@reicek/neataptic-ts 0.1.21 → 0.1.22

package/src/methods/README.md

@@ -28,10 +28,6 @@ rounded network outputs (thresholded at 0.5) against the target labels.
 It assumes target values are 0 or 1, and outputs are probabilities between 0 and 1.
 Note: This is equivalent to `1 - accuracy` for binary classification.
 
-Parameters:
-- `` - - An array of target values, expected to be 0 or 1.
-- `` - - An array of output values from the network, typically probabilities between 0 and 1.
-
 Returns: The proportion of misclassified samples (error rate, between 0 and 1).
 
 #### crossEntropy
@@ -47,10 +43,6 @@ predicted probability diverges from the actual label.
 It uses a small epsilon (PROB_EPSILON = 1e-15) to prevent `log(0)` which would result in `NaN`.
 Output values are clamped to the range `[epsilon, 1 - epsilon]` for numerical stability.
 
-Parameters:
-- `` - - An array of target values, typically 0 or 1 for binary classification, or probabilities for soft labels.
-- `` - - An array of output values from the network, representing probabilities (expected to be between 0 and 1).
-
 Returns: The mean cross-entropy error over all samples.
 
 #### focalLoss
@@ -60,12 +52,6 @@ Returns: The mean cross-entropy error over all samples.
 Calculates the Focal Loss, which is useful for addressing class imbalance in classification tasks.
 Focal loss down-weights easy examples and focuses training on hard negatives.
 
-Parameters:
-- `` - - Array of target values (0 or 1 for binary, or probabilities for soft labels).
-- `` - - Array of predicted probabilities (between 0 and 1).
-- `` - - Focusing parameter (default 2).
-- `` - - Balancing parameter (default 0.25).
-
 Returns: The mean focal loss.
 
 #### hinge
@@ -79,10 +65,6 @@ Hinge loss is used for training classifiers. It penalizes predictions that are
 not only incorrect but also those that are correct but not confident (i.e., close to the decision boundary).
 Assumes target values are encoded as -1 or 1.
 
-Parameters:
-- `` - - An array of target values, expected to be -1 or 1.
-- `` - - An array of output values from the network (raw scores, not necessarily probabilities).
-
 Returns: The mean hinge loss.
 
 #### labelSmoothing
@@ -92,11 +74,6 @@ Returns: The mean hinge loss.
 Calculates the Cross Entropy with Label Smoothing.
 Label smoothing prevents the model from becoming overconfident by softening the targets.
 
-Parameters:
-- `` - - Array of target values (0 or 1 for binary, or probabilities for soft labels).
-- `` - - Array of predicted probabilities (between 0 and 1).
-- `` - - Smoothing factor (between 0 and 1, e.g., 0.1).
-
 Returns: The mean cross-entropy loss with label smoothing.
 
 #### mae
@@ -108,10 +85,6 @@ Calculates the Mean Absolute Error (MAE), another common loss function for regre
 MAE measures the average of the absolute differences between predictions and actual values.
 Compared to MSE, it is less sensitive to outliers because errors are not squared.
 
-Parameters:
-- `` - - An array of target numerical values.
-- `` - - An array of output values from the network.
-
 Returns: The mean absolute error.
 
 #### mape
@@ -125,10 +98,6 @@ for understanding the error relative to the magnitude of the target values.
 However, it has limitations: it's undefined when the target value is zero and
 can be skewed by target values close to zero.
 
-Parameters:
-- `` - - An array of target numerical values. Should not contain zeros for standard MAPE.
-- `` - - An array of output values from the network.
-
 Returns: The mean absolute percentage error, expressed as a proportion (e.g., 0.1 for 10%).
 
 #### mse
@@ -141,10 +110,6 @@ MSE measures the average of the squares of the errors—that is, the average
 squared difference between the estimated values and the actual value.
 It is sensitive to outliers due to the squaring of the error terms.
 
-Parameters:
-- `` - - An array of target numerical values.
-- `` - - An array of output values from the network.
-
 Returns: The mean squared error.
 
 #### msle
@@ -159,10 +124,6 @@ It measures the squared difference between the logarithms of the predicted and a
 Uses `log(1 + x)` instead of `log(x)` for numerical stability and to handle inputs of 0.
 Assumes both targets and outputs are non-negative.
 
-Parameters:
-- `` - - An array of target numerical values (assumed >= 0).
-- `` - - An array of output values from the network (assumed >= 0).
-
 Returns: The mean squared logarithmic error.
 
 #### softmaxCrossEntropy
@@ -382,10 +343,6 @@ rounded network outputs (thresholded at 0.5) against the target labels.
 It assumes target values are 0 or 1, and outputs are probabilities between 0 and 1.
 Note: This is equivalent to `1 - accuracy` for binary classification.
 
-Parameters:
-- `` - - An array of target values, expected to be 0 or 1.
-- `` - - An array of output values from the network, typically probabilities between 0 and 1.
-
 Returns: The proportion of misclassified samples (error rate, between 0 and 1).
 
 #### cosineAnnealing
@@ -434,10 +391,6 @@ predicted probability diverges from the actual label.
 It uses a small epsilon (PROB_EPSILON = 1e-15) to prevent `log(0)` which would result in `NaN`.
 Output values are clamped to the range `[epsilon, 1 - epsilon]` for numerical stability.
 
-Parameters:
-- `` - - An array of target values, typically 0 or 1 for binary classification, or probabilities for soft labels.
-- `` - - An array of output values from the network, representing probabilities (expected to be between 0 and 1).
-
 Returns: The mean cross-entropy error over all samples.
 
 #### exp
@@ -482,12 +435,6 @@ Returns: A function that takes the base learning rate and the current iteration
 Calculates the Focal Loss, which is useful for addressing class imbalance in classification tasks.
 Focal loss down-weights easy examples and focuses training on hard negatives.
 
-Parameters:
-- `` - - Array of target values (0 or 1 for binary, or probabilities for soft labels).
-- `` - - Array of predicted probabilities (between 0 and 1).
-- `` - - Focusing parameter (default 2).
-- `` - - Balancing parameter (default 0.25).
-
 Returns: The mean focal loss.
 
 #### hinge
@@ -501,10 +448,6 @@ Hinge loss is used for training classifiers. It penalizes predictions that are
 not only incorrect but also those that are correct but not confident (i.e., close to the decision boundary).
 Assumes target values are encoded as -1 or 1.
 
-Parameters:
-- `` - - An array of target values, expected to be -1 or 1.
-- `` - - An array of output values from the network (raw scores, not necessarily probabilities).
-
 Returns: The mean hinge loss.
 
 #### inv
@@ -534,11 +477,6 @@ Returns: A function that calculates the inversely decayed learning rate for a gi
 Calculates the Cross Entropy with Label Smoothing.
 Label smoothing prevents the model from becoming overconfident by softening the targets.
 
-Parameters:
-- `` - - Array of target values (0 or 1 for binary, or probabilities for soft labels).
-- `` - - Array of predicted probabilities (between 0 and 1).
-- `` - - Smoothing factor (between 0 and 1, e.g., 0.1).
-
 Returns: The mean cross-entropy loss with label smoothing.
 
 #### linearWarmupDecay
@@ -563,10 +501,6 @@ Calculates the Mean Absolute Error (MAE), another common loss function for regre
 MAE measures the average of the absolute differences between predictions and actual values.
 Compared to MSE, it is less sensitive to outliers because errors are not squared.
 
-Parameters:
-- `` - - An array of target numerical values.
-- `` - - An array of output values from the network.
-
 Returns: The mean absolute error.
 
 #### mape
@@ -580,10 +514,6 @@ for understanding the error relative to the magnitude of the target values.
 However, it has limitations: it's undefined when the target value is zero and
 can be skewed by target values close to zero.
 
-Parameters:
-- `` - - An array of target numerical values. Should not contain zeros for standard MAPE.
-- `` - - An array of output values from the network.
-
 Returns: The mean absolute percentage error, expressed as a proportion (e.g., 0.1 for 10%).
 
 #### mse
@@ -596,10 +526,6 @@ MSE measures the average of the squares of the errors—that is, the average
 squared difference between the estimated values and the actual value.
 It is sensitive to outliers due to the squaring of the error terms.
 
-Parameters:
-- `` - - An array of target numerical values.
-- `` - - An array of output values from the network.
-
 Returns: The mean squared error.
 
 #### msle
@@ -614,10 +540,6 @@ It measures the squared difference between the logarithms of the predicted and a
 Uses `log(1 + x)` instead of `log(x)` for numerical stability and to handle inputs of 0.
 Assumes both targets and outputs are non-negative.
 
-Parameters:
-- `` - - An array of target numerical values (assumed >= 0).
-- `` - - An array of output values from the network (assumed >= 0).
-
 Returns: The mean squared logarithmic error.
 
 #### reduceOnPlateau
@@ -703,7 +625,7 @@ but can slow down the search process.
 
 ### registerCustomActivation
 
-`(activationName: string, activationFunction: import("C:/NeatapticTS/src/methods/activation.utils").ActivationFunction) => void`
+`(activationName: string, activationFunction: import("src/methods/activation.utils").ActivationFunction) => void`
 
 ## methods/connection.ts
 
@@ -935,7 +857,7 @@ Useful for any stateless schedule strategy.
 
 ### createCosineAnnealingRateSchedule
 
-`(period: number, minimumRate: number) => import("C:/NeatapticTS/src/methods/rate.utils").RateSchedule`
+`(period: number, minimumRate: number) => import("src/methods/rate.utils").RateSchedule`
 
 Returns a cosine annealing learning rate schedule.
 
@@ -947,7 +869,7 @@ Returns: A learning rate schedule implementing cosine annealing.
 
 ### createCosineAnnealingWarmRestartsSchedule
 
-`(initialPeriod: number, minimumRate: number, periodGrowthMultiplier: number) => import("C:/NeatapticTS/src/methods/rate.utils").RateSchedule`
+`(initialPeriod: number, minimumRate: number, periodGrowthMultiplier: number) => import("src/methods/rate.utils").RateSchedule`
 
 Returns a cosine annealing schedule with warm restarts and growing cycles.
 
@@ -960,7 +882,7 @@ Returns: A learning rate schedule implementing SGDR-style warm restarts.
 
 ### createExponentialRateSchedule
 
-`(decayFactor: number) => import("C:/NeatapticTS/src/methods/rate.utils").RateSchedule`
+`(decayFactor: number) => import("src/methods/rate.utils").RateSchedule`
 
 Returns an exponential decay learning rate schedule.
 
@@ -971,7 +893,7 @@ Returns: A learning rate schedule implementing exponential decay.
 
 ### createFixedRateSchedule
 
-`() => import("C:/NeatapticTS/src/methods/rate.utils").RateSchedule`
+`() => import("src/methods/rate.utils").RateSchedule`
 
 Returns a schedule that always yields the base learning rate.
 
@@ -979,7 +901,7 @@ Returns: A learning rate schedule that ignores iteration and returns baseRate.
 
 ### createInverseRateSchedule
 
-`(decayFactor: number, decayPower: number) => import("C:/NeatapticTS/src/methods/rate.utils").RateSchedule`
+`(decayFactor: number, decayPower: number) => import("src/methods/rate.utils").RateSchedule`
 
 Returns an inverse decay learning rate schedule.
 
@@ -991,7 +913,7 @@ Returns: A learning rate schedule implementing inverse decay.
 
 ### createLinearWarmupDecaySchedule
 
-`(totalStepCount: number, warmupStepCount: number | undefined, endRate: number) => import("C:/NeatapticTS/src/methods/rate.utils").RateSchedule`
+`(totalStepCount: number, warmupStepCount: number | undefined, endRate: number) => import("src/methods/rate.utils").RateSchedule`
 
 Returns a linear warmup followed by linear decay schedule.
 
@@ -1004,7 +926,7 @@ Returns: A learning rate schedule implementing warmup then decay.
 
 ### createReduceOnPlateauSchedule
 
-`(options: { factor?: number | undefined; patience?: number | undefined; minDelta?: number | undefined; cooldown?: number | undefined; minRate?: number | undefined; verbose?: boolean | undefined; } | undefined) => import("C:/NeatapticTS/src/methods/rate.utils").ReduceOnPlateauSchedule`
+`(options: { factor?: number | undefined; patience?: number | undefined; minDelta?: number | undefined; cooldown?: number | undefined; minRate?: number | undefined; verbose?: boolean | undefined; } | undefined) => import("src/methods/rate.utils").ReduceOnPlateauSchedule`
 
 Returns a ReduceLROnPlateau-style schedule that lowers the rate when no improvement is seen.
 
@@ -1015,7 +937,7 @@ Returns: A stateful schedule that reacts to lack of improvement.
 
 ### createStepRateSchedule
 
-`(decayFactor: number, decayStepSize: number) => import("C:/NeatapticTS/src/methods/rate.utils").RateSchedule`
+`(decayFactor: number, decayStepSize: number) => import("src/methods/rate.utils").RateSchedule`
 
 Returns a step decay learning rate schedule.
 

package/src/multithreading/README.md

@@ -26,24 +26,14 @@ emphasis on efficient evaluation and mutation of neural networks in parallel env
 
 Absolute activation function.
 
-Parameters:
-- `` - - The input value.
-
 Returns: The activated value.
 
 #### activateSerializedNetwork
 
-`(inputValues: number[], activationValues: number[], stateValues: number[], serializedNetwork: number[], activationFunctions: import("C:/NeatapticTS/src/multithreading/types").ActivationFn[]) => number[]`
+`(inputValues: number[], activationValues: number[], stateValues: number[], serializedNetwork: number[], activationFunctions: import("src/multithreading/types").ActivationFn[]) => number[]`
 
 Activates a serialized network.
 
-Parameters:
-- `` - - The input values.
-- `` - - The activations array.
-- `` - - The states array.
-- `` - - The serialized network data.
-- `` - - The activation functions.
-
 Returns: The output values.
 
 #### activations
@@ -56,9 +46,6 @@ A list of compiled activation functions in a specific order.
 
 Bent Identity activation function.
 
-Parameters:
-- `` - - The input value.
-
 Returns: The activated value.
 
 #### bipolar
@@ -67,9 +54,6 @@ Returns: The activated value.
 
 Bipolar activation function.
 
-Parameters:
-- `` - - The input value.
-
 Returns: The activated value.
 
 #### bipolarSigmoid
@@ -78,20 +62,14 @@ Returns: The activated value.
 
 Bipolar Sigmoid activation function.
 
-Parameters:
-- `` - - The input value.
-
 Returns: The activated value.
 
 #### deserializeDataSet
 
-`(serializedSet: number[]) => import("C:/NeatapticTS/src/multithreading/types").SerializedSample[]`
+`(serializedSet: number[]) => import("src/multithreading/types").SerializedSample[]`
 
 Deserializes a dataset from a flat array.
 
-Parameters:
-- `` - - The serialized dataset.
-
 Returns: The deserialized dataset as an array of input-output pairs.
 
 #### gaussian
@@ -100,14 +78,11 @@ Returns: The deserialized dataset as an array of input-output pairs.
 
 Gaussian activation function.
 
-Parameters:
-- `` - - The input value.
-
 Returns: The activated value.
 
 #### getBrowserTestWorker
 
-`() => Promise<import("C:/NeatapticTS/src/multithreading/types").TestWorkerConstructor>`
+`() => Promise<import("src/multithreading/types").TestWorkerConstructor>`
 
 Gets the browser test worker.
 
@@ -115,7 +90,7 @@ Returns: The browser test worker.
 
 #### getNodeTestWorker
 
-`() => Promise<import("C:/NeatapticTS/src/multithreading/types").TestWorkerConstructor>`
+`() => Promise<import("src/multithreading/types").TestWorkerConstructor>`
 
 Gets the node test worker.
 
@@ -127,9 +102,6 @@ Returns: The node test worker.
 
 Hard Tanh activation function.
 
-Parameters:
-- `` - - The input value.
-
 Returns: The activated value.
 
 #### identity
@@ -138,9 +110,6 @@ Returns: The activated value.
 
 Identity activation function.
 
-Parameters:
-- `` - - The input value.
-
 Returns: The activated value.
 
 #### inverse
@@ -149,9 +118,6 @@ Returns: The activated value.
 
 Inverse activation function.
 
-Parameters:
-- `` - - The input value.
-
 Returns: The activated value.
 
 #### logistic
@@ -160,9 +126,6 @@ Returns: The activated value.
 
 Logistic activation function.
 
-Parameters:
-- `` - - The input value.
-
 Returns: The activated value.
 
 #### relu
@@ -171,9 +134,6 @@ Returns: The activated value.
 
 Rectified Linear Unit (ReLU) activation function.
 
-Parameters:
-- `` - - The input value.
-
 Returns: The activated value.
 
 #### selu
@@ -182,9 +142,6 @@ Returns: The activated value.
 
 Scaled Exponential Linear Unit (SELU) activation function.
 
-Parameters:
-- `` - - The input value.
-
 Returns: The activated value.
 
 #### serializeDataSet
@@ -193,9 +150,6 @@ Returns: The activated value.
 
 Serializes a dataset into a flat array.
 
-Parameters:
-- `` - - The dataset to serialize.
-
 Returns: The serialized dataset.
 
 #### sinusoid
@@ -204,9 +158,6 @@ Returns: The serialized dataset.
 
 Sinusoid activation function.
 
-Parameters:
-- `` - - The input value.
-
 Returns: The activated value.
 
 #### softplus
@@ -215,9 +166,6 @@ Returns: The activated value.
 
 Softplus activation function. - Added
 
-Parameters:
-- `` - - The input value.
-
 Returns: The activated value.
 
 #### softsign
@@ -226,9 +174,6 @@ Returns: The activated value.
 
 Softsign activation function.
 
-Parameters:
-- `` - - The input value.
-
 Returns: The activated value.
 
 #### step
@@ -237,9 +182,6 @@ Returns: The activated value.
 
 Step activation function.
 
-Parameters:
-- `` - - The input value.
-
 Returns: The activated value.
 
 #### tanh
@@ -248,25 +190,14 @@ Returns: The activated value.
 
 Hyperbolic tangent activation function.
 
-Parameters:
-- `` - - The input value.
-
 Returns: The activated value.
 
 #### testSerializedSet
 
-`(serializedSampleSet: import("C:/NeatapticTS/src/multithreading/types").SerializedSample[], cost: (expected: number[], actual: number[]) => number, activationValues: number[], stateValues: number[], serializedNetwork: number[], activationFunctions: import("C:/NeatapticTS/src/multithreading/types").ActivationFn[]) => number`
+`(serializedSampleSet: import("src/multithreading/types").SerializedSample[], cost: (expected: number[], actual: number[]) => number, activationValues: number[], stateValues: number[], serializedNetwork: number[], activationFunctions: import("src/multithreading/types").ActivationFn[]) => number`
 
 Tests a serialized dataset using a cost function.
 
-Parameters:
-- `` - - The serialized dataset as an array of input-output pairs.
-- `` - - The cost function.
-- `` - - The activations array.
-- `` - - The states array.
-- `` - - The serialized network data.
-- `` - - The activation functions.
-
 Returns: The average error.
 
 #### workers
@@ -306,7 +237,7 @@ Returns: Absolute activation.
 
 ### activateSerializedNetwork
 
-`(inputValues: number[], activationValues: number[], stateValues: number[], serializedNetwork: number[], activationFunctions: import("C:/NeatapticTS/src/multithreading/types").ActivationFn[]) => number[]`
+`(inputValues: number[], activationValues: number[], stateValues: number[], serializedNetwork: number[], activationFunctions: import("src/multithreading/types").ActivationFn[]) => number[]`
 
 Activates a serialized network and produces outputs.
 
@@ -350,7 +281,7 @@ Returns: Bipolar sigmoid activation.
 
 ### deserializeDataSet
 
-`(serializedSet: number[]) => import("C:/NeatapticTS/src/multithreading/types").SerializedSample[]`
+`(serializedSet: number[]) => import("src/multithreading/types").SerializedSample[]`
 
 Deserializes a dataset from its flat representation.
 
@@ -480,7 +411,7 @@ Returns: Hyperbolic tangent activation.
 
 ### testSerializedSet
 
-`(serializedSampleSet: import("C:/NeatapticTS/src/multithreading/types").SerializedSample[], costFunction: (expected: number[], actual: number[]) => number, activationValues: number[], stateValues: number[], serializedNetwork: number[], activationFunctions: import("C:/NeatapticTS/src/multithreading/types").ActivationFn[]) => number`
+`(serializedSampleSet: import("src/multithreading/types").SerializedSample[], costFunction: (expected: number[], actual: number[]) => number, activationValues: number[], stateValues: number[], serializedNetwork: number[], activationFunctions: import("src/multithreading/types").ActivationFn[]) => number`
 
 Tests a serialized dataset using a cost function.
 

package/src/multithreading/workers/README.md

@@ -12,7 +12,7 @@ Utility class for managing workers in both Node.js and browser environments.
 
 #### getBrowserTestWorker
 
-`() => Promise<typeof import("C:/NeatapticTS/src/multithreading/workers/browser/testworker").TestWorker>`
+`() => Promise<typeof import("src/multithreading/workers/browser/testworker").TestWorker>`
 
 Loads the browser test worker dynamically.
 
@@ -20,7 +20,7 @@ Returns: A promise that resolves to the browser TestWorker class.
 
 #### getNodeTestWorker
 
-`() => Promise<typeof import("C:/NeatapticTS/src/multithreading/workers/node/testworker").TestWorker>`
+`() => Promise<typeof import("src/multithreading/workers/node/testworker").TestWorker>`
 
 Loads the Node.js test worker dynamically.
 

package/src/multithreading/workers/browser/README.md

@@ -24,9 +24,6 @@ computationally expensive tasks, such as network evaluation, to separate threads
 
 Creates a string representation of the worker's blob.
 
-Parameters:
-- `` - - The cost function to be used by the worker.
-
 Returns: The blob string.
 
 #### evaluate
@@ -35,9 +32,6 @@ Returns: The blob string.
 
 Evaluates a network using the worker process.
 
-Parameters:
-- `` - - The network to evaluate.
-
 Returns: A promise that resolves to the evaluation result.
 
 #### terminate

package/src/multithreading/workers/node/README.md

@@ -45,9 +45,6 @@ Evaluates a neural network using the worker process.
 The network is serialized and sent to the worker for evaluation. The worker
 sends back the evaluation result, which is returned as a promise.
 
-Parameters:
-- `` - - The neural network to evaluate. It must implement a `serialize` method.
-
 Returns: A promise that resolves to the evaluation result.
 
 #### terminate