Biological Models TRIZ Cases
- Autosave Trigger Configuration Using Machine Learning Value Metrics
A machine learning model generates autosave value triggers based on data attributes and environmental conditions.
- Phrase-Based End-to-End Text-to-Speech Synthesis
Parallel phrase generation overcomes autoregressive speed limits by processing segments simultaneously with context embeddings to maintain style consistency.
- Dynamic NPC Action Generation from Player History
Information processing device generates non-player character actions from player interaction history.
- Machine Learning Model Selection via Data Group Distance
A data predicting apparatus selects machine learning models by measuring distances between input data and predefined groups.
- Unified Transformer for Variable-Length Satellite Imagery
A unified transformer processes variable-length satellite imagery time series to generate mid-cycle agricultural inferences.
- GNSS Residual Grid Localization via Graph Neural Networks
Residual grids aggregate pseudorange measurements to estimate location, reducing positioning time in urban environments with non-line-of-sight signals.
- ML-Based Network Service Attack Mitigation
Machine learning models analyze operational data to identify negative factors and disconnect suspected connections from network services.
- Anomaly Detection in 5G Network Slices via Modular Validation
Modular validation separates attack and failure anomalies using common measurement data to enhance network security without increasing device complexity.
- GAN Discriminator Learning via Vector Space Optimization
A GAN learning system acquires discrimination images and calculates feature vectors to automate training.
- Personalized Feature Vectors for Search Relevance and Conversion
Merging content item features with user activity data resolves the contradiction between search result relevance and conversion rate.
- Tropical Geometry Interpretable Machine Learning Model
Tropical geometry transforms complex parameters into interpretable structures, resolving the trade-off between high accuracy and model transparency.
- Hierarchical LSTM Audio Classification with Transient Detection
Hierarchical audio classification system using parallel transient detectors to reset LSTM states at class transitions.
- AI-Based ATM Servicing System for Error Code Analysis
Machine learning models analyze ATM error codes and service histories to determine servicing actions.
- Prediction Rationale Analysis Apparatus for Model Discrepancy Detection
A prediction rationale analysis apparatus identifies common properties across multiple models to derive unified rationales.
- Exemplar Vector Selection for ML Anomaly Detection
Inferential exemplar selection system balances prognostic accuracy against compute cost by automatically determining the quantity of exemplar vectors.
- Emotional Speech Synthesis Using Face Feature Extraction
A speech processing apparatus generates emotional voice output using extracted face feature points from video frames.
- Data Augmentation Using Linear Interpolation and Random Numbers
A data augmentation method combines input and output samples using linear interpolation and random numbers to generate extended training datasets.
- Data Attributes Recommendation Engine for Analytics Platform Optimization
A machine learning model selects relevant data attributes to optimize analytics platform resource utilization and reduce redundant storage.
- SATCOM Data Compression System for Aircraft Bandwidth Efficiency
A data compression system segments web page files into lossless and lossy categories to achieve a two-fold size reduction.
- CNN Omics Data Mapping via Spatial Ordering
A mapping apparatus transforms omics input data into N-dimensional spatial representations for convolutional neural network processing.
- Differential Fuzzing for LLM Code Verification
Differential fuzzing compares LLM-generated executables against reference binaries using identical inputs to detect behavioral inconsistencies.
- AI Virtual Assistant LLM Streaming for Real-Time Video Responses
An AI virtual assistant streams synthetic host video responses generated from large language model processing of user audio input.