Conference Report on IEEE BIOSIG 2023
The 22nd edition of the International Conference of the Biometrics Special Interest Group (BIOSIG) took place at Fraunhofer IGD in Darmstadt, Germany from September 20 to 22. A total number of 87 participants attended BIOSIG 2023. The conference featured 18 oral presentations and 15 poster presentations, as well as 3 keynotes by Frøy Løvåsdal (National Police Directorate, Norway), Arun Ross (Michigan State University), and Ignacio López Moreno (Google NYC). This event was the last part of the Darmstadt Biometric Week.
The conference was opened with a session on bias and explainability in biometric systems. All three presentations focused on facial recognition, being the first presentation dedicated to a new explainability metric. Subsequently, two speakers discussed the impact of quantization on bias and how to measure bias across different demographic groups in terms of False Non-Match Rates (FNMR). After a short break, hand-based recognition was the main topic: first tattoos and their impact on recognition accuracy, and then palmprint recognition for children.
| The first keynote talk was given by Frøy Løvåsdal, from the National Police Directorate of Norway. She addressed the challenges of human examiners within the police forces and described the most recent actions undertaken to tackle such issues. She described not only technical aspects related to equipment, but also the complexity of for instance developing and implementing adequate training for the officers. This highlighted the long way that still needs to be covered, and the necessity for further collaboration between officers in the field and stakeholders with a more technical background. |
The keynote was followed by presentations devoted to finger vein recognition, keystroke dynamics, and several aspects of voice recognition, namely: synthetic speech and whether humans can detect it, voice morphing, and fairness and privacy aspects.
Before the social event, further conference contributions were presented in the poster session. The topics included different aspects of systems based on faces, contactless fingerprints, veins, silhouettes, or handwriting. Also more general topics such as explainability, fairness, or template protection were covered.
Friday started with the keynote talk of Arun Ross, from Michigan State University (MSU), on deep learning and trust. The significant improvements achieved with deep learning algorithms in different aspects of biometrics, ranging from basic recognition accuracy for constrained and unconstrained scenarios, to privacy and security aspects of these systems (e.g., presentation attack detection), were described. Then, the main issue towards a further deployment and acceptance of biometric recognition was highlighted: trust in the technology. Different aspects contributing to trust were then considered: explainability, bias, and compliance to policies and regulations. To conclude the talk, the next steps towards building more trust were discussed. |
|
Before the social event, further conference contributions were presented in the poster session. The topics included different aspects of vascular biometrics, face, fingerprint, iris and gait recognition, biometric quality, ageing, and presentation attack detection.
| The conference concluded with the keynote talk of Ignacio López Moreno, from Google NYC. He gave an overview of speech-based technologies in Google, and how algorithms have evolved in the last decade, from a technical and research perspective. He focused on different aspects, from speaker to speech recognition, from the first handcrafted features employed years ago to the latest deep learning methods. He also discussed how the subject’s privacy is protected, using the latest federated learning methods in the state of the art. |
As in previous editions of the BIOSIG conference, participants of the conference themselves voted for the best paper and the best poster presented at the conference. The BIOSIG 2022 “Best Paper Award” was given to Pedro Neto (University of Porto) for the interesting presentation “Compressed Models Decompress Race Biases: What Quantized ModelsForget for Fair Face Recognition“. The winner of the “Best Poster Award” was Satya Sai Siva Rama Krishna Akula (University of Missouri-Kansas City) for the poster “A Wrist-worn Diffuse Optical Tomography Biometric System”.
The BIOSIG conference was preceded by several satellite events in the Darmstadt Biometric Week:
- TeleTrusT Biometric Working Group Meeting
- EAB Research Projects Conference (EAB-RPC) 2023
- EAB General Assembly
- EAB Biometrics Max Snijder, Research, and Industry Awards 2023
The 2023 BIOSIG conference was jointly organized by the Competence Center for Applied Security Technology (CAST) and the special interest group BIOSIG of the Gesellschaft für Informatik e.V. (GI).
The conference was technically co-sponsored by IEEE Biometric Council and the papers will be added to IEEE Xplore.
For more details on the next edition of the conference, see: www.biosig.org