Academic Whitepapers & Literature Analysis
Explore the theoretical grounding, empirical evidence, and peer-reviewed literature behind CORROBORATE and the CHIMERA adversary model.
The Processed Replay Blindspot: Why Modern Bot Defenses Fail
Analyzing the unresolved gap highlighted by IEEE studies (Salman & Bicakci 2026, Sadeghpour & Vlajic 2024)
Traditional bot detectors rely on synthetic-movement regularities or historical duplicates. When an attacker cuts and splices real human session fragments, single-channel detectors are blind.
The Biomechanics of Human Attention: Grounding Cross-Modal Coupling
How KLM homing times and gaze-cursor latencies define natural human interaction
Why humans cannot type and move a mouse simultaneously, and how the ~400ms Card-Moran-Newell homing operator gives us a mathematical foundation for seam localization.
Inside CHIMERA: Synthesizing Frankenstein Sessions
Pause-boundary segmentation, similarity retargeting, and cross-person keystroke overlays
A deep dive into how CHIMERA splices genuine sessions at pause boundaries to deceive single-channel classifiers while preserving velocity continuity.
From Audio-Visual Deepfakes to Web Biometrics
Adapting self-supervised dissonance detection (Chugh et al., Feng et al.) to web sessions
How CORROBORATE borrows the 'train-on-genuine-only' segment-then-aggregate structure from audio-visual deepfake forensics to prevent overfitting to attack generators.