Movatterモバイル変換

[0]ホーム
URL:

Skip to content

Navigation Menu

Sign in
Appearance settings

Search code, repositories, users, issues, pull requests...

Provide feedback

We read every piece of feedback, and take your input very seriously.

Saved searches

Use saved searches to filter your results more quickly

 Sign up
Appearance settings

Commitde62f66

Browse files
authored
PyGAD 2.10.1 Documentation
1. In the `gene_space` parameter, any `None` value (regardless of its index or axis), is replaced by a randomly generated number based on the 3 parameters `init_range_low`, `init_range_high`, and `gene_type`. So, the `None` value in `[..., None, ...]` or `[..., [..., None, ...], ...]` are replaced with random values. This gives more freedom in building the space of values for the genes.2. All the numbers passed to the `gene_space` parameter are casted to the type specified in the `gene_type` parameter.3. The `numpy.uint` data type is supported for the parameters that accept integer values.4. In the `pygad.kerasga` module, the `model_weights_as_vector()` function uses the `trainable` attribute of the model's layers to only return the trainable weights in the network. So, only the trainable layers with their `trainable` attribute set to `True` (`trainable=True`), which is the default value, have their weights evolved. All non-trainable layers with the `trainable` attribute set to `False` (`trainable=False`) will not be evolved. Thanks to [Prof. Tamer A. Farrag](https://github.com/tfarrag2000) for pointing about that at [GitHub](ahmedfgad/KerasGA#1).
1 parent6deb2c0 commitde62f66

File tree

4 files changed

+66
-28
lines changed

4 files changed

+66
-28
lines changed

‎docs/source/Footer.rst

Lines changed: 33 additions & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -461,7 +461,7 @@ Release Date: 03 January 2021
461461

462462
2. Support of adaptive mutation where the mutation rate is determined
463463
by the fitness value of each solution. Read the `Adaptive
464-
Mutation <https://pygad.readthedocs.io/en/latest/README_pygad_ReadTheDocs.html#adaptive-mutation>`__
464+
Mutation <https://pygad.readthedocs.io/en/latest/README_pygad_torchga_ReadTheDocs.html#adaptive-mutation>`__
465465
section for more details. Also, read this paper: `Libelli, S.
466466
Marsili, and P. Alba. "Adaptive mutation in genetic algorithms."
467467
Soft computing 4.2 (2000):
@@ -509,6 +509,38 @@ Release Date: 03 January 2021
509509
``cal_pop_fitness()`` method is called to calculate the fitness
510510
values of the population.
511511

512+
.. _header-n698:
513+
514+
PyGAD 2.10.1
515+
------------
516+
517+
Release Date: 10 January 2021
518+
519+
1. In the ``gene_space`` parameter, any ``None`` value (regardless of
520+
its index or axis), is replaced by a randomly generated number based
521+
on the 3 parameters ``init_range_low``, ``init_range_high``, and
522+
``gene_type``. So, the ``None`` value in ``[..., None, ...]`` or
523+
``[..., [..., None, ...], ...]`` are replaced with random values.
524+
This gives more freedom in building the space of values for the
525+
genes.
526+
527+
2. All the numbers passed to the ``gene_space`` parameter are casted to
528+
the type specified in the ``gene_type`` parameter.
529+
530+
3. The ``numpy.uint`` data type is supported for the parameters that
531+
accept integer values.
532+
533+
4. In the ``pygad.kerasga`` module, the ``model_weights_as_vector()``
534+
function uses the ``trainable`` attribute of the model's layers to
535+
only return the trainable weights in the network. So, only the
536+
trainable layers with their ``trainable`` attribute set to ``True``
537+
(``trainable=True``), which is the default value, have their weights
538+
evolved. All non-trainable layers with the ``trainable`` attribute
539+
set to ``False`` (``trainable=False``) will not be evolved. Thanks to
540+
`Prof. Tamer A. Farrag<https://github.com/tfarrag2000>`__ for
541+
pointing about that at
542+
`GitHub<https://github.com/ahmedfgad/KerasGA/issues/1>`__.
543+
512544
.. _header-n554:
513545

514546
PyGAD Projects at GitHub

‎docs/source/README_pygad_ReadTheDocs.rst

Lines changed: 2 additions & 2 deletions
Original file line numberDiff line numberDiff line change
@@ -132,10 +132,10 @@ The ``pygad.GA`` class constructor supports the following parameters:
132132
be used unchanged in the next generation. ``Adaptive`` mutation is
133133
supported starting from PyGAD 2.10.0. For more information about
134134
adaptive mutation, go the the `Adaptive
135-
Mutation <https://pygad.readthedocs.io/en/latest/README_pygad_ReadTheDocs.html#adaptive-mutation>`__
135+
Mutation <https://pygad.readthedocs.io/en/latest/README_pygad_torchga_ReadTheDocs.html#adaptive-mutation>`__
136136
section. For example about using adaptive mutation, check the `Use
137137
Adaptive Mutation in
138-
PyGAD <https://pygad.readthedocs.io/en/latest/README_pygad_ReadTheDocs.html#use-adaptive-mutation-in-pygad>`__
138+
PyGAD <https://pygad.readthedocs.io/en/latest/README_pygad_torchga_ReadTheDocs.html#use-adaptive-mutation-in-pygad>`__
139139
section.
140140

141141
- ``mutation_probability=None``: The probability of selecting a gene

‎docs/source/README_pygad_kerasga_ReadTheDocs.rst

Lines changed: 30 additions & 24 deletions
Original file line numberDiff line numberDiff line change
@@ -1,4 +1,4 @@
1-
.. _header-n0:
1+
.. _header-n194:
22

33
``pygad.kerasga`` Module
44
========================
@@ -25,7 +25,7 @@ The contents of this module are:
2525

2626
More details are given in the next sections.
2727

28-
.. _header-n13:
28+
.. _header-n207:
2929

3030
Steps Summary
3131
=============
@@ -45,7 +45,7 @@ follows:
4545

4646
6. Run the genetic algorithm.
4747

48-
.. _header-n28:
48+
.. _header-n222:
4949

5050
Create Keras Model
5151
==================
@@ -93,7 +93,7 @@ This is the same model created using the Functional API.
9393
9494
Feel free to add the layers of your choice.
9595

96-
.. _header-n44:
96+
.. _header-n238:
9797

9898
``pygad.kerasga.KerasGA`` Class
9999
===============================
@@ -103,7 +103,7 @@ an initial population for the genetic algorithm based on a Keras model.
103103
The constructor, methods, and attributes within the class are discussed
104104
in this section.
105105

106-
.. _header-n46:
106+
.. _header-n240:
107107

108108
``__init__()``
109109
--------------
@@ -116,7 +116,7 @@ parameters:
116116
- ``num_solutions``: Number of solutions in the population. Each
117117
solution has different parameters of the model.
118118

119-
.. _header-n53:
119+
.. _header-n247:
120120

121121
Instance Attributes
122122
-------------------
@@ -134,15 +134,15 @@ Here is a list of all instance attributes:
134134
- ``population_weights``: A nested list holding the weights of all
135135
solutions in the population.
136136

137-
.. _header-n63:
137+
.. _header-n257:
138138

139139
Methods in the ``KerasGA`` Class
140140
--------------------------------
141141

142142
This section discusses the methods available for instances of the
143143
``pygad.kerasga.KerasGA`` class.
144144

145-
.. _header-n65:
145+
.. _header-n259:
146146

147147
``create_population()``
148148
~~~~~~~~~~~~~~~~~~~~~~~
@@ -152,14 +152,14 @@ genetic algorithm as a list of solutions where each solution represents
152152
different model parameters. The list of networks is assigned to the
153153
``population_weights`` attribute of the instance.
154154

155-
.. _header-n67:
155+
.. _header-n261:
156156

157157
Functions in the ``pygad.kerasga`` Module
158158
=========================================
159159

160160
This section discusses the functions in the ``pygad.kerasga`` module.
161161

162-
.. _header-n69:
162+
.. _header-n263:
163163

164164
``pygad.kerasga.model_weights_as_vector()``
165165
--------------------------------------------
@@ -170,13 +170,19 @@ holding all model weights. The reason for representing the model weights
170170
as a vector is that the genetic algorithm expects all parameters of any
171171
solution to be in a 1D vector form.
172172

173+
This function filters the layers based on the ``trainable`` attribute to
174+
see whether the layer weights are trained or not. For each layer, if its
175+
``trainable=False``, then its weights will not be evolved using the
176+
genetic algorithm. Otherwise, it will be represented in the chromosome
177+
and evolved.
178+
173179
The function accepts the following parameters:
174180

175181
- ``model``: The Keras model.
176182

177183
It returns a 1D vector holding the model weights.
178184

179-
.. _header-n76:
185+
.. _header-n270:
180186

181187
``pygad.kerasga.model_weights_as_matrix()``
182188
-------------------------------------------
@@ -190,7 +196,7 @@ parameters:
190196

191197
It returns the restored model weights after reshaping the vector.
192198

193-
.. _header-n84:
199+
.. _header-n278:
194200

195201
Examples
196202
========
@@ -199,7 +205,7 @@ This section gives the complete code of some examples that build and
199205
train a Keras model using PyGAD. Each subsection builds a different
200206
network.
201207

202-
.. _header-n86:
208+
.. _header-n280:
203209

204210
Example 1: Regression Example
205211
-----------------------------
@@ -296,7 +302,7 @@ subsections discuss each part in the code.
296302
abs_error= mae(data_outputs, predictions).numpy()
297303
print("Absolute Error :", abs_error)
298304
299-
.. _header-n89:
305+
.. _header-n283:
300306

301307
Create a Keras Model
302308
~~~~~~~~~~~~~~~~~~~~
@@ -328,7 +334,7 @@ The model can also be build using the Keras Sequential Model API.
328334
model.add(dense_layer1)
329335
model.add(output_layer)
330336
331-
.. _header-n94:
337+
.. _header-n288:
332338

333339
Create an Instance of the ``pygad.kerasga.KerasGA`` Class
334340
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
@@ -344,7 +350,7 @@ Change this number according to your needs.
344350
keras_ga= pygad.kerasga.KerasGA(model=model,
345351
num_solutions=10)
346352
347-
.. _header-n97:
353+
.. _header-n291:
348354

349355
Prepare the Training Data
350356
~~~~~~~~~~~~~~~~~~~~~~~~~
@@ -369,7 +375,7 @@ output.
369375
[1.3],
370376
[2.5]])
371377
372-
.. _header-n100:
378+
.. _header-n294:
373379

374380
Build the Fitness Function
375381
~~~~~~~~~~~~~~~~~~~~~~~~~~
@@ -400,7 +406,7 @@ function to calculate the fitness value.
400406
401407
return solution_fitness
402408
403-
.. _header-n104:
409+
.. _header-n298:
404410

405411
Create an Instance of the ``pygad.GA`` Class
406412
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
@@ -435,7 +441,7 @@ accepts <https://pygad.readthedocs.io/en/latest/README_pygad_ReadTheDocs.html#in
435441
keep_parents=keep_parents,
436442
on_generation=callback_generation)
437443
438-
.. _header-n108:
444+
.. _header-n302:
439445

440446
Run the Genetic Algorithm
441447
~~~~~~~~~~~~~~~~~~~~~~~~~
@@ -508,7 +514,7 @@ The next code measures the trained model error.
508514
509515
Absolute Error : 0.013740465
510516
511-
.. _header-n124:
517+
.. _header-n318:
512518

513519
Example 2: XOR Binary Classification
514520
------------------------------------
@@ -679,7 +685,7 @@ Here is some information about the trained model. Its fitness value is
679685
680686
Accuracy :1.0
681687
682-
.. _header-n144:
688+
.. _header-n338:
683689

684690
Example 3: Image Multi-Class Classification (Dense Layers)
685691
----------------------------------------------------------
@@ -789,7 +795,7 @@ cross entropy.
789795
cce= tensorflow.keras.losses.CategoricalCrossentropy()
790796
solution_fitness=1.0/ (cce(data_outputs, predictions).numpy()+0.00000001)
791797
792-
.. _header-n149:
798+
.. _header-n343:
793799

794800
Prepare the Training Data
795801
~~~~~~~~~~~~~~~~~~~~~~~~~
@@ -839,7 +845,7 @@ Here are some statistics about the trained model.
839845
Categorical Crossentropy : 0.23823906
840846
Accuracy : 0.9852192
841847
842-
.. _header-n164:
848+
.. _header-n358:
843849

844850
Example 4: Image Multi-Class Classification (Conv Layers)
845851
---------------------------------------------------------
@@ -974,7 +980,7 @@ each input sample is 100x100x3.
974980
975981
model= tensorflow.keras.Model(inputs=input_layer,outputs=output_layer)
976982
977-
.. _header-n170:
983+
.. _header-n364:
978984

979985
Prepare the Training Data
980986
~~~~~~~~~~~~~~~~~~~~~~~~~

‎docs/source/conf.py

Lines changed: 1 addition & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -22,7 +22,7 @@
2222
author='Ahmed Fawzy Gad'
2323

2424
# The full version, including alpha/beta/rc tags
25-
release='2.10.0'
25+
release='2.10.1'
2626

2727
master_doc='index'
2828

0 commit comments

Comments
 (0)
[8]ページ先頭
©2009-2025 Movatter.jp