Table 1 Comparisons for recent works.
Authors | Focus | Technique | Advantages | Limitations |
|---|---|---|---|---|
Chen et al.27 | Semantic-based vulnerability detection | Residual Graph Convolutional Networks (GCN) with Edge Attention | Captures contextual information in smart contract code effectively | Edge attention mechanism increases computational complexity |
Ma et al.28 | Graph-based vulnerability detection | Hierarchical Graph Attention Network (HGAT) | Improves detection accuracy by considering hierarchical relationships | Requires high-quality labeled data for optimal performance |
Jie et al.29 | Blockchain programs for IoT use contract-oriented languages, executed automatically | Blockchain technology | Enhancing IoT solutions with technical capabilities | Traditional methods rely on rules, struggle with false positives, low accuracy |
Osei et al.31 | Wide and DL for vulnerability detection | Wide and Deep Neural Network (WDNN) | Combines memorization and generalization capabilities | May struggle with unseen vulnerability patterns |
Sharma et al.32 | IoT security layers discussed, DL model proposed for intrusion detection, explainable AI used for high accuracy | Intrusion Detection System (IDS) | Detect intrusions in IoT networks, prevent malicious activities efficiently | IoT systems face daily attacks, identification and mitigation needed for network protection |
Zhen et al.33 | Smart contract vulnerability detection | Dual Attention Graph Neural Network (DA-GNN) | Improves accuracy by leveraging dual attention for better feature extraction | Computationally intensive due to complex graph processing |
He et al.21 | Enhancing smart contract security | Pre-trained language models for vulnerability detection | Leverages NLP advancements for precise vulnerability identification | Requires extensive training data and high computational resources |
Wu et al.25 | Smart contract vulnerability detection using hybrid attention | Combine self-attention and convolutional layers | Improved accuracy, handles complex patterns well | Computationally expensive, struggles with large datasets |
Wu et al.34 | Improved dual-channel technique for identifying vulnerabilities in smart contracts | Dual-channel approach for syntactic and semantic analysis | Enhanced detection performance and faster processing | Increased computational cost and scalability issues |
Mothukuri et al.35 | Trust scoring and vulnerability detection in DeFi projects | XGBoost | Multi-perspective evaluation with Enhanced transparency and trust | May not scale efficiently across all DeFi platforms |
Yazdinejad et al.26 | Threat detection in IoT networks using blockchain | DNN | Decentralized architecture enhances security | High resource requirements for fuzzy logic and blockchain computation |
