fitted(to:eventHandler:)
Fits a time series classifier model to a sequence of examples.
Declaration
func fitted(to input: some Sequence<AnnotatedFeature<MLShapedArray<Scalar>, Label>>, eventHandler: EventHandler? = nil) async throws -> TimeSeriesClassifier<Scalar, Label>.ModelParameters
- input:
A sequence of annotated features for training. Each feature’s shape should be
[sequenceLength, featureSize]and each annotation should be one-hot encoded with shape[labelCount]. - eventHandler:
An event handler.
Return Value
The fitted time series classifier model.
Discussion
The training process partitions the input into random batches according to the batch size configuration parameter. Training stops when the validation loss stops improving or when the maximum number of iterations is reached.