Relational Databases TRIZ Cases
- Multi-Modal Data Labeling Through Inter- and Intra-Modal Transformation
Inter-modal and intra-modal transformations combine estimated labels to reduce manual work when labeling multi-modal training data.
- Data Package Header Validation for Fast Threat Detection
Header classifiers trace each data package’s transmission chain to flag unexpected intermediaries before deeper security analysis.
- Balanced Cluster Division for Computing-in-Memory Vector Search
Unequal k-means clusters limit bit-line use; overlapping data division matches CIM capacity and reduces CPU transfers during similarity search.
- Low-Level Mobile Activity Detection for MVNO Bandwidth Forecasting
AI compares low-level mobile logs to infer activities, helping MVNOs forecast bandwidth with limited MNO data.
- Machine Learning Conversation Platform with State-Based Emotion Recognition
Chat programs often miss emotions and prior events; machine learning enriches responses and routes them through state transitions for personalized interaction.
- Machine Learning Listing Segmentation for Scalable Resource Allocation
K-means clustering segments listing availability patterns to target resources and communications while scaling marketplace operations.
- Model-Based Cohort Clustering Visualization for Diverse Data
Cluster visualizations help choose cohort counts and compare results while logarithmic normalization and category weighting handle diverse, long-tailed data.
- Federated Data Retrieval for Decentralized Generative AI Pipelines
A federated manager routes queries to decentralized structured and unstructured sources, helping LLMs use proprietary data without retraining.
- Infoscenting Fields for Guided Multi-Fact Data Visualization
Learn how an object-model interface uses visual feedback and field guidance to clarify relevant selections across multi-fact data sources.
- Unstructured Data Classification With Key-Datum Validation
Rule-based email processing struggles with varied content; classifiers, extractors, and validation models automate accurate downstream outputs.
- Centralized Metadata Repository with Relevancy-Scoped Retrieval
Relevancy identifiers let a central repository return only each client's needed metadata, reducing retrieval overhead and error risks.
- Data Investigation Knowledge Graphs for Reproducible Analysis
Automated queries build a knowledge graph from multiple data sources, preserving investigation steps for accurate review and reuse.
- Knowledge Graph Temporal Queries with Confidence-Tracked Relationships
Confidence points and temporal metadata help knowledge graphs answer queries when relationship boundaries are undefined, reducing manual upkeep.
- Data File Clustering with KD-Classifier Trees for Fewer Rewrites
Uniformly distributed updates can trigger broad file rewrites; a KD-classifier tree routes records by key-values to limit affected files.
- Hybrid Table Secondary Indexes for Concurrent OLTP-OLAP Workloads
Separate transactional and analytical table paths coordinate secondary indexes to preserve integrity during concurrent OLTP and OLAP work.
- Wearable Biological Extraction Classification for Guided Recommendations
Classifying wearable biological extraction and activity data against user fingerprints helps display compatible guided recommendations in a clearer GUI.
- On-Device Transaction Tags for Private Record Search
Pre-generated on-device tags and indexes improve transaction searches, reduce bandwidth use, and keep sensitive data local.