|
Abstract:
In this workshop, we will further explore the role of AI in education using the common language of the National Education Lab AI (NOLAI). The common language is introduced and explained in the light of NOLAI’ co-creation program. In this program, we develop novel AI applications together with schools, scientists and companies. In the workshop, I will show how our co-creation projects are designed and developed with the help of these models. Participants will learn how to describe AI solutions in our common language and discuss possible futures of AI in education and the consequences thereof. Our common language contains three main models:
- The Detect-Diagnose-Act model helps understand how AI is functioning. It describes the input data (detect), the constructs being interpreted (diagnose) and how the AI is acting (act).
- The adaptivity model outlines 3 approaches to personalization, namely step, task and curriculum adaptivity.
- The 6 layers of the automation model describes control division between the teacher and the AI. This indicates how AI automation and teacher autonomy are related in particular applications.
Bio:
Inge Molenaar is the Director of the National Education Lab AI (NOLAI) and a Professor of Education and Artificial Intelligence at the Behavioural Science Institute, Radboud University, Netherlands. With over 20 years of experience in technology-enhanced learning, she has successfully navigated roles ranging from entrepreneur to academic.
Dr. Molenaar's research focuses on innovative, technology-empowered approaches to optimize human learning and teaching. Central to her work is the application of data, learning analytics, and artificial intelligence to understand the dynamics of learning over time.
Envisioning Hybrid Human-AI Learning Technologies, Dr. Molenaar aims to augment human intelligence with artificial intelligence, empowering both learners and educators to make education more efficient, effective, and responsive to individual needs. Her research group, the Adaptive Learning Lab, investigates how self-regulated learning can be supported through technology. At the National Education Lab AI, she develops new educational practices using AI and explores the responsible use of AI in education.
She is a recipient of several prestigious grants, including an ERC Starting Grant and funding for the National AI and Education Lab (NOLAI) and holds Master's degrees in Cognitive Psychology and International Business Studies, as well as a PhD in Educational Sciences from the University of Amsterdam.
|
|
Abstract:
Experiment 1. Autoregressive modelling with AI
Read More
Linear Regression Models:
Linear methods like AR, ARX, and ARIMA are popular classical techniques for time series forecasting. But these traditional approaches also have some constraints:
- Focus on linear relationships and inability to find complex nonlinear ones.
- Fixed lag observations and incapacity to make feature pre-processing.
- Missing data & noise are not supported.
- Working with univariate time series only, but common real-world problems have multiple input variables.
- One-step predictions while many real-world problems require predictions with a long time horizon.
Machine Learning for Time Series Forecasting:
- Xgboost Regression
- Linear Regression
- Decision Trees (DT)
- Regression Random Forest (RF) Regression
The learning process is based on the following steps:
- Algorithms are fed data. (In this step you can provide additional information to the model, for example, by performing feature extraction).
- Train a model using this data.
- Test and deploy the model.
- Utilize the deployed model to automate predictive tasks.
Experiment 2. Spiral drawing
Read More
Aims:
- investigate differences between internally vs externally guided movements
- investigate movement adaptation
- investigate effects of speed on motor performance/accuracy
- investigate effects of levodopa/DBS
Background:
The occurrence of subthalamic beta bursts parallel reductions in movement velocity during free drawing in patients with PD. Deep brain stimulation modulates this relationship. (Bange et al., 2024) However, the pathological mechanisms of how subthalamic beta bursts affect movement remain elusive. Here, we aim to investigate the communication between STN and M1, and how this relates to drawing performance during free drawing (as an internally guided movement, IGM) and dynamic template guided drawing (as an externally guided movement, EGM). We further investigate how dynamic template guided drawing affects free drawing, and if adaptations in movement performance are modulated by levodopa substitution and DBS. We hypothesize that beta bursts affect drawing performance during free drawing. This association should be abolished during dynamical template guided drawing with slow speed but reoccur when shifting the speed- accuracy relationship towards faster movements. Furthermore, cuing the patients’ drawings by providing dynamically evolving templates should lead to an improvement of free drawing performance, specifically when patients receive levodopa substitution.
Experimental procedures:
Subjects first draw 15 free spirals (IGM). Then they trace 15 dynamic template guided spirals (EGM) at varying speed (very slow, slow, medium, fast, very fast), followed by drawing another 15 free spirals. See Fig. 2 for an overview.
Equipment: EEG, EMG, Graphics tablet
Drawing task (Fig. 2)
- 15 trials of free drawing, 15 trials of dynamical template guided drawing, 15 trials of free drawing
- Free drawing should be performed as fast as possible, without crossing their own trace. Six revelations within a square of 12cm*12cm.
- Conditions:
- Medication: on/off (performed on different days)
- DBS: on/off
- Drawing: free/ dynamical template guided
- Speed: very slow, slow, medium, fast, very fast
- Duration: ~30 min (stim on and off)
Figure 2. Experimental procedures.
Variables/outcomes:
Spirals:
- Drawing velocity
- Drawing accuracy
- Multiscale Sample Entropy
- Correlation with bradykinesia scores
Brain activity:
- STN/M1 spectral power
- STN/M1 cross frequency coupling
- STN/M1 beta bursts (duration, amplitude, rate)
- STN-M1 coherence
- STN-M1 cross frequency coupling
Experiment 3. Gait
Read More
Aims
A1: Identify individual subjects and their current physiological/psychological/health state
Background
The ability to walk is crucial in human life. Gait analysis is an established method for exploring neurophysiological mechanisms and pathological associations in science and medicine. High acquisition costs and limited transfer in highly standardized laboratory settings often hinder effective and timely knowledge dissemination into daily life. Furthermore, individual and state- dependent aspects are frequently neglected.
Applying an integrative, multidimensional, and holistic characterization of individual gait patterns, we aim to identify altered gait characteristics in PD. The multimodal use of various sensor technologies to analyze time-continuous parameters will reveal previously unknown coordination-patterns within gait and their interactions with intra- and inter-individual aspects.
Long-term, this will enable a cost-effective, non-invasive examination of gait in natural environments, significantly scaling up neuroscientific and clinical research and improving accessibility to the public, benefiting not only medicine but various other life domains.
Experimental procedures
Subjects will walk under various conditions (normal, med-On, med-Off, DBS-On, DBS-Off, different emotional states, physical exhaustion, …, see table 2). 5 cameras will record gait for the 3D Pose estimation. Accelerometry and EEG will also be applied. Gait speed is controlled.
Table 2. Different Walking Conditions
| |
Continuous |
Or trial based (20 trials) (needs to be determined) |
| SelfSelect* |
2 min |
3 min |
| SelfSelect +10% |
2 min |
3 min |
| SelfSelect -10% |
2 min |
3 min |
| 2 km/h |
2 min |
3 min |
| 3 km/h |
2 min |
3 min |
| 6 Emotions** |
6 × 2 trials × 90s = 18 min |
20 trials × 6 × 20s = 40 min |
| Physical |
3 × 240s = 12 min |
20 × 15s = 5 min |
| Total time |
~40 min |
~60 min |
|
* 6 meters at 3 km/h ≈ 7.2 seconds.
** Rating of emotion at the end of each trial?
|
Bio:
Muthuraman Muthuraman was born in Chennai, India, in 1980. He received the B.E. degree in electronics and communication engineering from the University of Madras, Madras, India, in 2002, and the MS in digital communications in Christian Albrecht’s University, Kiel, Germany in 2006. Ph.D. degree in biomedical engineering from the technical faculty and Department of Neurology of Christian Albrecht’s University, Kiel, Germany, in 2010. In 2010, he joined the Department of Neurology, University of Kiel, as a Post-doc, and in 2013 became a senior post-doc. Since December 2016, he has been with the Department of Neurology, Johannes Gutenberg University Mainz, where he is an Assistant Professor, and the head of the department biomedical statistics and multimodal signal processing unit. Currently from 2024 he is heading the group Informatics for Medical Technology (IMT) in Augsburg as an associate professor and second affiliation to Julius Maximilian university of Würzburg in the department of Neurology and head of the group Neural Engineering with Signal Analytics and Artificial Intelligence (NESA-AI). His current research interests include mathematical methods for time series analysis and source analysis on oscillatory signals, sleep, function of oscillatory activity in central motor systems, biomedical statistics, connectivity analyses, multimodal signal processing and analyses of EEG, MEG, fMRI and EMG, structural and network analyses on anatomical MRI and DTI, functional network analyses on PET imaging, machine learning and deep learning, network analyses on proteomic and genomic data, RNA, mRNA and Spatial transcriptomics.
|
|
Abstract:
We are now in the Age of Artificial Intelligence, which is transforming our lives at an unprecedented pace. AI tools and applications enhance our daily activities, from getting information to managing finances. However, AI also amplifies existing social challenges, necessitating a societal discussion on its future. Effective governance frameworks, based on ethical and moral values, are crucial. UNESCO's Recommendation on the Ethics of AI, adopted by 194 countries, aims to ensure AI delivers fair, sustainable, and inclusive outcomes. This involves protecting human rights and dignity, and moving beyond self-regulation.
UNESCO's policy approach acknowledges the varying stages of AI development across countries, providing targeted support. The Readiness Assessment Methodology (RAM) and Ethical Impact Assessment tool, launched in December 2022, help countries assess their AI readiness across five dimensions: Legal and Regulatory, Social and Cultural, Economic, Scientific and Educational, and Technological and Infrastructural. These tools offer both individual and comparative insights, guiding countries in building capacities and strengthening policies.
The goal is to develop AI ethically, benefiting humanity and the planet. By using tools like RAM, UNESCO supports Member States in establishing robust AI regulations, moving closer to this ethical vision.
The purpose of the Ethics of AI workshop is to introduce the audience with the guiding principles and values embedded in the 2021 recommendation on the ethics of AI and to introduce the Readiness Assessment Methodology and Tools with the hope to count the Sultanate of Oman and other Member States amongst the active implementers of the recommendation.
Bio:
Amina Hamshari is the Regional Advisor for Social and Human Sciences at the UNESCO Regional Office in Doha for GCC States and Yemen. She holds a postgraduate diploma in Social Sciences (Contemporary History) from the University of Paris X-Nanterre, France. Amina's career spans significant roles in Palestine, where she coordinated curriculum development and emergency projects for UNDP-PAPP and the UN Special Coordinator for the Middle East Peace Process. At UNESCO Headquarters, she specialized in intercultural dialogue programs, has designed the “Writing Peace” programme and “Arab Latinos!”.
Amina has designed and conducted many projects to promote social justice and reconciliation through the reappropriation of space and narratives such as the Regioanl Conference "HERstory: Heroines in the Liberation Struggles in Southern Africa" as co-architect and organizer with female political leaders and young people from 8 countries. As the initiator of the "Art-Lab for Human Rights and Dialogue" at UNESCO, Amina has leveraged the arts to support vulnerable communities and promote human dignity. Her innovative approach has led to the successful implementation of pilots worldwide, addressing exclusion and fostering social cohesion. Additionally, Amina is now working on the development of emergency preparedness and mitigation plans for sports ecosystems within UNESCO’s Fit for Life Flagship, ensuring resilience and sustainability in the face of crises.
She has also supported the implementation of the 2021 UNESCO Recommendation on the ethics of artificial intelligence in 6 countries in Southern Africa and currently in the Gulf States and Yemen.
|