TrainingExample
data class TrainingExample(val input: FloatArray, val expectedOutput: FloatArray, val timestamp: Long)
Training example data structure for neural network learning.
Encapsulates a single training instance with input features, expected output values, and temporal metadata for analysis.
Components:
input: Feature vector inputSize representing interaction contextexpectedOutput: Target values outputSize for supervised learningtimestamp: Unix epoch timestamp for temporal analysis and cleanup
Usage Patterns:
Created during trainStep from InteractionData
Stored in trainingHistory for analysis and replay
Used during backpropagation for weight updates
Parameters
input
Feature vector extracted from user interaction data
expectedOutput
Target prediction values for supervised learning
timestamp
Creation time for temporal analysis and data lifecycle management