Each engine is optimized for a specific signal type and use case. Pick the right one for your application, or let AutoML choose for you.
Battle-tested, general-purpose engines. Start here.
Advanced engines optimized for specific signal types. Higher accuracy on the right problem.
Vivi KAN 4 (highest accuracy), Vivi Wave 4 (oscillatory signals), or Vivi Focus 4 (fastest inference). All competitive with CNN on CWRU bearing dataset.
Vivi 4 (most stable), Vivi 3 (well-tested), or Vivi Cerebellum 4 (predictive). One-class learning: train on normal data only, detect deviations automatically.
Vivi Spike 4 (19KB Flash, event-driven) or Vivi Hyperbolic 4 (10KB Flash). Fit on the smallest MCUs for battery-powered deployment.
Upload your data — AutoML tests all engines, picks the best one for your specific problem and hardware constraints. Automatic architecture search.
Every engine is benchmarked on industry-standard datasets with documented methodology.