ExplainabilityTransparencyTrustworthinesscs.CV
An Efficient Ensemble Explainable AI (XAI) Approach for Morphed Face Detection
Rudresh Dwivedi, Ritesh Kumar, Deepak Chopra, Pranay Kothari, Manjot Singh
April 23, 2023
Pattern Recognition Letters
20 citations
1 influential
Key Takeaways
-
Morphed face attacks pose a real security threat to biometric systems: The paper addresses a genuine vulnerability where attackers create synthetic images by merging faces to fool authentication systems used in critical applications like border control and criminal databases.
-
AI systems used for security need to be transparent and explainable: The researchers highlight that while deep learning models perform well at detecting morphed faces, they operate as "black boxes" that don't reveal why they make decisions, which can lead to incorrect judgments and undermine trust in security systems.
-
Multiple explanation techniques together provide better insights than one alone: The study combines three different visual explanation methods (Saliency maps, Class Activation Maps, and Grad-CAM) to create more comprehensive explanations of how the AI model identifies morphed faces, suggesting that layered approaches to AI transparency may be more reliable.
-
Explainability is essential for high-stakes biometric applications: The authors emphasize that for security-critical uses—such as border control and criminal identification—stakeholders need to understand and trust how AI systems reach their conclusions, not just accept their accuracy rates.
Abstract
The extensive utilization of biometric authentication systems have emanated attackers / imposters to forge user identity based on morphed images. In this attack, a synthetic image is produced and merged with genuine. Next, the resultant image is user for authentication. Numerous deep neural convolutional architectures have been proposed in literature for face Morphing Attack Detection (MADs) to prevent such attacks and lessen the risks associated with them. Although, deep learning models achieved optimal results in terms of performance, it is difficult to understand and analyse these networks since they are black box/opaque in nature. As a consequence, incorrect judgments may be made. There is, however, a dearth of literature that explains decision-making methods of black box deep learning models for biometric Presentation Attack Detection (PADs) or MADs that can aid the biometric community to have trust in deep learning-based biometric systems for identification and authentication in various security applications such as border control, criminal database establishment etc. In this work, we present a novel visual explanation approach named Ensemble XAI integrating Saliency maps, Class Activation Maps (CAM) and Gradient-CAM (Grad-CAM) to provide a more comprehensive visual explanation for a deep learning prognostic model (EfficientNet-B1) that we have employed to predict whether the input presented to a biometric authentication system is morphed or genuine. The experimentations have been performed on three publicly available datasets namely Face Research Lab London Set, Wide Multi-Channel Presentation Attack (WMCA), and Makeup Induced Face Spoofing (MIFS). The experimental evaluations affirms that the resultant visual explanations highlight more fine-grained details of image features/areas focused by EfficientNet-B1 to reach decisions along with appropriate reasoning.