waderianda/nutrition_VAE-RNN

Nutrition Recommendation Model

A VAE-RNN model for generating personalized nutrition recommendations based on user profiles and activity schedules.

Model Description

This model combines a Variational Autoencoder (VAE) with a Recurrent Neural Network (RNN) to generate personalized meal plans. The VAE encodes user profile information, while the RNN processes activity schedules to produce context-aware nutrition recommendations.

Usage

from inference import load_model, generate_recommendation

# Example user profile (biometric and health features)
user_profile = [...]  # 26-dimensional vector

# Example activity schedule (7 days, 10 features per day)
activity_schedule = [...]  # Shape: [7, 10]

# Generate recommendations
meal_plan = generate_recommendation(user_profile, activity_schedule)

Model Architecture

• VAE Component: Encodes user profile data into a latent space
• RNN Component: Processes activity schedules using GRU cells
• Decoder: Generates meal plans based on combined VAE and RNN outputs

Training Data

This model was trained on nutrition data including:

• User profiles (age, weight, height, health conditions)
• Activity schedules (exercise patterns, daily activities)
• Food logs (nutrient information, meal timing)