Advancements in Few-Shot Learning for Network Intrusion Detection

• •Few-shot learning is crucial for effective NIDS training amid limited labeled data.
• •A systematic review identified 21 studies employing various FSL techniques from 2022 to 2026.
• •Standardization in evaluation practices is needed to improve reproducibility and comparison.

Recent research highlights the challenges in training anomaly-based network intrusion detection systems (NIDS) due to the scarcity of labeled attack data. Few-shot learning (FSL) has emerged as a solution, allowing systems to learn from a limited number of samples. A systematic review of FSL approaches for NIDS published between 2022 and 2026 was conducted, identifying 21 relevant studies from an initial pool of 1,358. The review categorized the FSL techniques, datasets, and experimental parameters used, revealing that meta-learning and convolutional neural networks are the most common methods. The CIC-IDS2017 and CSE-CIC-IDS2018 datasets were frequently utilized, but the lack of standardized evaluation practices hampers reproducibility and comparison across studies. The findings indicate a pressing need for consensus on FSL methodologies to enhance the effectiveness of NIDS against evolving threats.