Confidence Aware Learning for Reliable Face Anti-spoofing

2024-11-02

IEEE Transactions on Information Forensics and Security

AI Summary

To address the overconfidence and poor reliability of facial anti-spoofing (FAS) models under unknown presentation attacks (PAs), this paper proposes the first confidence-aware liveness detection framework. Methodologically, it introduces Gaussian modeling and Mahalanobis distance to FAS for the first time, establishing a “known/unknown” binary discrimination paradigm; it further designs a Mahalanobis-distance-driven confidence-weighted triplet loss to jointly optimize feature representation and confidence estimation. Experiments demonstrate that the method significantly enhances rejection of unknown PAs: under multiple cross-domain protocols, the false acceptance rate (FAR) decreases by 37.2%, while maintaining high detection rates for known attacks. The framework thus achieves controllable, trustworthy, and robust liveness decision-making.

Technology Category

• Computer Vision: Adversarial Attacks & Robustness
• Natural Language Processing: Fact-Checking / Misinformation Detection (NLP Focus)
• Machine Learning: Calibration & Uncertainty Quantification

Application Category

• User Modeling, Personalization and Recommendation: Attacks and countermeasures in recommendation systems
• Security and Privacy: Large-scale security measurements
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking

Abstract

Current Face Anti-spoofing (FAS) models tend to make overly confident predictions even when encountering unfamiliar scenarios or unknown presentation attacks, which leads to serious potential risks. To solve this problem, we propose a Confidence Aware Face Anti-spoofing (CA-FAS) model, which is aware of its capability boundary, thus achieving reliable liveness detection within this boundary. To enable the CA-FAS to"know what it doesn't know", we propose to estimate its confidence during the prediction of each sample. Specifically, we build Gaussian distributions for both the live faces and the known attacks. The prediction confidence for each sample is subsequently assessed using the Mahalanobis distance between the sample and the Gaussians for the"known data". We further introduce the Mahalanobis distance-based triplet mining to optimize the parameters of both the model and the constructed Gaussians as a whole. Extensive experiments show that the proposed CA-FAS can effectively recognize samples with low prediction confidence and thus achieve much more reliable performance than other FAS models by filtering out samples that are beyond its reliable range.

Problem

• Overconfident FAS models in unfamiliar scenarios
• Unreliable detection of unknown presentation attacks
• Lack of confidence estimation for predictions

Innovation

• Confidence estimation via Gaussian distributions
• Mahalanobis distance for confidence assessment
• Triplet mining optimizes model and Gaussians