BRAIN.JS AI POC · USER GUIDE

Brain.js Optical Air-Defense Demonstration

This demonstration shows how a neural network is initialized, trained using synthetic images, and used to classify previously unseen targets.

Demonstration Workflow

9 STEPS

  1. Initialize the Network

    Click Initialize to create the Brain.js neural network and prepare the training environment.

  2. Train the Network

    Click Train to begin neural-network training.

  3. Wait for Training to Complete

    Training progress and network activity are displayed. Do not select targets until training finishes and the application indicates READY / TRAINED.

  4. Select a Target

    Use the < and > Target Palette controls to browse targets. Click a thumbnail to classify that previously unseen image.

  5. Review the Neural-Network Result

    Review the predicted target class and the neural network's confidence/output values.

  6. Review the Engagement Decision

    DRONE and AIRCRAFT may produce ENGAGE at 75% confidence or higher. BIRD, KITE, and BALLOON produce DO NOT ENGAGE.

  7. IFF Safety Inhibit

    A target identified as FRIEND always produces DO NOT ENGAGE, regardless of neural-network classification or confidence.

  8. Engagement Demonstration

    A qualifying DRONE engagement plays the CIWS demonstration. A qualifying AIRCRAFT engagement plays the Patriot demonstration.

  9. Reset / Retrain

    Reset clears the trained network and demonstration state. Initialize and train again before classifying targets.

Continue to Demonstration