Codebilibili/Traffic-Classification

Introduction

这个仓库关联论文：CENTIME: A Direct Comprehensive Traffic Features Extraction for Encrypted Traffic Classification

还有其他两篇关于流量检测的论文：

• An Explainable Machine Learning Framework for Intrusion Detection Systems
• An Encrypted Traffic Classification Framework Based on Convolutional Neural Networks and Stacked Autoencoders

如果觉得有帮助，欢迎引用上述论文。

Data Preprocessing

Run the following command in the root directory to perform data preprocessing:

python -m TrafficFlowClassification preprocess_pipeline

Model Train

Select the model in 'train.py' file in line 40,

model = resnet181D(model_path, pretrained=cfg.test.pretrained, num_classes=12, image_width=cfg.train.IMAGE_WIDTH).to(device)

Then run the following command in the root directory to train the model.

python -m TrafficFlowClassification train_pipeline