Biological Neural Network Models TRIZ Cases
- Neural Graph Receipt Decoding for Wrinkled Paper Alignment
Graph neural networks generate adjacency matrices from OCR vertex features to reconstruct text lines on wrinkled receipts with large data gaps.
- Base Station Autoconfiguration Using Neural Network Prediction
An autoconfiguration system uses machine learning to predict parameters from network planning data, eliminating manual entry errors and reducing setup time.
- In-store Consumer Behavior Metadata Aggregation and Verification
Aggregates in-store consumer behavior event metadata from electronic devices for AI analysis.
- Hybrid Power Plant Predictive Model Self-Tuning
A hybrid predictive model combines physics-based and data-driven approaches to forecast power plant performance.
- Plant Protein Modeling for Food Texture and Solubility
Algorithms model protein interactions to optimize texture and taste while preventing undesirable organoleptic properties in food products.
- Recommendation Engine Using Dynamic Neural Network Link Matrices
A recommendation engine builds a neural network of interrelationships among venues, reviewers, and users based on attributes and reviews.
- Neural Network Speech Recognition Using Latent Variable Filtering
A neural network processes feature vectors and latent variables to isolate speech components from distracting audio characteristics.