Tuğçe Arican from the University of Twente, will present her research about "Patch-Based Finger Vein Recognition Using an Auto Encoder". Auto Encoders are neural networks that use unsupervised learning methods, and aim to reconstruct their input through a series of compression and de-compression steps. Tuğçe will shed light on how the model learns a compact representation of the input, without the need of labeled data, during this process. Furthermore, Tuğçe stresses the convenience of this property if ground truth labels are hard-to-realize for large data sets, such as for finger vein data. The compact representation can be regarded as a feature vector which can be used to compare finger vein images. The contrast of veins in the images is very low. Therefore, instead of the full image, small finger vein patches will be utilized as an input to the Auto Encoder.
Curriculum vitæ
Tuğçe is a PhD.candidate at the University of Twente, at the Data Management Biometrics Group. She obtained her Bachelor degree in Computer Engineering at the Osmangazi University of Turkey. She was likewise granted with a scholarship provided by the Republic of Turkey Ministry of Education. She moved to Enschede and pursued a Masters degree in Computer Science at the University of Twente. Currently, Tuğçe is conducting research on finger vein recognition using unsupervised learning methods.
We thank our media partner:
