Human-Centered Artificial Intelligence Lab (Bielefeld University)

📑 🚀 Our new paper in Behavior Research Methods What if your chat history could help research (without sharing a single word)? Online communication generates a wealth of data that could offer valuable insights into human behavior. But this data is mostly out of reach for research and for good reason: privacy and ethics matter. That's where Dona comes in - our data donation platform developed together with Paul-Julius Hillmann, Florian Martin, Erwin Böttinger and Hanna Drimalla. 📌 With Dona, people can donate their de-identified chat Data from WhatsApp, Facebook and Instagram! De-identified means only anonymized IDs, message lengths and timestamps. No content, no private details. 📌 But is such minimal data still meaningful for scientific research? Our paper shows how even meta-data can be helpful in identifying interaction patterns and temporal dynamics. And it's not just for our advanced analysis! 📌 The participants received personalized visualizations about their messaging behavior and found them insightful. Most spotted major life events just by looking at simple graphs 📈 📊 📌 Dona is suited for sensitive research, like understanding communication patterns in mental health research while fully protecting participant privacy. Participants benefit too by receiving personalized visualizations of their messaging habits. 🌍 Everything is open-access—the tool, the dataset (for researchers), the analysis code and the paper. Check it out: https://lnkd.in/e942Wk9V

🚀 Join our Human-Centered AI Lab at Universität Bielefeld as a PhD candidate! If you have a background in computer science, data science, or a related field and are passionate about developing AI to analyze human interaction, this is for you. Engage in meaningful research at the intersection of psychology and computer science. 👉 More details & application: https://lnkd.in/eRNCDTAq Know someone who might be interested? Feel free to share or reach out! 🤝

🌍 From Bielefeld to Bangalore: Highlights from Bhargav’s research stay! As part of his research stay at the Bangalore Institute of Technology (BIT) in India, Bhargav Acharya recently conducted a data collection study 📊to improve non-invasive methods for his ongoing research on heart rate estimation ❤️. During his time at BIT, Bhargav also gave a talk to students, sharing insights into remote photoplethysmography (rPPG) and and the broader implications of this technology. Thanks to the Bangalore Institute of Technology Institute of Technology, especially Dr.Aswath M U and Dr. Kalpana A B., for their support in facilitating this study. A special thanks to Naveen Kumar for his valuable assistance throughout the project.

✨ Join us for an insightful talk by Prof. Dr. Claudia Müller-Birn (@Freie Universität Berlin) on a human-centered approach to AI system design in healthcare! What to Expect? Prof. Dr. Müller-Birn will share insights into designing AI systems that prioritize human values and needs. Key topics include: 🤝 Human-AI collaboration and its impact on decision-making. 📊 Value-centered dataset creation, focusing on ethical considerations. 🤔 Participatory AI development, where patients, clinicians, and researchers co-design systems. Hosted by: Human-Centered AI Lab of Prof. Dr. Hanna Drimalla together with the SAIL Research Network.

In 2024, we sharpened our focus on human-centered AI that works for the people 👨💻👩💻. We presented our research at conferences and events in Glasgow, Manchester, New York, Yerevan, and Stockholm, connecting with researchers worldwide 🌍. A highlight was Hanna Drimalla's appointment as Professor of Human-Centered Artificial Intelligence (HCAI) 🎓, which is now also reflected in our new group name: HCAI Lab. During our team retreat, we reflected on AI that’s transparent, fair, and truly useful for people from all walks of life 🧐 Get ready to dive into fresh scientific adventures with us in 2025 🚀!

Project A06 researchers David Johnson, Jonas Paletschek, and Hanna Drimalla have collaborated with Olya Hakobyan to publish a new paper: "Explainable AI for Audio and Visual Affective Computing: A Scoping Review." 📄 The paper reviews how Explainable AI (#XAI) is being applied to audiovisual affective computing, which uses data such as facial expressions and vocal cues to recognize emotional states. The authors provide an overview of #XAI concepts relevant to affective computing, helping to shed light on how machine learning models in this domain make decisions. While interest in interpretability of machine learning models is growing, there is still a gap in the research addressing this topic. The review highlights encouraging developments in using XAI for audiovisual affective computing, though the application of methods remains limited. The paper concludes with recommendations for enhancing interpretability in future affective machine learning research. ➡️ You can find a short interview with Dr David Johnson and further information here: https://lnkd.in/eqtQRjK8

Our Kids-SIT at MUM Conference! 🎉 Last week at the International Conference on Mobile and Ubiquitous Multimedia (MUM) in Stockholm, our team member William S. presented our demo paper 📝: "Introducing the „Simulated Interaction Task for Children“ (Kids-SIT). The Kids-SIT is a video-based tool to analyze social behavior in children by studying eye gaze 👀, facial expressions 😊 and head movements 🧑💻. ✨ This tool helps standardize how we understand these interactions – valuable for both research and practice! 💡 It was great to share our work, connect with others 🤝, and get feedback! 🙌 Curious to learn more? Check out the paper here 👉 https://lnkd.in/ez4XqQYy 

🎉 📑 🚀 Exciting news—our review paper has just been accepted by the IEEE Transactions on Affective Computing! But first, let me take you through a playful experiment ⤵️ After a bit of convincing, ChatGPT analyzed the facial expressions in my picture, using descriptions of my eyes, eyebrows, etc. When I asked it to highlight these features in the original image, it generated the result below. I have to admit, I find myself disagreeing with ChatGPT’s assessment! This playful experiment highlights an important challenge in the field of explainable AI (xAI): understanding how AI models make decisions. While this was just a fun exercise, David Johnson, me, Jonas Paletschek and Hanna Drimalla conducted a more rigorous analysis of how xAI is applied in affective computing, e.g. AI applications in emotion recognition. Our review paper highlighs the current state, the promise and limitations specific to this field: https://lnkd.in/eJsnt9ua #xai #affectivecomputing

🎉Once again, we had the pleasure of hosting a guest speaker at our Research Colloquium! Yesterday, Dr. Philipp Müller (Deutsches Forschungszentrum für Künstliche Intelligenz, DFKI) joined us for an engaging talk. 📌In his presentation, “Multi-modal Behaviour Analysis: Bridging the Gap between Computer Science and Psychology,” Dr. Müller explored how bringing together psychology and computer science theories and methods can be very insightful🌍 . He shared interesting insights on human and machine face detection, engagement detection and assessing emotion regulation strategies —highlighting the opportunities of uniting 🧠 psychology and 🤖 computer science.