Dist‐ <scp>PGQBM</scp> : Distributed Pyramidal Graph Attention‐Enabled Deep Learning Model for Manipulated Timestamp Detection From Raw Disk Image

ABSTRACT Timestamp manipulation affects the reliability of the digital forensic investigation process. Hence, numerous traditional approaches are utilized to perform meaningful investigation, but they are limited in their inability to eradicate the occurrence of redundant details, increased computational complexity, and reduced interpretability. Thus, an effective Distributed Pyramidal Graph attention‐enabled Quantum Bidirectional Long short term Memory (Dist‐PGQBM) is proposed. The extraction of the optimal features mitigates the occurrence of redundant data, thereby enhancing the accurate detection of the manipulated timestamp. The distributed learning strategy followed by the research aids in the parallel processing of complex features with the effective reduction in the computational resource requirement. Moreover, the Pyramidal Graph attention (PGA) incorporated in the research enhanced the ability of the proposed model to handle complex temporal patterns via the effective handling of the noisy inputs with increased interpretability. Nevertheless, the quantum layer integrated with the Dist‐PGQBM made the manipulated timestamp detection a faster process. Experimental results show that the proposed Dist‐PGQBM method outperforms the other existing methods by attaining the high accuracy of 97.76%, precision of 98.18%, recall of 97.34%, and Mean Absolute Error (MAE) of 1.67 for 90% of training using the msuhanov/dfir_ntfs dataset.