Orange: Stochastic Gradient Descent - Revision history

Onnowpurbo: /* Contoh */

2020-04-05T11:22:09Z

Contoh

Onnowpurbo: /* Contoh */

2020-04-05T11:21:34Z

Contoh

Onnowpurbo at 11:21, 5 April 2020

2020-04-05T11:21:17Z

Onnowpurbo: /* Output */

2020-04-05T11:19:57Z

Output

Onnowpurbo at 21:41, 6 March 2020

2020-03-06T21:41:10Z

Onnowpurbo at 04:21, 28 January 2020

2020-01-28T04:21:33Z

Onnowpurbo at 03:12, 23 January 2020

2020-01-23T03:12:11Z

Onnowpurbo: Created page with "Sumber: https://docs.biolab.si//3/visual-programming/widgets/model/stochasticgradient.html Minimize an objective function using a stochastic approximation of gradient descen..."

2020-01-16T07:11:47Z

Created page with "Sumber: https://docs.biolab.si//3/visual-programming/widgets/model/stochasticgradient.html Minimize an objective function using a stochastic approximation of gradient descen..."

New page

Sumber: https://docs.biolab.si//3/visual-programming/widgets/model/stochasticgradient.html

Minimize an objective function using a stochastic approximation of gradient descent.

Inputs

Data: input dataset

Preprocessor: preprocessing method(s)

Outputs

Learner: stochastic gradient descent learning algorithm

Model: trained model

The Stochastic Gradient Descent widget uses stochastic gradient descent that minimizes a chosen loss function with a linear function. The algorithm approximates a true gradient by considering one sample at a time, and simultaneously updates the model based on the gradient of the loss function. For regression, it returns predictors as minimizers of the sum, i.e. M-estimators, and is especially useful for large-scale and sparse datasets.

../../_images/StochasticGradientDescent-stamped.png

Specify the name of the model. The default name is “SGD”.

Algorithm parameters:

Classification loss function:

Hinge (linear SVM)

Logistic Regression (logistic regression SGD)

Modified Huber (smooth loss that brings tolerance to outliers as well as probability estimates)

Squared Hinge (quadratically penalized hinge)

Perceptron (linear loss used by the perceptron algorithm)

Squared Loss (fitted to ordinary least-squares)

Huber (switches to linear loss beyond ε)

Epsilon insensitive (ignores errors within ε, linear beyond it)

Squared epsilon insensitive (loss is squared beyond ε-region).

Regression loss function:

Squared Loss (fitted to ordinary least-squares)

Huber (switches to linear loss beyond ε)

Epsilon insensitive (ignores errors within ε, linear beyond it)

Squared epsilon insensitive (loss is squared beyond ε-region).

Regularization norms to prevent overfitting:

None.

Lasso (L1) (L1 leading to sparse solutions)

Ridge (L2) (L2, standard regularizer)

Elastic net (mixing both penalty norms).

Regularization strength defines how much regularization will be applied (the less we regularize, the more we allow the model to fit the data) and the mixing parameter what the ratio between L1 and L2 loss will be (if set to 0 then the loss is L2, if set to 1 then it is L1).

Learning parameters.

Learning rate:

Constant: learning rate stays the same through all epochs (passes)

Optimal: a heuristic proposed by Leon Bottou

Inverse scaling: earning rate is inversely related to the number of iterations

Initial learning rate.

Inverse scaling exponent: learning rate decay.

Number of iterations: the number of passes through the training data.

If Shuffle data after each iteration is on, the order of data instances is mixed after each pass.

If Fixed seed for random shuffling is on, the algorithm will use a fixed random seed and enable replicating the results.

Produce a report.

Press Apply to commit changes. Alternatively, tick the box on the left side of the Apply button and changes will be communicated automatically.

Examples

For the classification task, we will use iris dataset and test two models on it. We connected Stochastic Gradient Descent and Tree to Test & Score. We also connected File to Test & Score and observed model performance in the widget.

../../_images/StochasticGradientDescent-classification.png

For the regression task, we will compare three different models to see which predict what kind of results. For the purpose of this example, the housing dataset is used. We connect the File widget to Stochastic Gradient Descent, Linear Regression and kNN widget and all four to the Predictions widget.

../../_images/StochasticGradientDescent-regression.png

==Referensi==

* https://docs.biolab.si//3/visual-programming/widgets/model/stochasticgradient.html

==Pranala Menarik==

* [[Orange]]

@@ Line 61: / Line 61: @@
 ==Contoh==
-Dalam task classification, kita akan menggunakan dataset iris dan test dua model ke data tersebut. Kita akan connect widget Stochastic Gradient Descent dan widget Tree ke widget Test & Score. Kita akan juga connect widget File ke widget Test & Score dan mengamati performance model di widget.
+Dalam task classification, kita akan menggunakan dataset iris dan men-test dua model ke data tersebut. Kita akan connect widget Stochastic Gradient Descent dan widget Tree ke widget Test & Score. Kita akan juga connect widget File ke widget Test & Score dan mengamati performance model di widget.
 [[File:StochasticGradientDescent-classification.png|center|600px|thumb]]

@@ Line 63: / Line 63: @@
 Dalam task classification, kita akan menggunakan dataset iris dan test dua model ke data tersebut. Kita akan connect widget Stochastic Gradient Descent dan widget Tree ke widget Test & Score. Kita akan juga connect widget File ke widget Test & Score dan mengamati performance model di widget.
-[[File:StochasticGradientDescent-classification.png|center|200px|thumb]]
+[[File:StochasticGradientDescent-classification.png|center|600px|thumb]]
 Dalam task regression, kita akan membandingkan tiga (3) model yang berbeda untuk melihat masing-masing akan menghasilkan prediksi seperti apa. Untuk kebutuhan ini, dataset housing akan digunakan. Kita connect widget File ke widget Stochastic Gradient Descent, widget Linear Regression dan widget kNN dan ke semuanya di connect ke widget Predictions.
-[[File:StochasticGradientDescent-regression.png|center|200px|thumb]]
+[[File:StochasticGradientDescent-regression.png|center|600px|thumb]]
 ==Referensi==

@@ Line 2: / Line 2: @@
-Meminimalkan objective function menggunakan stochastic approximation dari gradient descent.
+Widget Stochastic Gradient Descent meminimalkan objective function menggunakan stochastic approximation dari gradient descent.
 ==Input==
@@ Line 61: / Line 61: @@
 ==Contoh==
-Untuk task classification, kita akan menggunakan dataset iris dan test dua model ke data tersebut. Kita akan connect widget Stochastic Gradient Descent dan widget Tree ke widget Test & Score. Kita akan juga connect widget File ke widget Test & Score dan mengamati performance model di widget.
+Dalam task classification, kita akan menggunakan dataset iris dan test dua model ke data tersebut. Kita akan connect widget Stochastic Gradient Descent dan widget Tree ke widget Test & Score. Kita akan juga connect widget File ke widget Test & Score dan mengamati performance model di widget.
 [[File:StochasticGradientDescent-classification.png|center|200px|thumb]]
-Untuk task regression, kita akan membandingkan tiga (3) model yang berbeda untuk melihat masing-masing akan menghasilkan prediksi seperti apa. Untuk kebutuhan ini, dataset housing akan digunakan. Kita connect widget File ke widget Stochastic Gradient Descent, Linear Regression dan kNN dan ke semuanya di connect ke widget Predictions.
+Dalam task regression, kita akan membandingkan tiga (3) model yang berbeda untuk melihat masing-masing akan menghasilkan prediksi seperti apa. Untuk kebutuhan ini, dataset housing akan digunakan. Kita connect widget File ke widget Stochastic Gradient Descent, widget Linear Regression dan widget kNN dan ke semuanya di connect ke widget Predictions.
 [[File:StochasticGradientDescent-regression.png|center|200px|thumb]]

@@ Line 14: / Line 14: @@
   Model: trained model
-Widget Stochastic Gradient Descent menggunakan stochastic gradient descent yang akan meminimalkan loss function terpilih dengan fungsi liniar. Algoritma akan mengira-ngira gradient yang benar dengan mempertimbangkan satu sampel pada satu waktu, dan secara bersamaan memperbarui model berdasarkan pada gradien dari loss function. Untuk regression, dia akan memberikan predictor sebagai minimizers of the sum, yaitu, M-estimators, dan sangat bermanfaat untuk dataset large-scale dan sparse.
+Widget Stochastic Gradient Descent menggunakan stochastic gradient descent yang akan meminimalkan loss function terpilih dengan fungsi linier. Algoritma akan mengira-ngira gradient yang benar dengan mempertimbangkan satu sampel pada satu waktu, dan secara bersamaan memperbarui model berdasarkan pada gradien dari loss function. Untuk regression,
 [[File:StochasticGradientDescent-stamped.png|center|200px|thumb]]

@@ Line 2: / Line 2: @@
-Minimize an objective function using a stochastic approximation of gradient descent.
+Meminimalkan objective function menggunakan stochastic approximation dari gradient descent.
 ==Input==
@@ Line 14: / Line 14: @@
   Model: trained model
-The Stochastic Gradient Descent widget uses stochastic gradient descent that minimizes a chosen loss function with a linear function. The algorithm approximates a true gradient by considering one sample at a time, and simultaneously updates the model based on the gradient of the loss function. For regression, it returns predictors as minimizers of the sum, i.e. M-estimators, and is especially useful for large-scale and sparse datasets.
+Widget Stochastic Gradient Descent menggunakan stochastic gradient descent yang akan meminimalkan loss function terpilih dengan fungsi liniar. Algoritma akan mengira-ngira gradient yang benar dengan mempertimbangkan satu sampel pada satu waktu, dan secara bersamaan memperbarui model berdasarkan pada gradien dari loss function. Untuk regression, dia akan memberikan predictor sebagai minimizers of the sum, yaitu, M-estimators, dan sangat bermanfaat untuk dataset large-scale dan sparse.
 [[File:StochasticGradientDescent-stamped.png|center|200px|thumb]]
@@ Line 60: / Line 60: @@
 ==Contoh==
-For the classification task, we will use iris dataset and test two models on it. We connected Stochastic Gradient Descent and Tree to Test & Score. We also connected File to Test & Score and observed model performance in the widget.
+Untuk task classification, kita akan menggunakan dataset iris dan test dua model ke data tersebut. Kita akan connect widget Stochastic Gradient Descent dan widget Tree ke widget Test & Score. Kita akan juga connect widget File ke widget Test & Score dan mengamati performance model di widget.
 [[File:StochasticGradientDescent-classification.png|center|200px|thumb]]
-For the regression task, we will compare three different models to see which predict what kind of results. For the purpose of this example, the housing dataset is used. We connect the File widget to Stochastic Gradient Descent, Linear Regression and kNN widget and all four to the Predictions widget.
+Untuk task regression, kita akan membandingkan tiga (3) model yang berbeda untuk melihat masing-masing akan menghasilkan prediksi seperti apa. Untuk kebutuhan ini, dataset housing akan digunakan. Kita connect widget File ke widget Stochastic Gradient Descent, Linear Regression dan kNN dan ke semuanya di connect ke widget Predictions.
 [[File:StochasticGradientDescent-regression.png|center|200px|thumb]]

@@ Line 4: / Line 4: @@
 Minimize an objective function using a stochastic approximation of gradient descent.
-Inputs
+==Input==
-    Data: input dataset
+ Data: input dataset
-    Preprocessor: preprocessing method(s)
+ Preprocessor: preprocessing method(s)
-Outputs
+==Output==
-    Learner: stochastic gradient descent learning algorithm
+ Learner: stochastic gradient descent learning algorithm
-    Model: trained model
+ Model: trained model
 The Stochastic Gradient Descent widget uses stochastic gradient descent that minimizes a chosen loss function with a linear function. The algorithm approximates a true gradient by considering one sample at a time, and simultaneously updates the model based on the gradient of the loss function. For regression, it returns predictors as minimizers of the sum, i.e. M-estimators, and is especially useful for large-scale and sparse datasets.
@@ Line 18: / Line 18: @@
 [[File:StochasticGradientDescent-stamped.png|center|200px|thumb]]
-    Specify the name of the model. The default name is “SGD”.
+* Specify the name of the model. The default name is “SGD”.
-    Algorithm parameters:
+* Algorithm parameters:
-        Classification loss function:
-            Hinge (linear SVM)
+** Classification loss function:
-            Logistic Regression (logistic regression SGD)
+*** Hinge (linear SVM)
-            Modified Huber (smooth loss that brings tolerance to outliers as well as probability estimates)
+*** Logistic Regression (logistic regression SGD)
-            Squared Hinge (quadratically penalized hinge)
+*** Modified Huber (smooth loss that brings tolerance to outliers as well as probability estimates)
-            Perceptron (linear loss used by the perceptron algorithm)
+*** Squared Hinge (quadratically penalized hinge)
-            Squared Loss (fitted to ordinary least-squares)
+*** Perceptron (linear loss used by the perceptron algorithm)
-            Huber (switches to linear loss beyond ε)
+*** Squared Loss (fitted to ordinary least-squares)
-            Epsilon insensitive (ignores errors within ε, linear beyond it)
+*** Huber (switches to linear loss beyond ε)
-            Squared epsilon insensitive (loss is squared beyond ε-region).
+*** Epsilon insensitive (ignores errors within ε, linear beyond it)
-        Regression loss function:
+*** Squared epsilon insensitive (loss is squared beyond ε-region).
-            Squared Loss (fitted to ordinary least-squares)
-            Huber (switches to linear loss beyond ε)
+** Regression loss function:
-            Epsilon insensitive (ignores errors within ε, linear beyond it)
+*** Squared Loss (fitted to ordinary least-squares)
-            Squared epsilon insensitive (loss is squared beyond ε-region).
+*** Huber (switches to linear loss beyond ε)
-    Regularization norms to prevent overfitting:
+*** Epsilon insensitive (ignores errors within ε, linear beyond it)
-        None.
+*** Squared epsilon insensitive (loss is squared beyond ε-region).
-        Lasso (L1) (L1 leading to sparse solutions)
-        Ridge (L2) (L2, standard regularizer)
+* Regularization norms to prevent overfitting:
-        Elastic net (mixing both penalty norms).
+** None.
-    Regularization strength defines how much regularization will be applied (the less we regularize, the more we allow the model to fit the data) and the mixing parameter what the ratio between L1 and L2 loss will be (if set to 0 then the loss is L2, if set to 1 then it is L1).
+** Lasso (L1) (L1 leading to sparse solutions)
-    Learning parameters.
+** Ridge (L2) (L2, standard regularizer)
-        Learning rate:
+** Elastic net (mixing both penalty norms).
-            Constant: learning rate stays the same through all epochs (passes)
+* Regularization strength defines how much regularization will be applied (the less we regularize, the more we allow the model to fit the data) and the mixing parameter what the ratio between L1 and L2 loss will be (if set to 0 then the loss is L2, if set to 1 then it is L1).
-            Optimal: a heuristic proposed by Leon Bottou
+* Learning parameters.
-            Inverse scaling: earning rate is inversely related to the number of iterations
+** Learning rate:
-        Initial learning rate.
+*** Constant: learning rate stays the same through all epochs (passes)
-        Inverse scaling exponent: learning rate decay.
+*** Optimal: a heuristic proposed by Leon Bottou
-        Number of iterations: the number of passes through the training data.
+*** Inverse scaling: earning rate is inversely related to the number of iterations
-        If Shuffle data after each iteration is on, the order of data instances is mixed after each pass.
+** Initial learning rate.
-        If Fixed seed for random shuffling is on, the algorithm will use a fixed random seed and enable replicating the results.
+** Inverse scaling exponent: learning rate decay.
-    Produce a report.
+** Number of iterations: the number of passes through the training data.
-    Press Apply to commit changes. Alternatively, tick the box on the left side of the Apply button and changes will be communicated automatically.
+** If Shuffle data after each iteration is on, the order of data instances is mixed after each pass.
 ==Contoh==