Namespace NeuralNetworks.Core
Classes
- ActivationRoutines
Provides activation routines for neural network computations.
- ArrayFactoryRoutines
Provides factory and initialization routines for creating float arrays with range-based values, uniformly distributed random values, and normally distributed random values.
- ArrayRoutines
Provides extension methods for comparing the shape of single- and multi-dimensional float arrays.
- ArraySlicingRoutines
Provides array slicing and row/column manipulation routines for supported multidimensional array types.
- DataLoaderRoutines
Provides routines for loading separator-delimited text data into two-dimensional float arrays.
- GenericArrayRoutines
Provides utility methods for working with generic types, including span conversion, shape validation, and row-count retrieval.
- MatrixRoutines
Provides extension methods for float array operations, including arithmetic, elementwise transformations, shape-preserving initialization, clipping, transposition, and matrix multiplication.
- PermutationRoutines
Provides routines for randomly permuting samples along the first dimension of multidimensional arrays, including in-place and cloned permutations, and synchronized permutations of paired feature and label arrays.
- PreprocessingRoutines
Provides preprocessing routines for numeric array data, including scaling and standardization operations for 2D and 4D float arrays.
- ReshapeRoutines
Provides reshape routines for converting between 4D and 2D float arrays by flattening and restoring trailing dimensions.
- StatisticRoutines
Provides extension methods for computing common statistics on single- and multi-dimensional float arrays, including argmax, min, max, sum, mean, standard deviation, and per-column aggregates.