Deep Residual Network With Adaptive Learning Framework for Fingerprint Liveness Detection
Why this work is in the frame
A frame that forgets how it found something cannot be audited. These are the routes that admitted this work.
Bibliographic record
Abstract
Today, fingerprint recognition technology has aroused wide attention in the society, especially in the application of identity authentication with a smartphone as a carrier. However, the disadvantage of these devices is that the identification sensors are vulnerable to spoofing attacks from artificial replicas made from clay, gelatin, silicon, etc. To resolve it, a feasible anti-deception countermeasure, called fingerprint liveness detection (FLD), has been proposed. Different from most shallow feature methods, the deep convolutional neural network (DCNN)-based FLD methods have been widely explored with the properties of fast operation, few parameters, and end-to-end feature self-learning. Meanwhile, DCNN faces a pair of contradictory problems, on the one hand, the training accuracy will keep rising with the increasement of multilayer perceptron (MLP), finally tends to a stable value. Continue to increase the number of MLP, results will decline. Much research, on the other hand, shows that the number of MLP is the foundation for realizing a high performance detection. Hereby, we apply deep residual network (DRN) to FLD for the first time to solve the contradiction mentioned in this paper. Next, to eliminate the interference of invalid regions of given images, a region-of-interest (ROI) extraction algorithm is put forward. Afterward, to avoid the parameters learned plunging into local optimization, adaptive learning-based DRNs (ALDRNs), which automatically adjust the learning rate if those monitoring parameters (verification accuracy) are stable, are explored. Finally, we propose a novel texture enhancement based on the local gradient pattern (LGP) method to improve the generalization of a model classifier as well. Experimental results on three benchmark data sets: LivDet 2011, 2013, and 2015, show that our results outperform the state-of-the-art FLD methods.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
Full frame distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.000 | 0.000 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.000 | 0.001 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.000 | 0.000 |
| Open science | 0.000 | 0.000 |
| Research integrity | 0.000 | 0.000 |
| Insufficient payload (model declined to judge) | 0.000 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from it