📚 📥 Download Documentation

Access the complete research memo in PDF format

🖥️ Digital Version Optimized for screen reading

🖨️ Print Version Formatted for physical printing

Binary classification and prediction system that implements a voting mechanism among multiple machine learning models. The client-server architecture allows processing of temporally and multidimensionally structured data. Each model is implemented as an independent microservice, and is responsible for its own data preprocessing and prediction logic.

The system is divided into modular components:

• Orchestrates requests to the various classification models
• Implements voting mechanisms (initially by majority)
• Provides a visual interface for configuration and visualization
• Manages timeouts and model availability
• Asynchronously handles model responses

• Independent APIs implementing six algorithms:
  • SVM
  • LSTM
  • XGBoost
  • One-Class SVM
  • Isolation Forest
  • Custom CNN
• Each model performs its own data preprocessing
• Returns predictions and confidence levels
• Implements healthcheck to verify availability