@inproceedings{alsheikh2026_114337,
  title = {Proceedings of the Second Workshop on AI in Production},
  author = {Alsheikh, Ahmad and Fischer, Andreas and Banf, Michael and Kuhn, Johannes and Fichtel, Lars and Seuß, Dominik and Marchl, Andreas and Kasper, Maximilian and Geis, Simon and Deutel, Mark and Plinge, Axel and Mattick, Alexander and Sturm, Louis and Müller, Felix and Loftus, Sebastian and Asif, Haris and Mutschler, Christopher and Pahl, Jasper and Göttl, Quirin and Mayat, Nils and Schönekehs, Chris and Padrón Hinrichs, Marcos and Weisser, Tim and Schmitt, Robert},
  editor = {Knoblach, Judith and Krockert, Martin and Bahr, Lukas and Voigt, Erik},
  year = {2026},
  publisher = {Otto-Friedrich-Universität},
  address = {Bamberg},
  doi = {10.20378/irb-114337},
  urn = {urn:nbn:de:bvb:473-irb-114337x},
  url = {https://fis.uni-bamberg.de/handle/uniba/114337},
  language = {eng},
  abstract = {The 2nd Workshop on Artificial Intelligence in Production (AIP) was held as part of the 48th German Conference on Artificial Intelligence in Potsdam on September 16, 2025.
The workshop was organised jointly by academic and industrial partners from the University of Bamberg, HTW Dresden, and the BMW Group. The aim of the AIP workshop is to bring together researchers and practitioners from Artificial Intelligence (AI) and production to exchange ideas, experiences, and solutions for applying AI across production planning, control, and optimisation. By fostering interdisciplinary dialogue and sharing transferable approaches, the workshop seeks to advance effective, scalable, and sustainable AI applications in production and related domains, and strengthening the connection between academic research and industrial practice.
In the 2025 edition, seven researchers and practitioners presented their latest work in the field. The workshop received eight submissions, of which seven were accepted after a peer-review process involving two members of the programme committee per paper. We would like to thank all authors, reviewers, and participants for their valuable contributions, which made the workshop a success.},
  keywords = {AI, Production}
}

@article{rieger2025_112139,
  title = {FMC-Net : A Human-Guided Deep Learning Framework for Adaptable and Transparent Facial Expression Recognition in Real-World Scenarios},
  author = {Rieger, Ines and Pahl, Jasper and Schmid, Ute},
  year = {2025},
  pages = {1-19},
  publisher = {Otto-Friedrich-Universität},
  address = {Bamberg},
  urn = {urn:nbn:de:bvb:473-irb-112139x},
  url = {https://fis.uni-bamberg.de/handle/uniba/112139},
  language = {eng},
  abstract = {We introduce FMC-Net, a facial expression recognition (FER) framework that leverages the hierarchical relationship between discrete facial muscle movements, known as Action Units (AUs), and Facial Expressions (FEs) by integrating two complementary constraint layers. This framework couples data-driven learning with psychology-grounded structure. First, a training-time correlation constraint aligns the two tasks within a multi-task network by softly regularizing a target statistical relationship. This can improve sample efficiency and generalization, particularly under limited or biased data. Second, an inference-time fuzzy rule layer maps the networks probabilistic AU predictions to FEs using compact, human-editable from psychological research, yielding transparent, per-decision attributions. An ensemble then combines the model and rule-based pathways and exposes a disagreement-based risk score for human-in-the-loop triage. This two-layer constraint integration addresses the limitations of single-mechanism approaches: training-time constraints shape the learned representations but lack case-wise transparency, while inference-time rules explain decisions but cannot improve the underlying features. Experiments across diverse datasets, including in-the-wild video and cross-dataset evaluation, validate our approach. Constraint-guided training consistently produces models that outperform competitive baselines, while the rule-based pathway can provide transparency and actionable risk signals towards reliable deployment. The proposed methodology is also generalizable to other machine learning tasks with interdependent outputs.},
  keywords = {Deep learning, Facial expression recognition, Constraint integration, Human-guided AI}
}

@article{rieger2025_112138,
  title = {FMC-Net : A Human-Guided Deep Learning Framework for Adaptable and Transparent Facial Expression Recognition in Real-World Scenarios},
  author = {Rieger, Ines and Pahl, Jasper and Schmid, Ute},
  year = {2025},
  journal = {Applied intelligence : the international journal of artificial intelligence, neural networks, and complex problem-solving technologies},
  volume = {55},
  number = {18, 1127},
  pages = {1-19},
  publisher = {Springer Science + Business Media B.V},
  address = {Dordrecht [u.a.]},
  issn = {1573-7497},
  doi = {10.1007/s10489-025-07017-9},
  url = {https://fis.uni-bamberg.de/handle/uniba/112138},
  language = {eng},
  abstract = {We introduce FMC-Net, a facial expression recognition (FER) framework that leverages the hierarchical relationship between discrete facial muscle movements, known as Action Units (AUs), and Facial Expressions (FEs) by integrating two complementary constraint layers. This framework couples data-driven learning with psychology-grounded structure. First, a training-time correlation constraint aligns the two tasks within a multi-task network by softly regularizing a target statistical relationship. This can improve sample efficiency and generalization, particularly under limited or biased data. Second, an inference-time fuzzy rule layer maps the networks probabilistic AU predictions to FEs using compact, human-editable from psychological research, yielding transparent, per-decision attributions. An ensemble then combines the model and rule-based pathways and exposes a disagreement-based risk score for human-in-the-loop triage. This two-layer constraint integration addresses the limitations of single-mechanism approaches: training-time constraints shape the learned representations but lack case-wise transparency, while inference-time rules explain decisions but cannot improve the underlying features. Experiments across diverse datasets, including in-the-wild video and cross-dataset evaluation, validate our approach. Constraint-guided training consistently produces models that outperform competitive baselines, while the rule-based pathway can provide transparency and actionable risk signals towards reliable deployment. The proposed methodology is also generalizable to other machine learning tasks with interdependent outputs.},
  keywords = {Deep learning, Facial expression recognition, Constraint integration, Human-guided AI}
}

@inproceedings{rieger2022_58025,
  title = {CorrLoss : Integrating Co-Occurrence Domain Knowledge for Affect Recognition},
  author = {Rieger, Ines and Pahl, Jaspar and Finzel, Bettina and Schmid, Ute},
  year = {2022},
  booktitle = {2022 26th International Conference on Pattern Recognition (ICPR): Proceedings},
  pages = {798-804},
  publisher = {IEEE},
  isbn = {978-1-6654-9062-7},
  doi = {10.1109/ICPR56361.2022.9956319},
  url = {https://fis.uni-bamberg.de/handle/uniba/58025},
  language = {eng}
}

@inproceedings{pahl2022_55613,
  title = {Female, white, 27? : Bias Evaluation on Data and Algorithms for Affect Recognition in Faces},
  author = {Pahl, Jaspar and Rieger, Ines and Möller, Anna and Wittenberg, Thomas and Schmid, Ute},
  year = {2022},
  pages = {973-987},
  publisher = {Otto-Friedrich-Universität},
  address = {Bamberg},
  urn = {urn:nbn:de:bvb:473-irb-556138},
  url = {https://fis.uni-bamberg.de/handle/uniba/55613},
  language = {eng},
  abstract = {Nowadays, Artificial Intelligence (AI) algorithms show a strong performance for many use cases, making them desirable for real-world scenarios where the algorithms provide high-impact decisions. However, one major drawback of AI algorithms is their susceptibility to bias and resulting unfairness. This has a huge influence for their application, as they have a higher failure rate for certain subgroups. In this paper, we focus on the field of affective computing and particularly on the detection of bias for facial expressions. Depending on the deployment scenario, bias in facial expression models can have a disadvantageous impact and it is therefore essential to evaluate the bias and limitations of the model. In order to analyze the metadata distribution in affective computing datasets, we annotate several benchmark training datasets, containing both Action Units and categorical emotions, with age, gender, ethnicity, glasses, and beards. We show that there is a significantly skewed distribution, particularly for ethnicity and age. Based on this metadata annotation, we evaluate two trained state-of-the-art affective computing algorithms. Our evaluation shows that the strongest bias is in age, with the best performance for persons under 34 and a sharp decrease for older persons. Furthermore, we see an ethnicity bias with varying direction depending on the algorithm, a slight gender bias and worse performance for facial parts occluded by glasses.},
  keywords = {affective computing, action units, categorical emotions, metadata post-annotation, bias, fairness, data evaluation,, algorithm evaluation}
}

@inproceedings{finzel2022_55380,
  title = {Deriving Temporal Prototypes from Saliency Map Clusters for the Analysis of Deep-Learning-based Facial Action Unit Classification},
  author = {Finzel, Bettina and Kollmann, Rene and Rieger, Ines and Pahl, Jaspar and Schmid, Ute},
  year = {2022},
  pages = {86-97},
  publisher = {Otto-Friedrich-Universität},
  address = {Bamberg},
  urn = {urn:nbn:de:bvb:473-irb-553809},
  url = {https://fis.uni-bamberg.de/handle/uniba/55380},
  language = {eng},
  abstract = {Reliably determining the emotional state of a person is a difficult task for both humans as well as machines. Automatic detection and evaluation of facial expressions is particularly important if people are unable to express their emotional state themselves, for example due to cognitive impairments. Identifying the presence of Action Units in a human’s face is a psychologically validated approach of quantifying which emotion is expressed. To automate the detection process of Action Units Neural Networks have been trained. However, the black-box nature of Deep Neural Networks provides no insight on the relevant features identified during the decision process. Approaches of Explainable Artificial Intelligence have to be applied to provide an explanation why the network came to a certain conclusion. In this work "Layer-Wise Relevance Propagation" (LRP) in combination with the meta analysis approach "Spectral Relevance Analysis" (SpRAy) is used to derive temporal prototypes from predictions in video sequences. Temporal prototypes provide an aggregated view on the prediction of the network by grouping together similar frames by considering relevance. Additionally, a specific visualization method for temporal prototypes is presented that highlights the most relevant areas for a prediction of an Action Unit. A quantitative evaluation of our approach shows that temporal prototypes aggregate temporal information well. The proposed method can be used to generate concise visual explanations for a sequence of interpretable saliency maps. Based on the above, this work shall provide the foundation for a new temporal analysis method as well as an explanation approach that is supposed to help researchers and experts to gain a deeper understanding of how the underlying network decides which Action Units are active in a particular emotional state.},
  keywords = {Temporal Prototypes, Facial Action Unit Detection, Layer-wise Relevance Propagation, Spectral Clustering, Affective Computing, Correlation Analysis}
}

@inproceedings{pahl2022_55036,
  title = {Female, white, 27? : Bias Evaluation on Data and Algorithms for Affect Recognition in Faces},
  author = {Pahl, Jaspar and Rieger, Ines and Möller, Anna and Wittenberg, Thomas and Schmid, Ute},
  year = {2022},
  booktitle = {FAccT '22: 2022 ACM Conference on Fairness, Accountability, and Transparency : 2022 Proceeding},
  pages = {973-987},
  publisher = {Association for Computing Machinery},
  isbn = {978-1-4503-9352-2},
  doi = {10.1145/3531146.3533159},
  url = {https://fis.uni-bamberg.de/handle/uniba/55036},
  language = {eng},
  abstract = {Nowadays, Artificial Intelligence (AI) algorithms show a strong performance for many use cases, making them desirable for real-world scenarios where the algorithms provide high-impact decisions. However, one major drawback of AI algorithms is their susceptibility to bias and resulting unfairness. This has a huge influence for their application, as they have a higher failure rate for certain subgroups. In this paper, we focus on the field of affective computing and particularly on the detection of bias for facial expressions. Depending on the deployment scenario, bias in facial expression models can have a disadvantageous impact and it is therefore essential to evaluate the bias and limitations of the model. In order to analyze the metadata distribution in affective computing datasets, we annotate several benchmark training datasets, containing both Action Units and categorical emotions, with age, gender, ethnicity, glasses, and beards. We show that there is a significantly skewed distribution, particularly for ethnicity and age. Based on this metadata annotation, we evaluate two trained state-of-the-art affective computing algorithms. Our evaluation shows that the strongest bias is in age, with the best performance for persons under 34 and a sharp decrease for older persons. Furthermore, we see an ethnicity bias with varying direction depending on the algorithm, a slight gender bias and worse performance for facial parts occluded by glasses.},
  keywords = {affective computing, action units, categorical emotions, metadata post-annotation, bias, fairness, data evaluation,, algorithm evaluation}
}

@inproceedings{finzel2021_55037,
  title = {Deriving Temporal Prototypes from Saliency Map Clusters for the Analysis of Deep-Learning-based Facial Action Unit Classification},
  author = {Finzel, Bettina and Kollmann, Rene and Rieger, Ines and Pahl, Jaspar and Schmid, Ute},
  year = {2021},
  booktitle = {Proceedings of the LWDA 2021 Workshops: FGWM, KDML, FGWI-BIA, and FGIR},
  pages = {86-97},
  publisher = {RWTH Aachen},
  address = {Aachen, Germany},
  url = {https://fis.uni-bamberg.de/handle/uniba/55037},
  language = {eng},
  abstract = {Reliably determining the emotional state of a person is a difficult task for both humans as well as machines. Automatic detection and evaluation of facial expressions is particularly important if people are unable to express their emotional state themselves, for example due to cognitive impairments. Identifying the presence of Action Units in a human’s face is a psychologically validated approach of quantifying which emotion is expressed. To automate the detection process of Action Units Neural Networks have been trained. However, the black-box nature of Deep Neural Networks provides no insight on the relevant features identified during the decision process. Approaches of Explainable Artificial Intelligence have to be applied to provide an explanation why the network came to a certain conclusion. In this work "Layer-Wise Relevance Propagation" (LRP) in combination with the meta analysis approach "Spectral Relevance Analysis" (SpRAy) is used to derive temporal prototypes from predictions in video sequences. Temporal prototypes provide an aggregated view on the prediction of the network by grouping together similar frames by considering relevance. Additionally, a specific visualization method for temporal prototypes is presented that highlights the most relevant areas for a prediction of an Action Unit. A quantitative evaluation of our approach shows that temporal prototypes aggregate temporal information well. The proposed method can be used to generate concise visual explanations for a sequence of interpretable saliency maps. Based on the above, this work shall provide the foundation for a new temporal analysis method as well as an explanation approach that is supposed to help researchers and experts to gain a deeper understanding of how the underlying network decides which Action Units are active in a particular emotional state.},
  keywords = {Temporal Prototypes, Facial Action Unit Detection, Layer-wise Relevance Propagation, Spectral Clustering, Affective Computing, Correlation Analysis}
}

@inproceedings{pahl2020_50306,
  title = {Multi-label Learning with Missing Values using Combined Facial Action Unit Datasets},
  author = {Pahl, Jaspar and Rieger, Ines and Seuß, Dominik},
  year = {2020},
  publisher = {arXiv},
  url = {https://fis.uni-bamberg.de/handle/uniba/50306},
  language = {eng}
}

@inproceedings{rieger2020_50307,
  title = {Unique Class Group Based Multi-Label Balancing Optimizer for Action Unit Detection},
  author = {Rieger, Ines and Pahl, Jaspar and Seuß, Dominik},
  year = {2020},
  booktitle = {2020: 15th IEEE International Conference on Automatic Face and Gesture Recognition (FG 2020), proceedings},
  pages = {619-623},
  publisher = {IEEE},
  isbn = {978-1-7281-3079-8},
  doi = {10.1109/FG47880.2020.00101},
  url = {https://fis.uni-bamberg.de/handle/uniba/50307},
  language = {eng}
}

