@am/neuralnetwork@1.0.0

Built and signed on GitHub Actions

Built and signed on GitHub Actions

Represents a neural network model composed of a sequence of layers.

This module provides a basic Tensor class for numerical operations, primarily supporting 2D tensors (matrices).

Abstract base class for activation function layers. All activation layers must implement the forward and backward methods.

Represents a fully connected (dense) layer in a neural network. Implements both Layer and TrainableLayer interfaces.

Rectified Linear Unit (ReLU) activation function. Outputs the input directly if it is positive, otherwise, it outputs zero. Formula: ReLU(x) = max(0, x)

Sigmoid activation function. Outputs values between 0 and 1. Formula: S(x) = 1 / (1 + e^(-x))

Softmax activation function. Typically used in the output layer of a multi-class classification network. Converts a vector of K real numbers into a probability distribution of K possible outcomes. Formula: Softmax(x_i) = e^(x_i) / sum(e^(x_j)) for j = 1 to K.

Hyperbolic Tangent (Tanh) activation function. Outputs values between -1 and 1. Formula: tanh(x) = (e^x - e^(-x)) / (e^x + e^(-x))

BinaryCrossEntropyLoss calculates the binary cross-entropy loss between predictions and target values. This loss is commonly used for binary classification tasks.

CrossEntropyLoss calculates the cross-entropy loss between predictions and target values. This loss is commonly used for classification tasks.

Calculates the Hinge Loss between predictions and targets. Commonly used for "maximum-margin" classification, most notably for support vector machines (SVMs). For a predicted score y and a true label t (either -1 or 1): L = max(0, 1 - t * y)

Calculates the Huber Loss between predictions and targets. Huber Loss is less sensitive to outliers than MSE and behaves like MAE for large errors.

Calculates the Mean Absolute Error (MAE) between predictions and target values. MAE is defined as the average of the absolute differences between predicted and actual values. Formula: MAE = (1/n) * Σ|prediction_i - target_i|

Calculates the Mean Squared Error (MSE) between predictions and target values. MSE is defined as the average of the squared differences between predicted and actual values. Formula: MSE = (1/n) * Σ(prediction_i - target_i)^2

Adagrad (Adaptive Gradient Algorithm) optimizer. Adagrad is an optimizer with parameter-specific learning rates, which are adapted relative to how frequently a parameter gets updated during training. The more a parameter receives updates, the smaller the updates will be.

Adam (Adaptive Moment Estimation) optimizer. Adam is an optimization algorithm that can be used instead of the classical stochastic gradient descent procedure to update network weights iteratively based on training data.

Implements the RMSprop (Root Mean Square Propagation) optimization algorithm.

Implements the Stochastic Gradient Descent (SGD) optimization algorithm.