Kernel Methods TRIZ Cases
- Network Controllability Analysis for Misinformation Attack Detection
A network controllability analysis system computes metrics to identify compromised nodes in mobile wireless networks.
- Cross Descriptor Learning for Unstructured Data Labeling
A cross descriptor learning system generates labels for unlabeled exemplars using multiple descriptors to enable automatic metadata enrichment.
- Convolutional Neural Network Incremental Learning via Kernel Similarity
A convolutional neural network updates weights by grouping similar kernels to adapt to new data distributions.
- Speech Interaction Apparatus Dynamic Wait Time Management
An interaction apparatus measures elapsed time and subtracts it based on user utterance context to control additional actions.
- Physical Layer Authentication via SVM Channel Feature Vectors
SVM classifies wireless channel feature vectors to authenticate terminals at the physical layer.
- Classifier-Based Process Variable Measurement Selection
AI classifier selects optimal measurement method for process variables, resolving accuracy-reliability trade-offs across varying media conditions.
- Nonvolatile Memristor Temporal Kernel Device With Adjustable RC Circuit
A nonvolatile memristor temporal kernel device integrates a parallel resistor and capacitor to enable adjustable time constants.
- Metadata-Driven Feature Engineering for Log Data Processing
A metadata-driven feature engineering system generates unit operations to process log data into vectors, eliminating separate vector creation programs.
- File Event Correlation for Intrusion Alert Remediation
A system correlates file events with intrusion detection alerts to generate event-correlation data.
- Unsupervised ML Anomaly Detection for Network Security
Machine learning models identify outlier processes in computer networks using unsupervised training techniques.
- Time Series Data Imputation Using Extremeness Scores
A data imputation system generates substitute values for missing elements in sparse time series datasets.
- Data Privacy Enforcer Automates Personal Data Removal
A data privacy enforcer system identifies user applications holding personal information and directs individuals to specific privacy settings.
- Neural Network Generation Using Predictive Classifier for Faster Convergence
A trained classifier predicts neural network reliability to adjust genetic algorithm mutation parameters, accelerating convergence on high-performing models.
- Context-Aware Recommendation System for Analytics
A machine learning recommendation system predicts user intent to deliver relevant analytics tools and data without manual queries.
- Neural Network Activation Subtraction for Occluded Object Classification
An activation region subtractor removes foreign object neuron activations within a convolutional neural network forward pass.
- Local Feature Embedding for Fine-Grained Visual Pattern Recognition
Local feature embedding captures fine-grained visual details via max pooling, resolving the trade-off between extraction precision and computational overhead.