{"data":{"id":"2b3293a4-48c8-45bd-b066-1d1c9d406f91","title":"Toward High Accuracy and Strong Security: Cancellable Templates for Multimodal Biometric Recognition Based on Feature Fusion","summary":"This research proposes a secure system for multimodal biometric recognition (using multiple biological measurements like palmprints and vein patterns together) that protects user privacy by creating cancellable templates (protected versions of biometric data that can't be reversed to recover the original). The system uses a neural network to combine features from different biometric sources and a novel random projection method to generate templates that are irreversible, revocable (can be replaced if compromised), and resistant to various attacks.","solution":"The proposed mitigation involves a 'novel cancellable random projection method, which generates protected templates through SoftMax-based random projection (SoRP) combined with other hashing algorithms, effectively avoiding the reversibility problem of random projection under certain conditions.' The framework is designed to satisfy security requirements including 'irreversibility, revocability, unlinkability, and resistance to various attacks.'","labels":["research","security"],"sourceUrl":"http://ieeexplore.ieee.org/document/11653473","publishedAt":"2026-08-12T13:16:39.000Z","cveId":null,"cweIds":null,"cvssScore":null,"cvssSeverity":null,"severity":"info","attackType":[],"issueType":"research","affectedPackages":null,"affectedVendors":[],"affectedVendorsRaw":[],"classifierModel":"claude-haiku-4-5-20251001","classifierPromptVersion":"v3","cvssVector":null,"attackVector":null,"attackComplexity":null,"privilegesRequired":null,"userInteraction":null,"exploitMaturity":null,"epssScore":null,"patchAvailable":null,"disclosureDate":"2026-08-12T13:16:39.000Z","capecIds":null,"crossRefCount":0,"attackSophistication":"moderate","impactType":["confidentiality","integrity"],"aiComponentTargeted":"model","llmSpecific":false,"classifierConfidence":0.75,"researchCategory":"peer_reviewed","atlasIds":null}}