Julia Wąsala
AI Researcher | Science communicator
"Trustworthy AI sytems can only be achieved through dialogue and close collaboration between Machine Learners and domain experts.
About Me
I'm an AI researcher at the Leiden Institute for Advanced Computer Science (LIACS)and the Space Research Organisation Netherlands (SRON). I design and evaluate machine learning methods for Earth Observation, for instance to find large greenhouse gas emissions from space.
Public engagement with research is very important to me. I write about AI and Earth Observation on my Substack newsletter [NL] and on my blog [EN].
Nothing beats speaking to people directly, so I regularly speak and participate in public AI events.
Research
I research how we can automatically create machine learning methods for Earth Observation applications, such as detecting large greenhouse gas emissions from space. I want to better understand how we can predict the performance and trustworthiness of machine learning models in real life. We all know how AI sysems can break down in the real world, but the well-known problem of bias looks very different in satellite data.
My recent publications (full list on LinkedIn):
- AutoMergeNet: AutoML-Based M-Source Satellite Data Fusion Evaluated With Atmospheric Case Studies.
- Mitigating Representation Bias Caused by Missing Pixels in Methane Plume Detection.
More about my research on:
Click to expand
“AI is not only interesting because of the societal impact. How AI systems work is a story of human problem-solving and scientific discovery.
Public Engagement
I aim to make my research findings as accessible as possible. AI research is so much more than ChatGPT. Did you know that not all AI use that much energy? And that satellite data can also be biased? How you write a good algorithm is not just societally relevant, it's also a fun puzzle.
My main goal in public engagement is connecting with people. I do that through writing, talks, workshop, leading discussions and engaging with kids through events like hackathons. I'm inerested in everybody's take on AI, from all ages and backgrounds.
- Using AI to track greenhouse gas emissions - Leiden University
- Safe and powerful AutoAI solutions amplify future research and innovation in many domains - EU project.
- AI is a solution and a problem for climate change - article in Vrij Nederland
“My research, communication and teaching are driven by human connection. I don’t want to give a lecture, I want you to feel my enthusiasm!
Teaching
Machine Learning for Earth Observation is a relatively new field. As a result, I had to learn a lot of lessons the hard way. In my teaching, I help students navigate the world of satellite data and the intricacies of evaluating machine learning pipelines for real-world applications.
- I supervise Machine Learning for Earth Observation Master's thesis. Find available thesis topics here.
- Machine Learning Bachelor's course
- Urban Computing Master's course.
I share many of my teaching materials on my Blog