# devchi1l/poison-resilient-continual-nids architecture

> Architecture diagram of the GitHub repository devchi1l/poison-resilient-continual-nids, generated by GitDiagram from its file tree, README and a sample of source files (last updated 2026-10-06). It is an AI-made overview: check details against the source.

- Interactive diagram: https://gitdiagram.com/devchi1l/poison-resilient-continual-nids
- Repository: https://github.com/devchi1l/poison-resilient-continual-nids

## Overview

This repository is a research prototype plan for a poison-resilient, novelty-aware continual-learning NIDS, not an implemented or runnable system. Its documented target workflow takes flow-level CSV data through preprocessing, Transformer training and classification, confidence-based UNKNOWN detection, Task 1/Task 2 evaluation, and a clean-versus-label-flip-poisoned comparison, then records metrics. The README also describes replay support and reproducibility metadata. These components are documented as planned ownership areas; the tree contains no implementation files, so their paths are null. Edges reflect README-documented interfaces and intended flow, not observed runtime calls.

## Diagram (Mermaid)

```mermaid
flowchart TD

subgraph group_data["Data preparation"]
  node_preprocessing["Preprocessing"]
  node_task_splits["Task splits"]
end

subgraph group_learning["Model and continual learning"]
  node_transformer["Tabular Transformer"]
  node_training["Training and checkpoints"]
  node_continual["Continual update"]
  node_replay["Replay support"]
  node_inference["Classification"]
end

subgraph group_robustness["Novelty and evaluation"]
  node_novelty["UNKNOWN detection"]
  node_poisoning["Label-flip experiment"]
  node_evaluation["Metrics and comparison"]
  node_manifest["Result manifest"]
end

node_researcher(("Researcher"))
node_flow_csv["Flow CSV"]

node_researcher -->|"supplies data"| node_flow_csv
node_flow_csv -->|"loads and preprocesses"| node_preprocessing
node_preprocessing -->|"provides features"| node_task_splits
node_task_splits -->|"provides training data"| node_training
node_training -->|"trains"| node_transformer
node_task_splits -->|"forms tasks"| node_continual
node_continual -->|"updates model"| node_training
node_continual -->|"supports replay"| node_replay
node_transformer -->|"predicts classes"| node_inference
node_inference -->|"provides confidence"| node_novelty
node_task_splits -->|"provides labels"| node_poisoning
node_poisoning -->|"supplies changed labels"| node_training
node_inference -->|"provides predictions"| node_evaluation
node_novelty -->|"provides unknown mask"| node_evaluation
node_continual -->|"reports task performance"| node_evaluation
node_evaluation -->|"records metrics"| node_manifest
node_researcher -->|"reviews results"| node_manifest

classDef toneNeutral fill:#f8fafc,stroke:#334155,stroke-width:1.5px,color:#0f172a
classDef toneBlue fill:#dbeafe,stroke:#2563eb,stroke-width:1.5px,color:#172554
classDef toneAmber fill:#fef3c7,stroke:#d97706,stroke-width:1.5px,color:#78350f
classDef toneMint fill:#dcfce7,stroke:#16a34a,stroke-width:1.5px,color:#14532d
classDef toneRose fill:#ffe4e6,stroke:#e11d48,stroke-width:1.5px,color:#881337
classDef toneIndigo fill:#e0e7ff,stroke:#4f46e5,stroke-width:1.5px,color:#312e81
classDef toneTeal fill:#ccfbf1,stroke:#0f766e,stroke-width:1.5px,color:#134e4a
class node_preprocessing,node_task_splits toneBlue
class node_transformer,node_training,node_continual,node_replay,node_inference toneAmber
class node_novelty,node_poisoning,node_evaluation,node_manifest toneMint
class node_researcher,node_flow_csv toneIndigo
```

## Components

### Data preparation

- **Preprocessing**
- **Task splits**

### Model and continual learning

- **Tabular Transformer**
- **Training and checkpoints**
- **Continual update**
- **Replay support**
- **Classification**

### Novelty and evaluation

- **UNKNOWN detection**
- **Label-flip experiment**
- **Metrics and comparison**
- **Result manifest**

### Other components

- **Researcher**
- **Flow CSV**

## Connections

- Researcher → Flow CSV: supplies data (evidence: `README.md`)
- Flow CSV → Preprocessing: loads and preprocesses (evidence: `README.md`)
- Preprocessing → Task splits: provides features (evidence: `README.md`)
- Task splits → Training and checkpoints: provides training data (evidence: `README.md`)
- Training and checkpoints → Tabular Transformer: trains (evidence: `README.md`)
- Task splits → Continual update: forms tasks (evidence: `README.md`)
- Continual update → Training and checkpoints: updates model (evidence: `README.md`)
- Continual update → Replay support: supports replay (evidence: `README.md`)
- Tabular Transformer → Classification: predicts classes (evidence: `README.md`)
- Classification → UNKNOWN detection: provides confidence (evidence: `README.md`)
- Task splits → Label-flip experiment: provides labels (evidence: `README.md`)
- Label-flip experiment → Training and checkpoints: supplies changed labels (evidence: `README.md`)
- Classification → Metrics and comparison: provides predictions (evidence: `README.md`)
- UNKNOWN detection → Metrics and comparison: provides unknown mask (evidence: `README.md`)
- Continual update → Metrics and comparison: reports task performance (evidence: `README.md`)
- Metrics and comparison → Result manifest: records metrics (evidence: `README.md`)
- Researcher → Result manifest: reviews results (evidence: `README.md`)
