Traffic Embedding: Improving the Transferability of Deep Learning Model in Anomaly Detection
Bibliographic record
Abstract
This paper explores the use of deep learning techniques in enhancing network anomaly detection systems (ADS) with emphasis on the transferability of model across datasets. Although previous research has shown that machine learning and deep learning models can effectively detect various types of attacks with high accuracy, these approaches typically require dataset-specific implementations. In contrast, this paper presents a novel framework that transcends such limitations by eliminating the need for complex feature engineering and model tuning for individual datasets. Namely, the same proposed model can be effectively deployed across different network environments without modification. The proposed model integrates Convolutional Neural Networks (CNNs) and self-attention mechanisms to process and classify network traffic data derived from packet captures (PCAP) files. CNNs excel at identifying local patterns in sequential data, enabling the model to extract byte-level features from data and generating comprehensive packet representations. The self-attention mechanism then identifies correlations among embedded packet features, highlighting critical patterns that indicate potential anomalies. To further enhance robustness of the proposed model, this paper incorporates the Variational Autoencoder (VAE) structure in the training phase. The experiment results demonstrate that the model has strong generalization capabilities, achieving $94 \%$ precision in test data while requiring only $15 \%$ of the available training samples.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.002 | 0.006 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.001 | 0.001 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.001 | 0.002 |
| Open science | 0.001 | 0.001 |
| Research integrity | 0.001 | 0.002 |
| Insufficient payload (model declined to judge) | 0.001 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".