Natural Language Analysis TRIZ Cases
- Machine Learning Impact Analysis for Software Release Quality
Automated impact analysis replaces manual testing by predicting affected UI links via machine learning, reducing upgrade time while maintaining reliability.
- Wavelet Feature Compression for Distributed Speech Recognition Bandwidth
Discrete wavelet transform filters speech features into frequency components, transmitting only low-frequency data to reduce bandwidth.
- Logarithmic Quantization Training for Neural Network Accuracy
Gradient descent optimizes logarithmic thresholds during training to preserve accuracy while reducing hardware complexity and memory footprint.
- Electronic Book Supplemental Content Generation via Noun Context Matching
A system generates supplemental content for electronic books by identifying nouns and related words in displayed text to create relevant multimedia segments.
- Intention Statement Visualization for Consumer Intent Analysis
Automated modules extract and visualize consumer intentions to reduce manual analysis time.
- Multi-die Dot-Product Engine for Large-Scale ML Inference
Multi-die dot-product engine distributes neural network weights across multiple chips to support large-scale machine learning inference.
- Information Extraction Using Entity-Attribute Graph Correlation
An information extraction method uses an entity-attribute graph to co-extract entities and attributes based on correlation coefficients.