cyber-attack-detector

MVP project for cyber attack detection based on transaction data and machine learning models.

Project goal

The project is aimed at building a minimal working prototype for detecting suspicious activity and cyber attacks in transaction flows.

Project structure

• src/data - data generation and dataset preparation scripts
• src/models - model training and validation scripts
• src/inference - inference scripts for prediction
• data - input and generated data
• artifacts - processed datasets and intermediate artifacts
• models - trained model files
• tests - basic tests
• docs - project documentation

Installation

Create and activate a virtual environment, then install dependencies:

python -m venv .venv
.venv\Scripts\activate
pip install -r requirements.txt

Pipeline

Run the following steps in order:

python src/data/generate_sbp_transactions.py
python src/data/prepare_dataset.py
python src/models/train_binary.py
python src/models/train_multiclass.py
python src/models/check_models.py

Inference

A minimal inference example is available in:

python src/inference/predict.py

MVP status

Current MVP includes:

• synthetic transaction generation
• dataset preparation
• binary attack detection model
• multiclass attack classification model
• basic prediction flow

Limitations

This version is a minimal prototype and does not yet include a production API, dashboard, or deployment pipeline.