Classifier
Block Catalogue / Machine Learning
Multi-class gesture classifier with two training modes: batch retraining from a Trigger Buffer, or per-sample incremental updates driven live by a Trigger.
Input and output
| Property | Value |
|---|---|
| Type | classifier, classifierblock |
| Inputs | 2; see usage below |
| Consumes | Vector<double> (Count == InputDim) on the feature edge; int segment count from a TriggerBuffer; Vector<double> (argmax = class) or null from a Trigger |
| Publishes | Vector<double> of class probabilities, length NumClasses |
Parameters
| # | Name | Type | Default | Meaning |
|---|---|---|---|---|
| 0 | modelType | string enum | required | Softmax | SoftmaxRFF | LDA | KNN | RandomForest | LinearSVM | Threshold; unknown throws |
| 1 | inputDim | int | required | Feature vector length |
| 2 | numClasses | int | required | Initial class count |
| 3 | trainingMode | string | Buffer |
Buffer or Incremental; unparseable falls back to Buffer silently |
Params null or Count < 3 throws. Indices 4+ are read by nothing.
Requirements and use
Choose batch or incremental training and provide matching labels. Capture/train before interpreting predicted classes; use held-out observations to assess the result.
For parameter positions and connection rules, see Pipeline Configuration. For capturing output, see Recording.
Example
Train a simple classifier provides complete JSON for a one-feature, two-class KNN example. Buffer mode retrains automatically when a completed segment count increases; Force Retrain fits the stored data again. Reopening JSON does not restore the fitted classifier or training segments.
Implementation
API reference. Model defaults above describe configuration loading; palette hints may expose a subset or select different defaults.