Paul Prasse
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Paul Prasse has a total of 55 co-authors. Here are the top co-authors by number of shared publications:
- Lena Ann Jäger
Department of Computational Linguistics, University of Zurich, Zurich, Switzerland
11 shared publications - Tobias Scheffer
University of Potsdam, Potsdam, Germany
11 shared publications - David Robert Reich
Department of Computational Linguistics, University of Zurich, Zurich, Switzerland
10 shared publications - Matthias Lienhard
Max Planck Institute for Molecular Genetics
4 shared publications - Ralf Herwig
Max Planck Institute for Molecular Genetics, Berlin, Germany
4 shared publications - Daniel G. Krakowczyk
University of Potsdam, Potsdam, Germany
3 shared publications
Latest Publications
- Open Access
- 25 May 2025
- Published by Association for Computing Machinery (ACM)
- p. 1-10
Neural Additive Models Uncover Predictive Gaze Features in Reading
- 25 May 2025
- Published by Association for Computing Machinery (ACM)
- p. 1-10
Open AccessNeural Additive Models Uncover Predictive Gaze Features in Reading
- Open Access
- 25 May 2025
- Published by Association for Computing Machinery (ACM)
- p. 1-10
Detection of Alcohol Inebriation from Eye Movements using Remote and Wearable Eye Trackers
- 25 May 2025
- Published by Association for Computing Machinery (ACM)
- p. 1-10
Open AccessDetection of Alcohol Inebriation from Eye Movements using Remote and Wearable Eye Trackers
- Open Access
- 25 May 2025
- Published by Association for Computing Machinery (ACM)
- p. 1-3
The More the Merrier: Boost Your Dataset Visibility and Discover Eye-Tracking Datasets with pymovements
- 25 May 2025
- Published by Association for Computing Machinery (ACM)
- p. 1-3
Open AccessThe More the Merrier: Boost Your Dataset Visibility and Discover Eye-Tracking Datasets with pymovements
- Open Access
- 25 May 2025
- Published by Association for Computing Machinery (ACM)
- p. 1-7
Proxy-Based Pre-Training for Eye-Tracking Applications
- 25 May 2025
- Published by Association for Computing Machinery (ACM)
- p. 1-7
Open AccessProxy-Based Pre-Training for Eye-Tracking Applications
- Open Access
- 22 May 2025
- Published by Association for Computing Machinery (ACM) in Proceedings of the ACM on Human-Computer Interaction
- Vol. 9 (3) , 1-15
Evaluating Gaze Event Detection Algorithms: Impacts on Machine Learning-based Classification and Psycholinguistic Statistical Modeling
Eye movements offer valuable, non-invasive insights into cognitive processes and are widely used in both psycholinguistic research and machine-learning applications, such as assessing reading comprehension and cognitive load. These applications typically rely on fixations and saccades detected through gaze event algorithms, which may be either proprietary or open-source. The impact of different gaze event detection algorithms on subsequent analysis is underexplored and often overlooked. This study investigates how two threshold-based algorithms, I-DT and I-VT, influence both machine-learning classification tasks and psycholinguistic statistical modeling. Using diverse datasets-including stationary, remote, and VR eye-tracking data across multiple sampling frequencies-our findings show significant differences in downstream performance. For ML tasks, I-DT generally outperforms I-VT, with I-VT being highly sensitive to threshold choices. In psycholinguistic analysis, results confirm established findings only when thresholds align with established fixation metrics, emphasizing the importance of appropriate threshold selection for meaningful analysis. Our code is publicly available: https://github.com/aeye-lab/eye-movement-preprocessing.
- 22 May 2025
- Published by Association for Computing Machinery (ACM) in Proceedings of the ACM on Human-Computer Interaction
- Vol. 9 (3) , 1-15
Open AccessEvaluating Gaze Event Detection Algorithms: Impacts on Machine Learning-based Classification and Psycholinguistic Statistical Modeling
Eye movements offer valuable, non-invasive insights into cognitive processes and are widely used in both psycholinguistic research and machine-learning applications, such as assessing reading comprehension and cognitive load. These applications typically rely on fixations and saccades detected through gaze event algorithms, which may be either proprietary or open-source. The impact of different gaze event detection algorithms on subsequent analysis is underexplored and often overlooked. This study investigates how two threshold-based algorithms, I-DT and I-VT, influence both machine-learning classification tasks and psycholinguistic statistical modeling. Using diverse datasets-including stationary, remote, and VR eye-tracking data across multiple sampling frequencies-our findings show significant differences in downstream performance. For ML tasks, I-DT generally outperforms I-VT, with I-VT being highly sensitive to threshold choices. In psycholinguistic analysis, results confirm established findings only when thresholds align with established fixation metrics, emphasizing the importance of appropriate threshold selection for meaningful analysis. Our code is publicly available: https://github.com/aeye-lab/eye-movement-preprocessing.
- Open Access
- 4 April 2024
- Published by Oxford University Press (OUP) in NAR Genomics and Bioinformatics
- Vol. 6 (2) , lqae043
Cancer drug sensitivity estimation using modular deep Graph Neural Networks
Computational drug sensitivity models have the potential to improve therapeutic outcomes by identifying targeted drugs components that are tailored to the transcriptomic profile of a given primary tumor. The SMILES representation of molecules that is used by state-of-the-art drug-sensitivity models is not conducive for neural networks to generalize to new drugs, in part because the distance between atoms does not generally correspond to the distance between their representation in the SMILES strings. Graph-attention networks, on the other hand, are high-capacity models that require large training-data volumes which are not available for drug-sensitivity estimation. We develop a modular drug-sensitivity graph-attentional neural network. The modular architecture allows us to separately pre-train the graph encoder and graph-attentional pooling layer on related tasks for which more data are available. We observe that this model outperforms reference models for the use cases of precision oncology and drug discovery; in particular, it is better able to predict the specific interaction between drug and cell line that is not explained by the general cytotoxicity of the drug and the overall survivability of the cell line. The complete source code is available at https://zenodo.org/doi/10.5281/zenodo.8020945. All experiments are based on the publicly available GDSC data.
- 4 April 2024
- Published by Oxford University Press (OUP) in NAR Genomics and Bioinformatics
- Vol. 6 (2) , lqae043
Open AccessCancer drug sensitivity estimation using modular deep Graph Neural Networks
Computational drug sensitivity models have the potential to improve therapeutic outcomes by identifying targeted drugs components that are tailored to the transcriptomic profile of a given primary tumor. The SMILES representation of molecules that is used by state-of-the-art drug-sensitivity models is not conducive for neural networks to generalize to new drugs, in part because the distance between atoms does not generally correspond to the distance between their representation in the SMILES strings. Graph-attention networks, on the other hand, are high-capacity models that require large training-data volumes which are not available for drug-sensitivity estimation. We develop a modular drug-sensitivity graph-attentional neural network. The modular architecture allows us to separately pre-train the graph encoder and graph-attentional pooling layer on related tasks for which more data are available. We observe that this model outperforms reference models for the use cases of precision oncology and drug discovery; in particular, it is better able to predict the specific interaction between drug and cell line that is not explained by the general cytotoxicity of the drug and the overall survivability of the cell line. The complete source code is available at https://zenodo.org/doi/10.5281/zenodo.8020945. All experiments are based on the publicly available GDSC data.
- Open Access
- 3 April 2024
- Published by Elsevier in JAACAP Open
- Vol. 3 (2) , 192-204
Saliency Models Reveal Reduced Top-Down Attention in Attention-Deficit/Hyperactivity Disorder: A Naturalistic Eye-Tracking Study
- 3 April 2024
- Published by Elsevier in JAACAP Open
- Vol. 3 (2) , 192-204
Open AccessSaliency Models Reveal Reduced Top-Down Attention in Attention-Deficit/Hyperactivity Disorder: A Naturalistic Eye-Tracking Study
- Open Access
- 4 March 2024
- Published by Elsevier in Computers & Graphics
- Vol. 119, 103901
Improving cognitive-state analysis from eye gaze with synthetic eye-movement data
- 4 March 2024
- Published by Elsevier in Computers & Graphics
- Vol. 119, 103901
Open AccessImproving cognitive-state analysis from eye gaze with synthetic eye-movement data
- Open Access
- 30 May 2023
- Published by Association for Computing Machinery (ACM)
- p. 1-9
SP-EyeGAN: Generating Synthetic Eye Movement Data with Generative Adversarial Networks
Neural networks that process the raw eye-tracking signal can outperform traditional methods that operate on scanpaths preprocessed into fixations and saccades. However, the scarcity of such data poses a major challenge. We, therefore, present SP-EyeGAN, a neural network that generates synthetic raw eye-tracking data. SP-EyeGAN consists of Generative Adversarial Networks; it produces a sequence of gaze angles indistinguishable from human micro- and macro-movements. We demonstrate how the generated synthetic data can be used to pre-train a model using contrastive learning. This model is fine-tuned on labeled human data for the task of interest. We show that for the task of predicting reading comprehension from eye movements, this approach outperforms the previous state-of-the-art.
- 30 May 2023
- Published by Association for Computing Machinery (ACM)
- p. 1-9
Open AccessSP-EyeGAN: Generating Synthetic Eye Movement Data with Generative Adversarial Networks
Neural networks that process the raw eye-tracking signal can outperform traditional methods that operate on scanpaths preprocessed into fixations and saccades. However, the scarcity of such data poses a major challenge. We, therefore, present SP-EyeGAN, a neural network that generates synthetic raw eye-tracking data. SP-EyeGAN consists of Generative Adversarial Networks; it produces a sequence of gaze angles indistinguishable from human micro- and macro-movements. We demonstrate how the generated synthetic data can be used to pre-train a model using contrastive learning. This model is fine-tuned on labeled human data for the task of interest. We show that for the task of predicting reading comprehension from eye movements, this approach outperforms the previous state-of-the-art.
- Open Access
- 30 May 2023
- Published by Association for Computing Machinery (ACM)
- p. 1-8
Bridging the Gap: Gaze Events as Interpretable Concepts to Explain Deep Neural Sequence Models
Recent work in XAI for eye tracking data has evaluated the suitability of feature attribution methods to explain the output of deep neural sequence models for the task of oculomotric biometric identification. These methods provide saliency maps to highlight important input features of a specific eye gaze sequence. However, to date, its localization analysis has been lacking a quantitative approach across entire datasets. In this work, we employ established gaze event detection algorithms for fixations and saccades and quantitatively evaluate the impact of these events by determining their concept influence. Input features that belong to saccades are shown to be substantially more important than features that belong to fixations. By dissecting saccade events into sub-events, we are able to show that gaze samples that are close to the saccadic peak velocity are most influential. We further investigate the effect of event properties like saccadic amplitude or fixational dispersion on the resulting concept influence.
- 30 May 2023
- Published by Association for Computing Machinery (ACM)
- p. 1-8
Open AccessBridging the Gap: Gaze Events as Interpretable Concepts to Explain Deep Neural Sequence Models
Recent work in XAI for eye tracking data has evaluated the suitability of feature attribution methods to explain the output of deep neural sequence models for the task of oculomotric biometric identification. These methods provide saliency maps to highlight important input features of a specific eye gaze sequence. However, to date, its localization analysis has been lacking a quantitative approach across entire datasets. In this work, we employ established gaze event detection algorithms for fixations and saccades and quantitatively evaluate the impact of these events by determining their concept influence. Input features that belong to saccades are shown to be substantially more important than features that belong to fixations. By dissecting saccade events into sub-events, we are able to show that gaze samples that are close to the saccadic peak velocity are most influential. We further investigate the effect of event properties like saccadic amplitude or fixational dispersion on the resulting concept influence.