hajo Rijgersberg The re-use of data in Machine Learning is a laborious and ad hoc process. We explore several options to make the reuse and openness of data more operational for ML applications. AI Model for Predicting Pest Occurrence and Optimizing Pesticide Application Timing By Xinxin Wang & Cheng Liu Insect pests significantly reduce rice production in India, with pesticide use being a common but often suboptimal control measure. Climate change intensifies these pest challenges, highlighting the need for adaptive and precise management strategies. We develop an AI model to predict pest occurrence and determine the optimal pesticide application timing in rice cultivation in India. Hybrid AI models to predict tomatoes? shelf life using mixed-effect random forest (MERF) and physics informed neural network (PINNs) By Charlotte Harbers & Xuezhen Guo Understanding the shelf life of tomatoes is crucial for reducing product waste, which in turn has a significant impact on mitigating greenhouse gas (GHG) emissions, making it highly relevant to climate change efforts. We develop models using MERF (Mixed Effects Random Forest) and PINNs (Physics-Informed Neural Networks) techniques to predict the shelf life of tomatoes. These models can be applied to a broader range of fresh produce, which can offer a substantial contribution to climate change mitigation by optimizing postharvest chains and reducing waste.
11:30-12:30 in Theatre Cerise
Infrastructure and governance of (farm level) model collaboration
Tamara ten Den, Hugo Scherer, John Helming, Pepijn van Oort
This session consists of two parts. 1) In the KB MAST project we coupled two models, FarmDyn and Farmmaps KPI calculations. In this first part of the session we will discuss our approach to coupling these models, the infrastructure, the benefits of coupling these models, and the lessons we learned along the way. 2) In the second part of the session we will look back to model collaboration work between the farm models FarmDyn, DairyWise, Nutrient balance Arable farming tool (NA) and RothC. Model collaboration enables answering more complex questions and enlarges the scope of the analysis. We will discuss the challenges and the need for a governance structure and networking to foster model collaboration. Together we will develop a general list of recommendations for our management to improve the governance of model collaboration.
15:20-16:20 in Kernhem 2
Linking models with the circular bioeconomy indicator framework and SDGs
Iris Vural Gursel, Berien Elbersen, Sjaak Conijn
As part of the KB MAST project, a circular bioeconomy indicator framework has been developed. The aim of this session is to inform about this framework encompassing indicators covering different circularity and bioeconomy principles, and their connections with SDG targets. We will present a preliminary overview of the indicators that are addressed by some selected WUR models. Furthermore, as part of an open discussion we invite WUR modellers to provide information on the type of indicators and data generated from their models, explore synergies/trade-offs and identify possible gaps in indicator coverage in WUR in the context of circular bioeconomy.
15:20-16:20 in Kernhem 1
Operating a low-cost sensor network to monitor gaseous nitrogen compounds (NO2, NH3)
Jasper Fabius, Francisco Souza, Daniel Bertocci, Shaojie Zhuang (all from OnePlanet Research Center)
In this session, we explore NitroSense ? a low-cost sensor device designed for measuring airborne nitrogen for environmental applications. We begin by discussing NitroSense's development, challenges, and real-world testbeds like Liefstinghsbroek, Agro-innovatiecentrum de Marke, and an experiment with drone-based measurements. Then, we address the challenges posed by the large volume of data generated by low-cost sensors. Pushing the performance of low-cost sensors using explainable AI for calibration. Finally, we focus on real-world applications where sensor measurements and atmospheric dispersion models are integrated into an emission monitoring framework, specifically tracking high-emission events from animal barns and manuring activities in agricultural areas.
15:20-16:20 in Calluna
RDM services
Peter van der Togt, Joris Luijsterburg
The RDM Infrastructure team works on automating Research Data Management (RDM) solutions with iRODS. iRODS is an open-source RDM software. You can think on ingestion of data or archiving data to tape. In this interactive session we will first give you an update of the services offered. We show you some use cases and examples. In the second part of the session, we want to give you the opportunity to express your problems and challenges in processing large amounts of data. These use cases help us to prioritize our work and improve our services.
13:30-14:30 in Kernhem 3
Recognizing herb-rich grassland with remote sensing and pictures
Wouter Meijninger, Tim Visser, Ron Wehrens, Janne Kool
Herb-rich grasslands are essential for the conservation of biodiversity in agricultural landscapes. Therefore, multiple initiatives in the Netherlands aim to increase the amount of herb-rich grasslands by financially rewarding farmers who work towards developing herb-rich grasslands. These initiatives require monitoring data to determine if the grasslands are actually developing into herb-rich grasslands. Within this project, we explored whether we could predict the herb-richness of grasslands based on remote sensing markers and pictures taken in the field. We collected data on herb-richness on a continuous scale for thousands of grassland fields and selected remote sensing variables that could potentially play a role in predicting herb-richness. We used different statistical approaches -- including artificial intelligence -- to predict herb-richness as accurately as possible.
15:20-16:20 in Theatre Cerise
SciML: Gray-Box Physics-informed Neural Networks (PINNS) modelling in food-related domains
Xuezhen Guo, Ruud van der Sman
In this session, Xuezhen will present an overview of the basic principles of Scientific Machine Learning (SciML). He will use the groundbreaking 'AlphaFold2' model as an example to demonstrate how SciML works, emphasizing the key factors behind AlphaFold2's success.
Next, Ruud will give a detailed presentation on how Physics-Informed Neural Networks (PINNs) have been applied in his research. He will explain their use in areas such as combining neural networks with rheology models for 3D food printing, and how neural networks can be used alongside kinetic models to predict quality decay.
09:30-10:30 in Calluna
Supersize your workflows: laptop vs Anunna
Bert Klandermans, Leonardo Honfi Camilo, Alexander van Ittersum
In this session we will talk about some developments around making Web services available through our own supercomputer Anunna. Enabling applications that we are used to running on our own workstation (like SAS, R-Studio or Matlab). After that Bert (Dairy Campus @ ASG) tells about how he runs his containerized models effectively using Signularity on Anunna
11:30-12:30 in Kernhem 2
SURF Research cloud introduction training (2hrs; separate registration needed)
Carsten Schelp, Yuliia Orlova
A two hour introductory session to the SURF research cloud, a collaboration environment for creating shared workspaces across Dutch research institutes. Participants learn how to take their first steps into this SURF environment for sharing data, workflows and computation.
09:30-12:30 in Erica; please bring your laptop
Swiss knife chemometrics: hands on modelling any type of multivariate collinear data with a single algorithm
Puneet Mishra
The session will present the recently developed Swiss Knife chemometric regression algorithms by Puneet Mishra. The Swiss Knife algorithms allows you to do regression, classification, sensor data fusion, feature selection, higher order data analysis and loads of other things. Obviously, that is why it is called "Swiss Knife". For more information, see https://doi.org/10.1002/cem.3441.
11:30-12:30 in Calluna
Tangible Landscape, a GIS-based tool to design Future-proof landscape
Xiaolu Hu, Eline van Elburg, Jan Droesen, Marc Meijer
Tangible Landscape (TL) is a projection-augmented sandbox powered by a Geographical Information System (GIS) for real-time geospatial analysis and simulation[1]. TL couples a mock-up with a digital model in near-real time by using 3D sensing, enabling users to apply geospatial modelling, simulation and visualization. TL enables users to physically interact with the digital environment by removing the need for interaction with a mouse and keyboard by shifting to manipulation by touch and feel. It has been developed by NCSU since 2015, and now it is further developed by WUR trying to integrate TL with the hydrological model SFINCS.
13:30-14:30 in Calluna
The Autonomous Greenhouse: From Playground to Real Life
Stef Maree, Atam Gangaram Panday
In the Autonomous Greenhouse Challenge, five international teams are challenged to grow tomatoes fully autonomous with cutting-edge AI solutions. In this talk, we will discuss what challenges we encountered while transitioning from this playground? to all our research greenhouses. Join us to uncover how our innovative IT building blocks helped us overcome these hurdles, and how they can help you.
09:30-10:30 in Kernhem 2
The future of model collaboration: strategic choices, good practices, and example connection to AI
Saeed Moghayer, Pim Post, Jason Levin-Koopman, Xinxin Wang
Talk 1:Identifying Research Questions Demanding Model Collaboration: A WUR Perspective
Presenter: Saeed Moghayer
Abstract:?This project aims to identify complex research questions, such as those related to the nitrogen, climate, and biodiversity crises, which necessitate increased collaboration between research models. Through interviews with 19 respondents, primarily within Wageningen University & Research (WUR), this study explores the untapped potential for synergy between the WUR knowledge base and research questions across various domains. By considering respondents' job descriptions, disciplines, and affiliations, this research provides valuable insights into identifying and prioritizing critical areas for collaborative research and modelling.
Talk 2:Good practice in model collaboration, model clusters, and an example of subnational assessment of multiple SDG indicators in Pakistan
Presenters: Pim Post and Jason Levin-Koopman
Abstract: Model collaboration is essential to address complex research questions. It happens rather often but ad hoc and with little guidance. Good practices for model collaboration have been identified for several steps of the modelling cycle. One insight is that much focus often goes to the technical aspects of model collaboration, while the collaboration aspect, achieving shared system understanding, may be as important. Although collaborative learning is an important aspect, enhancing the value of model collaboration would require the investment in selected model clusters to deepen understanding in complex societal issues. An example of such cluster is the MAGNET-MagnetGrid-LPJmL-SSID cluster, which allows to assess effects of policy and climatic changes on a national as well as subnational level, across a range of themes. This is illustrated for a case study in Pakistan.
Talk 3:Foresight Scenario Modelling for Global Wheat Production Under Shared Socioeconomic Pathways: An Integrated CGE-AI Model Approach
Presenter: Xinxin Wang or Saeed Moghayer
Abstract: This study presents a novel approach to foresight global wheat production by integrating the MAGNET Computational General Equilibrium (CGE) model with advanced AI techniques. The AI model is expected to enhance MAGNET's capabilities by improving its predictive accuracy and enabling the handling of complex, non-linear relationships within large multi-year datasets. Through this integrated model, various Shared Socioeconomic Pathways (SSPs) are simulated, providing insights into how future socioeconomic scenarios may impact wheat production. The model's ability to incorporate diverse shocks, such as policy changes, climate change impacts, and technological advancements, further enhances its capacity to generate robust forecasts under different global conditions. This research offers a valuable tool for understanding the potential implications of future socioeconomic pathways on global wheat production, contributing to informed decision-making for food security and agricultural policy.
09:30-10:30 in Kernhem 1
Value Creation — from Research to Innovation
Yannick van Gelder, Ruud Borgart
A session led by Yannick van Gelder (WDCC), where he shares the stories of three research projects that transcended academia, offering insights into how research can create real-world impact. Whether you're a researcher or simply curious about the power of knowledge to drive change, this workshop will inspire and inform, with practical guidance on accessing the support you need.
13:30-14:30 in Erica
What do we need a Digital Strategy for?
Jene van der Heide, Roos Godefrooij
Did you know we (WUR) have a Digital Strategy? What does this entail and what does model and data management have to do with it? Since having a strategy is not enough, we need you to help implementing it. Therefore we also present to you a roadmap, but still, that doesn' t make our dreams come true. This session is about increasing your involvement in creating the pathway to a more digital minded WUR.
09:30-10:30 in Kernhem 4
Wageningen Model & Data Day 2024
Registration website for Wageningen Model & Data Day 2024WUR Events Teamdata@wur.nl
WUR Events Teamdata@wur.nlhttps://event.wur.nl/wageningen-model-and-data-day/subscribe
2024-10-17
2024-10-17
OfflineEventAttendanceMode
EventScheduled
Wageningen Model & Data Day 2024Wageningen Model & Data Day 20240.00EUROnlineOnly2019-01-01T00:00:00Z
Akoesticum Akoesticum Nieuwe Kazernelaan 4 - D 6711 JC E Netherlands
