Alfie sequence diagram automl

alfie_seq.md

sequenceDiagram

participant User as Intent Recognition

participant Controller

participant AutoML as AutoML Engine

participant Tabular as Tabular Engine

participant Vision as Vision Engine

participant AutoMLPlus as AutoML+ Engine

participant ModelStore as Model Store

participant DatasetStore as Dataset Store

participant SessionStore as Session Store

participant XAI as XAI Layer

participant UI as User Interface

participant SKG as Semantic Knowledge Graph

participant FairEval as Fairness Evaluation

  

User->>Controller: Extract task type & user input

Controller->>AutoML: Request requirements

AutoML-->>Controller: Return requirements JSON

Controller->>AutoML: Send config, data, task info

  

AutoML->>Tabular: For tabular data

AutoML->>Vision: For vision tasks

AutoML->>AutoMLPlus: For general/multi-modal tasks

  

Tabular->>Tabular: Uses hyperparameter search

Tabular->>Tabular: Uses model ensembles

  

Vision->>Vision: Uses transfer learning

Vision->>Vision: Avoids neural architecture search (cost)

AutoMLPlus->>Ensuring Unbiased AI in Autonomous Vehicles: Used by Ensuring Unbiased AI in Autonomous Vehicles

Vision->>Ensuring Unbiased AI in Autonomous Vehicles: Used by Ensuring Unbiased AI in Autonomous Vehicles

Vision->>Compliance Screener: Used by Compliance Screener

AutoMLPlus->>Website Accessibility Checker: Used by Website Accessibility Checker

  

AutoML->>ModelStore: Save model

AutoML->>DatasetStore: Save data

AutoML->>SessionStore: Save session info

  

ModelStore->>ModelStore: Store handler & ethics metadata

DatasetStore->>DatasetStore: Store splits & metadata

  

AutoML->>XAI: Trigger XAI Layer

XAI->>AutoML: Receive model, task, test data

XAI->>UI: Generate SHAP/GradCAM/Fairness metrics

UI->>User: Provide evaluation results

UI->>EndUser: Report transparency & trust

  

AutoML->>SKG: Query task context

SKG-->>AutoML: Return insights

SKG->>XAI: Feed insights

SKG->>FairEval: Inform fairness evaluation

FairEval->>AutoML: Send feedback

Loading