TY  - JOUR
T1  - Fine-Tuning Anomaly Classifier for Unbalanced Network Data
AU  - Joe, Mugon 
AU  - Kim, Miru 
AU  - Kwon, Minhae 
JO  - The Journal of Korean Institute of Communications and Information Sciences
PY  - 2024
DA  - 2024/1/1
DO  - 10.7840/kics.2024.49.7.911
KW  - Anomaly Detection
KW  - Anomaly Classifier
KW  - Autoencoder
KW  - Fine-Tuning
AB  - With the recent surge in network intrusion attempts, there is a growing emphasis on prompt and accurate responses. Given that each intrusion type necessitates a distinct response approach, accurately identifying network anomalies is crucial for an effective response. Research on classifying network anomalies using classification models with autoencoder-based anomaly detection has garnered significant attention. However, network data presents an unbalance problem for abnormal data, which is challenging to collect, leading to limited performance. This limitation arises from the disparity in distribution between the detected data of autoencoder-based anomaly detection and the training data of classification models. To address this issue, this study proposes a solution that employs fine-tuning techniques with the classification model. Simulation results demonstrate that the proposed system surpasses previous research results across all evaluation metrics.