Genetic Models TRIZ Cases
- ML Architecture Search Framework for Hardware-Aware Multi-Metric Optimization
Automatically searches ML architectures across hardware constraints and performance metrics to cut design time and compute use.
- Multi-Task Neural Network Training With Frozen Shared Blocks
Frozen shared blocks and task-specific updates let one neural network learn multiple tasks while limiting forgetting, overhead, and latency.
- FSM Stateflow Optimization for Correct and Usable Behavior
Iterative scoring with domain rules, genetic search, and local search improves FSM stateflow correctness, usability, and expected behavior.