How dwday/deep_learn_ids works
Main components
• Data Input / Preprocessing Module (even if not explicitly separated, assume that there is some form of data ingestion and preprocessing inside ids_detect.py).
• Model Definition / Architecture Component:
• That the deep learning model is defined in the “models” directory (specifically in models/models_ddos.py).
• That it uses Dense, Conv1D, and LSTM layers.
• Training and Evaluation Module:
• Main script “ids_detect.py” which orchestrates the training process.
• Evaluation outputs such as accuracy plots, training progress graphs, and a confusion matrix, which are stored in the “images” directory.
• (Optional) External Libraries and Frameworks:
• Use of Keras/TensorFlow as the underlying framework for model training.
Relationships and Interactions
• Data flows from an input source (network data or provided dataset) into the preprocessing stage.
• The preprocessed data is passed into the deep learning model (whose architecture is defined in models/models_ddos.py).
• The training process (conducted in ids_detect.py) utilizes these model definitions to learn and evaluate.
• Evaluation results are-generated and saved as images (accuracy plots, confusion matrices) in the “images” directory.
Architectural Patterns and Technologies
• The design pattern is essentially a pipeline for deep learning model training and evaluation.
• Separation of model definitions (in the “models” folder) from the execution logic (in ids_detect.py).
• If available, mention any specific data preprocessing or evaluation routines as part of the pipeline (even if they are implicit within the id_detect.py file).