We proudly present the first MIDL satellite event, specifically designed for young researchers: the MIDL Doctoral Symposium 2021. Develop and train your scientific skill set, get to know fellow PhD students from all over the world and enjoy participating in the MIDL conference while already knowing some colleagues.
The Doctoral Symposium will take place on 2 July 2021 as a virtual event. We will use Zoom and Gather.Town for optimal interaction during the day. Thanks to our sponsor, the participation is free of charge.
The number of participants is restricted to 50. The selection process will focus on the candidate’s motivation, the integration of candidates in the early stage of their PhD thesis without broad conference experience or network opportunities as well as diversity.
You can apply with the application form and your CV. The application deadline is 4 June 2021, 23:59 CET. The Doctoral Symposium is open to all PhD students, even if you do not participate in the MIDL conference. The decision notification will be sent via email by 14 June 2021.
(All times CET)
|09:00 - 09:15
|Talk: Model evaluation in applied machine learning
|09:15 - 10:15
|10:15 - 10:30
|10:30 - 10:45
|Talk: Visualization of Results
|10:45 - 11:15
|Activity: Poster Karaoke
|11:15 - 12:30
|12:30 - 13:30
|Workshop: Time- and Selfmanagement
|13:30 - 17:30
|17:30 - 19:00
|Activity: Social Event
Talk: Model Evaluation in Applied Machine Learning
This talk will give an overview of common pitfalls in model evaluation for (supervised) machine learning problems. It is well understood that methodological mistakes may lead to biased evaluation results and over-optimistic conclusions. However, many poor practices can still be observed in the research literature. The focus in this talk will be on issues related to the experimental design (e.g. data splitting) and the statistical analysis (e.g. adjustments for multiple comparisons). In this regard, some practical guidance is provided to improve model evaluation for your research project(s).
Max Westphal obtained his PhD in applied statistics in 2019 from the University of Bremen where he worked on “Model Selection and Evaluation in Supervised Machine Learning”. He then joined Fraunhofer MEVIS as a postdoctoral researcher for Data Science and Biostatistics. His research is focused on the development and evaluation of prediction models for medical applications. Such models can be utilized in clinical decision support systems, for instance by providing patient-individual risk predictions based on complex data sources.
Talk: Visualization of Results by Plotting
Most results in medical image analysis are not just single numbers, and thus need to be visualized in order to make sense of them. Frequently, plotting is the method of choice for visualizing numerical results. The most suitable format depends, of course, on the type of data to be visualized, but equally on the purpose of the presentation: do you want to check intermediate or final results for plausibility, get an overview over non-image data; or do you want to present them in a poster, presentation, or publication? This talk will give an overview of different basic types of plots, how to choose between them, and how to avoid some plotting pitfalls.
Ole Schwen obtained his PhD in Mathematics in 2010 from University of Bonn, where he worked on Composite Finite Element simulations for simulations in Biomechanics. Since joining Fraunhofer MEVIS, his research has included multiscale modeling and spatially resolved simulation of pharmacokinetics as well as data science in Computational Pathology.
Activity: Poster Karaoke
Train your presentations skills and excell in presenting posters you have known for 2 Minutes. Are you in for a challenge?
Workshop: Time- and Selfmanagement, Motivation and Leadership
Work in a modern scientific environment requires advanced transferable skills like self-, time- and project management, effective communication, inspirational leadership & team management and collaboration and negotiation. These skills can be improved upon in intense interactive workshops, ideally from scientists to scientist to allow for direct applicability into everyday work.
Prioritizing / project management
- Placing tasks onto an urgent and important matrix can be telling (Eisenhower matrix). Urgent is not important, and tasks can be classified into the four fields “just do it”, “make it happen”, “delegate or use the Pareto principle” and finally the “trash can”. Key is to make the “important-only” tasks happen by taking resources (time) from the “urgent-only” tasks. Using real-life examples we will discuss approaches to prioritize and taking decisions to become more effective in our everyday work.
- Helpful for time- and self-management is the pareto principle and focusing on the important vs the urgent (“rocks-and-sand” analogy).
- Do you plan your whole week? Do you fill your time 100%? Then how do you accommodate the unexpected? More importantly, how do you accommodate the opportunities? Employing the 80-20-20 rule is key.
- How do you let go? Using the famous “20 dollar auction”, we will address the “sunk-cost-fallacy” and train letting go of the good in favour of the better.
Motivation and Delegating
- How would you like to work? Do you enjoy freedom or would you rather be closely supervised? What style will you use when you become a leader? Find out! 😉
- When should you delegate? If you delegate, will the product be the same as when you would do it yourself? Delegating is not easy, it takes guts and letting go. But it can be rewarding – beyond just saving time for yourself: train others, delegate to someone who can do it better. Delegating is key to success.
- Where does motivation come from? Motivation is the essence of performance, so understanding the mechanics of motivation and promoting motivation in yourself and in others is the energy to overcome obstacles and to reach goals.
PD Dr.rer.nat. Daniel Mertens heads two research groups, one group at the German Cancer Research Centre (DKFZ) and a research group at the Ulm University Hospital. As a scientist, Daniel Mertens is the author of 92 publications that have been cited more than 4000 times by colleagues in publications. He has received more than €5 million in grants from external third-party funders to finance his research and coordinates international research networks (cancerepisys.org and leukemia-resistance.de). Since 2011, he has been training scientists, physicians, administrators and staff in the framework of "Scientists Need More!" in transferable skills (www.scientistsneedmore.de). So far, >7000 participants took part in international workshops in Europe, USA and Africa.
Activity: Social Event
Not convinced that virtual social events can be cool? Let’s prove you wrong! Enjoy a relaxed evening in our Gather.Town pub with your fellow PhD students from all over the world. This is also a great opportunity to meet research engineers and recruiters from our sponsor Euroimmun in a casual way.
- Alessa Hering, Fraunhofer MEVIS & Radboudumc Nijmegen
- Jannis Hagenah, CHI Lab, University of Oxford, UK
- Floris Ernst, University of Lübeck