.png){kind=icon}
.png){kind=icon}
Return
Artificial intelligence-driven automated forensic data anomaly detection using blockchain-based integrity verification
.png){kind=icon}
PRE
AI
DOI：10.1016/j.engappai.2026.115740.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
In Internet of Things (IoT) systems, blockchain and Artificial Intelligence (AI) techniques have a major trend due to their significant. However, security, efficiency, and scalability are challenged due to the integration of AI techniques, which generate redundant data. To overcome these challenges, a novel Efficient Channel Attention-assisted Dense Transformer (ECA-deTrans) based data anomaly detection with blockchain is proposed for forensic evidence integrity verification. Initially, the raw data is collected from the repository and pre-processed. Feature extraction is performed using feature extraction using a Hybrid Autoencoder with 1 Dimensional Convolutional Bidirectional Long Short-Term Memory (HAE-1DBiLSTM). Based on the extracted features, ECA-deTrans performs forensic anomaly detection. Then, the smart contract system performs blockchain based integrity verification using Keccak-based Secure Hash Algorithm-3 with Enhanced Logistic Map (SHA-3-ELMC) chaotic algorithm. The blockchain creates an immutable audit trail that guarantees accountability, transparency, and compliance by permanently storing all AI discoveries, smart contract operations, and verification outcomes. Based on the simulation, the accuracy of 99.74%, precision of 99.48%, recall of 99.63%, F1-score of 99.55%, Mean Average Error (MAE) of 0.58%, throughput of 28.251 transactions per second (tps), and latency of 0.395 s were achieved. The experimental results demonstrate the significant of the proposed approach for real-time applicability.
Journal
IF:
8
Papers:
5.3K
Citations:
3.5W
View more organizations