Mathematical Models TRIZ Cases
- Unsupervised Keyphrase Extraction via Biased Random Walks
A graph-based system extracts keyphrases using biased random walks guided by contextual themes and named entity bias weights.
- Forecasting Traffic Signal Switching Behavior Using Interpretable Machine Learning
A hierarchical state machine forecasts traffic signal switching behavior using interpretable machine learning models.
- Multitask Machine Learning Framework With Adaptive Parameter Sharing
A multitask machine learning framework jointly trains models using a similarity matrix to share parameters across related tasks.
- Learning Device Using Neural ODE Gradient Model for Control Policy
A learning device employs a neural ordinary differential equation to model state transitions and temporal changes in control targets.
- Support Case Prediction Model Using Telemetry Data
A prediction model generates next steps for support cases using client telemetry and contextual information.
- Semi-supervised Anomaly Detection Framework for Purpose-Oriented Filtering
A semi-supervised framework integrates manual grading to train a supervised classification model on top of an unsupervised anomaly detector.
- Domain Adaptation for Synthetic Perception Model Training
A discriminator learns domain-invariant representations to minimize divergence between simulated and real sensor data distributions.
- Alimentary Provisioning System Using Biological Data Filtering
A computing device generates alimentary instruction sets from biological extractions to select beneficial ingredient combinations.
- AI Model Selection via Misclassification Probability Analysis
An AI model selection method identifies optimal models using misclassification probabilities derived from additional training on difficult data.
- Cross-Geographical Predictive Data Analysis Using Simulated Inputs
Simulated predictive input data generation replaces raw inferior domain transmissions, reducing network bandwidth demands while maintaining prediction accuracy.
- Touch Pad Segmentation for Handheld Display Selection
A touch pad maps active screen regions to separate input areas, enabling precise element selection without finger contact.
- Snowfall Probability Distribution Forecasting System
A rules-based forecasting system generates snowfall probability distributions from ensemble weather data.
- MOS Sensor Array Digital Signature Database for Rapid Olfactory Analysis
A structured MOS sensor array database organizes multi-dimensional smell data into standardized digital signatures for efficient computer processing.
- Wrist-Worn Accelerometer Gait Detection Using Naive Bayes
A wrist-worn tri-axial accelerometer detects gaits using a naive Bayes classifier, reducing power consumption by discarding gyroscopes.
- Database Reconfiguration for Time-Varying Workloads
A reconfiguration system forecasts workload measurements to generate optimized configuration parameter sets for database nodes.
- Anomaly Detection Using Local Density and Frequency Measurements
Probabilistic classifier combines local density measurements for numeric attributes with frequency distributions for categorical data to identify anomalies.
- Optimization Apparatus Escaping Local Solutions via Stored Energy Data
An optimization apparatus stores rejected state energy differences to select new states and accelerate search convergence.
- Machine Learning Step-Size Adaptation via Stochastic Meta-Descent
A machine learning architecture adapts step-sizes using stochastic meta-descent for per-feature tuning.
- Data Boundary Deriving System for AI Anomaly Detection
A data boundary deriving system generates labeled learning data using probability density functions to support machine learning models.
- Certifiable Out-of-Distribution Generalization for Deep Neural Networks
Kernelized linear regression and stochastic perturbation learning derive a classifier that provides certified precision for out-of-distribution generalization.