FlowTransformer for Multi-Classification Task

About

This is a FlowTransformer-based framework with the following additional features:

1. Display the detection performance for each class in the evaluation set.
2. Support training and testing on multi-classification tasks.

Before diving in, please check out the original project for an overview of how the framework operates.

Reference Paper Information

• Title : FlowTransformer: A Transformer Framework for Flow-based Network Intrusion Detection Systems
• Link to Paper : FlowTransformer Paper

Use

• For binary classification : binary_classification_demo.ipynb
• For multi-classification : multi_classification_demo.ipynb

Directory Hierarchy

The list below highlights the modifications made compared to the original project :

|—— binary_classification_demo.ipynb  [updated]
|—— multi_classification_demo.ipynb  [updated]
|—— framework
|    |—— __init__.py
|    |—— base_classification_head.py
|    |—— base_input_encoding.py
|    |—— base_preprocessing.py
|    |—— base_sequential.py
|    |—— dataset_specification.py  [modified]
|    |—— enumerations.py
|    |—— flow_transformer_binary_classification.py  [updated]
|    |—— flow_transformer_multi_classification.py  [updated]
|    |—— flow_transformer_parameters.py
|    |—— framework_component.py
|    |—— model_input_specification.py  [modified]
|    |—— sequential_input_encoding.py
|    |—— utilities.py
|—— implementations
|    |—— __init__.py
|    |—— classification_heads.py
|    |—— input_encodings.py
|    |—— pre_processings.py
|    |—— transformers
|        |—— basic
|            |—— decoder_block.py
|            |—— encoder_block.py
|        |—— basic_transformers.py  [modified]
|        |—— named_transformers.py  [modified]
|—— main.py
|—— .gitignore

License

This project is based on FlowTransformer and licensed under the AGPL-3.0 License.