BinaryCrossEntropyLossFitnessCalculator<T, TInput, TOutput>
Models & Types · AiDotNet.FitnessCalculators
A fitness calculator that uses Binary Cross-Entropy Loss to evaluate model performance for binary classification problems.
Inheritance: FitnessCalculatorBase<T, TInput, TOutput> → BinaryCrossEntropyLossFitnessCalculator<T, TInput, TOutput>
For Beginners
This calculator helps you evaluate how well your model is performing on binary classification tasks (problems where you’re predicting one of two possible outcomes, like “yes/no” or “spam/not spam”).
Binary Cross-Entropy Loss measures the difference between your model’s predicted probabilities and the actual outcomes. Here’s how to understand it:
- It’s specifically designed for binary classification problems (where the answer is one of two possibilities)
- Lower values are better (0 would be a perfect model)
- It heavily penalizes confident but wrong predictions (e.g., if your model is 99% sure something is spam when it’s not)
- It’s commonly used in neural networks and logistic regression
Think of it like a scoring system that gives your model a harsh penalty when it’s very confident but wrong, and a small penalty when it’s uncertain. This encourages your model to be confident only when it has good reason to be.
Unlike accuracy (which just counts right vs. wrong), Binary Cross-Entropy Loss takes into account how confident your model was in its predictions.
Constructors
| Constructor | Summary |
|---|---|
BinaryCrossEntropyLossFitnessCalculator(DataSetType dataSetType) |
Initializes a new instance of the BinaryCrossEntropyLossFitnessCalculator class. |