pyerualjetwork 4.3.13.dev0__py3-none-any.whl → 4.5__py3-none-any.whl

pyerualjetwork/__init__.py

@@ -1,4 +1,4 @@
-__version__ = "4.3.13dev0"
+__version__ = "4.5"
 __update__ = """* Changes: https://github.com/HCB06/PyerualJetwork/blob/main/CHANGES
 * PyerualJetwork Homepage: https://github.com/HCB06/PyerualJetwork/tree/main
 * PyerualJetwork document: https://github.com/HCB06/PyerualJetwork/blob/main/Welcome_to_PyerualJetwork/PYERUALJETWORK_USER_MANUEL_AND_LEGAL_INFORMATION(EN).pdf

pyerualjetwork/data_operations.py

@@ -373,6 +373,26 @@ dtype=np.float32):
     return (Input / MaxAbs)
 
 
+def non_neg_normalization(
+    Input, 
+    dtype=np.float32
+):
+    """
+    Normalizes the input data [0-1] range.
+
+    Args:
+        Input (cupy): Input data to be normalized.
+
+        dtype (cupy.dtype): Data type for the arrays. cp.float32 by default. Example: cp.float64 or cp.float16. [fp32 for balanced devices, fp64 for strong devices, fp16 for weak devices: not reccomended!] (optional)
+
+    Returns:
+        (cupy) Scaled input data after normalization.
+    """
+
+    MaxAbs = np.max(np.abs(Input.astype(dtype, copy=False)))
+    return (Input + MaxAbs) / (2 * MaxAbs)
+
+
 def find_closest_factors(a):
 
     root = int(math.sqrt(a))

pyerualjetwork/data_operations_cuda.py

@@ -421,6 +421,27 @@ def normalization(
     MaxAbs = cp.max(cp.abs(Input.astype(dtype, copy=False)))
     return (Input / MaxAbs)
 
+
+def non_neg_normalization(
+    Input, 
+    dtype=cp.float32
+):
+    """
+    Normalizes the input data [0-1] range.
+
+    Args:
+        Input (cupy): Input data to be normalized.
+
+        dtype (cupy.dtype): Data type for the arrays. cp.float32 by default. Example: cp.float64 or cp.float16. [fp32 for balanced devices, fp64 for strong devices, fp16 for weak devices: not reccomended!] (optional)
+
+    Returns:
+        (cupy) Scaled input data after normalization.
+    """
+
+    MaxAbs = cp.max(cp.abs(Input.astype(dtype, copy=False)))
+    return (Input + MaxAbs) / (2 * MaxAbs)
+
+
 def find_closest_factors(a):
 
     root = int(math.sqrt(a))

pyerualjetwork/fitness_functions.py

@@ -1,3 +1,5 @@
+import numpy as np
+
 def wals(acc, loss, acc_impact, loss_impact):
     """
     The WALS(weighted accuracy-loss score) function calculates a weighted sum of accuracy and loss based on their respective impacts.
@@ -13,4 +15,7 @@ def wals(acc, loss, acc_impact, loss_impact):
      much the loss affects the final result compared to the accuracy impact
     :return: the weighted sum of accuracy and loss based on their respective impacts.
     """
-    return (acc * acc_impact) + (-loss * loss_impact)
+    loss += np.finfo(float).eps
+    loss_impact += np.finfo(float).eps
+    
+    return (acc * acc_impact) + ((loss_impact / loss) * loss_impact)

pyerualjetwork/plan.py

@@ -84,7 +84,7 @@ def fit(
 
 
 def learner(x_train, y_train, optimizer, fit_start=True, gen=None, batch_size=1, pop_size=None,
-           weight_evolve=False, neural_web_history=False, show_current_activations=False, auto_normalization=False,
+           weight_evolve=True, neural_web_history=False, show_current_activations=False, auto_normalization=False,
            neurons_history=False, early_stop=False, show_history=False, target_loss=None,
            interval=33.33, target_acc=None, loss='categorical_crossentropy', acc_impact=0.9, loss_impact=0.1,
            start_this_act=None, start_this_W=None, dtype=np.float32):
@@ -120,7 +120,7 @@ def learner(x_train, y_train, optimizer, fit_start=True, gen=None, batch_size=1,
                              interval=16.67)
         ```
         
-        weight_evolve (bool, optional): Activation combinations already optimizes by planeat genetic search algorithm. Should the weight parameters also evolve or should the weights be determined according to the aggregating learning principle of the PLAN algorithm? Default: False (PLAN)
+        weight_evolve (bool, optional): Activation combinations already optimizes by PLANEAT genetic search algorithm. Should the weight parameters also evolve or should the weights be determined according to the aggregating learning principle of the PLAN algorithm? Default: True (Evolves Weights)
         
         loss (str, optional): options: ('categorical_crossentropy' or 'binary_crossentropy') Default is 'categorical_crossentropy'.
 
@@ -136,7 +136,7 @@ def learner(x_train, y_train, optimizer, fit_start=True, gen=None, batch_size=1,
 
         gen (int, optional): The generation count for genetic optimization.
 
-        batch_size (float, optional): Batch size is used in the prediction process to receive train feedback by dividing the test data into chunks and selecting activations based on randomly chosen partitions. This process reduces computational cost and time while still covering the entire test set due to random selection, so it doesn't significantly impact accuracy. For example, a batch size of 0.08 means each train batch represents 8% of the train set. Default is 1. (%100 of train)
+        batch_size (float, optional): Batch size is used in the prediction process to receive train feedback by dividing the train data into chunks and selecting activations based on randomly chosen partitions. This process reduces computational cost and time while still covering the entire train set due to random selection, so it doesn't significantly impact accuracy. For example, a batch size of 0.08 means each train batch represents %8 of the train set. Default is 1. (%100 of train)
 
         pop_size (int, optional): Population size of each generation. Default: count of activation functions
 

pyerualjetwork/plan_cuda.py

@@ -83,7 +83,7 @@ def fit(
 
 
 def learner(x_train, y_train, optimizer, fit_start=True, gen=None, batch_size=1, pop_size=None,
-           weight_evolve=False, neural_web_history=False, show_current_activations=False, auto_normalization=False, target_acc=None,
+           weight_evolve=True, neural_web_history=False, show_current_activations=False, auto_normalization=False, target_acc=None,
            neurons_history=False, early_stop=False, show_history=False, loss='categorical_crossentropy',
            interval=33.33, target_loss=None, loss_impact=0.1, acc_impact=0.9,
            start_this_act=None, start_this_W=None, dtype=cp.float32, memory='gpu'):
@@ -119,7 +119,7 @@ def learner(x_train, y_train, optimizer, fit_start=True, gen=None, batch_size=1,
                                 interval=16.67)
         ```
         
-        weight_evolve (bool, optional): Activation combinations already optimizes by planeat genetic search algorithm. Should the weight parameters also evolve or should the weights be determined according to the aggregating learning principle of the PLAN algorithm? Default: False (PLAN)
+        weight_evolve (bool, optional): Activation combinations already optimizes by PLANEAT genetic search algorithm. Should the weight parameters also evolve or should the weights be determined according to the aggregating learning principle of the PLAN algorithm? Default: True (Evolves Weights)
 
         loss (str, optional): For visualizing and monitoring. PLAN neural networks doesn't need any loss function in training. options: ('categorical_crossentropy' or 'binary_crossentropy') Default is 'categorical_crossentropy'.
 
@@ -131,7 +131,7 @@ def learner(x_train, y_train, optimizer, fit_start=True, gen=None, batch_size=1,
         
         gen (int, optional): The generation count for genetic optimization.
 
-        batch_size (float, optional): Batch size is used in the prediction process to receive train feedback by dividing the train data into chunks and selecting activations based on randomly chosen partitions. This process reduces computational cost and time while still covering the entire test set due to random selection, so it doesn't significantly impact accuracy. For example, a batch size of 0.08 means each train batch represents 8% of the train set. Default is 1. (%100 of train)
+        batch_size (float, optional): Batch size is used in the prediction process to receive train feedback by dividing the train data into chunks and selecting activations based on randomly chosen partitions. This process reduces computational cost and time while still covering the entire train set due to random selection, so it doesn't significantly impact accuracy. For example, a batch size of 0.08 means each train batch represents %8 of the train set. Default is 1. (%100 of train)
 
         acc_impact (float, optional): Impact of accuracy for optimization [0-1]. Default: 0.9
 

pyerualjetwork/planeat.py

@@ -17,7 +17,7 @@ import random
 import math
 
 ### LIBRARY IMPORTS ###
-from .data_operations import normalization
+from .data_operations import normalization, non_neg_normalization
 from .ui import loading_bars, initialize_loading_bar
 from .activation_functions import apply_activation, all_activations
 
@@ -291,7 +291,7 @@ def evolver(weights,
    bar_format = loading_bars()[0]
    
    if bar_status: progress = initialize_loading_bar(len(bad_weights), desc="GENERATION: " + str(what_gen), bar_format=bar_format, ncols=50)
-   normalized_fitness = normalization(fitness, dtype=dtype)
+   normalized_fitness = non_neg_normalization(fitness, dtype=dtype)
 
    best_fitness = normalized_fitness[-1]
    epsilon = np.finfo(float).eps
@@ -708,28 +708,18 @@ def mutation(weight,
 
       if potential_weight_mutation > weight_mutate_prob:
 
-         start = 0
-         row_end = weight.shape[0]
-         col_end = weight.shape[1]
-
-         max_threshold = row_end * col_end
-
-         threshold = weight_mutate_threshold * (genome_fitness + epsilon)
-         new_threshold = threshold
-         
-         for _ in range(max_threshold):
+            row_end, col_end = weight.shape
+            max_threshold = row_end * col_end
+            threshold = weight_mutate_threshold / (genome_fitness + epsilon)
+   
+            n_mutations = min(int(threshold), max_threshold)
+                     
+            row_indices = np.random.randint(0, row_end, size=n_mutations)
+            col_indices = np.random.randint(0, col_end, size=n_mutations)
             
-            selected_row = int(random.uniform(start, row_end))
-            selected_col = int(random.uniform(start, col_end))
-
-            weight[selected_row, selected_col] = random.uniform(-1, 1)
-            new_threshold += threshold
-
-            if max_threshold > new_threshold:
-               pass
+            new_values = np.random.uniform(-1, 1, size=n_mutations)
             
-            else:
-               break
+            weight[row_indices, col_indices] = new_values
 
    activation_mutate_prob = 1 - activation_mutate_prob
    potential_activation_mutation = random.uniform(0, 1)

pyerualjetwork/planeat_cuda.py

@@ -19,7 +19,7 @@ import math
 
 
 ### LIBRARY IMPORTS ###
-from .data_operations_cuda import normalization
+from .data_operations_cuda import normalization, non_neg_normalization
 from .ui import loading_bars, initialize_loading_bar
 from .activation_functions_cuda import apply_activation, all_activations
 
@@ -292,7 +292,7 @@ def evolver(weights,
    bar_format = loading_bars()[0]
    
    if bar_status: progress = initialize_loading_bar(len(bad_weights), desc="GENERATION: " + str(what_gen), bar_format=bar_format, ncols=50)
-   normalized_fitness = normalization(fitness, dtype=dtype)
+   normalized_fitness = non_neg_normalization(fitness, dtype=dtype)
 
    best_fitness = normalized_fitness[-1]
    epsilon = cp.finfo(float).eps
@@ -709,28 +709,18 @@ def mutation(weight,
 
       if potential_weight_mutation > weight_mutate_prob:
 
-         start = 0
-         row_end = weight.shape[0]
-         col_end = weight.shape[1]
-         
-         max_threshold = row_end * col_end
-
-         threshold = weight_mutate_threshold * (genome_fitness + epsilon)
-         new_threshold = threshold
-
-         for _ in range(max_threshold):
+            row_end, col_end = weight.shape
+            max_threshold = row_end * col_end
+            threshold = weight_mutate_threshold / (genome_fitness + epsilon)
+   
+            n_mutations = min(int(threshold), max_threshold)
+                     
+            row_indices = cp.random.randint(0, row_end, size=n_mutations)
+            col_indices = cp.random.randint(0, col_end, size=n_mutations)
             
-            selected_row = int(random.uniform(start, row_end))
-            selected_col = int(random.uniform(start, col_end))
-
-            weight[selected_row, selected_col] = random.uniform(-1, 1)
-            new_threshold += threshold
-
-            if max_threshold > new_threshold:
-               pass
+            new_values = cp.random.uniform(-1, 1, size=n_mutations)
             
-            else:
-               break
+            weight[row_indices, col_indices] = new_values
 
    activation_mutate_prob = 1 - activation_mutate_prob
    potential_activation_mutation = random.uniform(0, 1)

{pyerualjetwork-4.3.13.dev0.dist-info → pyerualjetwork-4.5.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: pyerualjetwork
-Version: 4.3.13.dev0
+Version: 4.5
 Summary: PyerualJetwork is a machine learning library supported with GPU(CUDA) acceleration written in Python for professionals and researchers including with PLAN algorithm, PLANEAT algorithm (genetic optimization). Also includes data pre-process and memory manegament
 Author: Hasan Can Beydili
 Author-email: tchasancan@gmail.com

{pyerualjetwork-4.3.13.dev0.dist-info → pyerualjetwork-4.5.dist-info}/RECORD

@@ -1,9 +1,9 @@
-pyerualjetwork/__init__.py,sha256=Ss4ruAB1dEBCNP5cIy5VexRdmSObtZbaoQ7iprxifcQ,1284
+pyerualjetwork/__init__.py,sha256=xOgL47nXk4gItE1UKQ_uEBevdRI2RUjN5RVuB-BRlHM,1277
 pyerualjetwork/activation_functions.py,sha256=Ms0AGBqkJuCA42ht64MSQnO54Td_1eDGquedpoBDVbc,7642
 pyerualjetwork/activation_functions_cuda.py,sha256=5y1Ti3GDfDteQDCUmODwe7tAyDAUlDTKmIikChQ8d6g,7772
-pyerualjetwork/data_operations.py,sha256=Flteouu6rfSo2uHMqBHuzO02dXmbNa-I5qWmUpGTZ5Y,14760
-pyerualjetwork/data_operations_cuda.py,sha256=ZcjmLXE1-HVwedextYdJZ1rgrns1OfSekzFpr1a9m6o,17625
-pyerualjetwork/fitness_functions.py,sha256=9ZtCasBVjk3qQldi2pIzV6c5hTBJJyGCU0eVhJsoxPo,1145
+pyerualjetwork/data_operations.py,sha256=c2FWIdHJQw_h6QP_ir0yx22zP7a9CtRp249V8Ro9V-0,15357
+pyerualjetwork/data_operations_cuda.py,sha256=IqS0JoXGM0XiYfoFSAj9li1WWiBroNXIcN52JWhlNFk,18224
+pyerualjetwork/fitness_functions.py,sha256=GisM8mDJAivw8YailXXDAw2M-lW1MHwRnIlWUVe-UEg,1261
 pyerualjetwork/help.py,sha256=nQ_YbYA2RtuafhuvkreNpX0WWL1I_nzlelwCtvei0_Y,775
 pyerualjetwork/loss_functions.py,sha256=6PyBI232SQRGuFnG3LDGvnv_PUdWzT2_2mUODJiejGI,618
 pyerualjetwork/loss_functions_cuda.py,sha256=C93IZJcrOpT6HMK9x1O4AHJWXYTkN5WZiqdssPbvAPk,617
@@ -12,14 +12,14 @@ pyerualjetwork/metrics.py,sha256=q7MkhnZDRbCjFBDDfUgrl8lBYnUT_1ro1LxeBq105pI,607
 pyerualjetwork/metrics_cuda.py,sha256=73h9GC7XwmnFCVzFEEiPQfF8CwHIz2wsCbxpZrJtYgw,5061
 pyerualjetwork/model_operations.py,sha256=MCSCNYiiICRVZITobtS3ZIWmH5Q9gjyELuH32sAdgg4,12649
 pyerualjetwork/model_operations_cuda.py,sha256=NT01BK5nrDYE7H1x3KnSI8gmx0QTGGB0mP_LqEb1uuU,13157
-pyerualjetwork/plan.py,sha256=PXZhlty36nBLwyy7QAQDmdczGxvmvPc1h16dbbcTH60,22964
-pyerualjetwork/plan_cuda.py,sha256=CaHnRmP3hcg_Eh07KrCYmaSz5B1OdKVCpB4gedLNMG0,23956
-pyerualjetwork/planeat.py,sha256=oAu0a2rs6BQ2HUpLjKFR4m7xDbhUVOzLstop75u35Tg,38889
-pyerualjetwork/planeat_cuda.py,sha256=E_pJEsNKfiF97FvGZGSsxmdqzLAJub05enmneymozcg,38922
+pyerualjetwork/plan.py,sha256=XEptIEWfWbqXEy92Eil5q_uujEiIBWnjejAGD7Lh0-w,22975
+pyerualjetwork/plan_cuda.py,sha256=rFHZa8jsgfol0uUo1rfCKAGtb-B4ivfzqqlnynsfMzQ,23966
+pyerualjetwork/planeat.py,sha256=DVJGptIPYKyz4ePwqRnbbwgHwQzbxMoBci0Te8kfCzk,38802
+pyerualjetwork/planeat_cuda.py,sha256=-GxZY8aMPayuYhwhcsRVn4LWq604o36VxkEtdoBum98,38835
 pyerualjetwork/ui.py,sha256=wu2BhU1k-w3Kcho5Jtq4SEKe68ftaUeRGneUOSCVDjU,575
 pyerualjetwork/visualizations.py,sha256=1QSisYAaGnO5kug6qo1qOssTkQM-MgR7L8h4c3sczzU,28294
 pyerualjetwork/visualizations_cuda.py,sha256=PYRqj4QYUbuYMYcNwO8yaTPB-jK7E6kZHhTrAi0lwPU,28749
-pyerualjetwork-4.3.13.dev0.dist-info/METADATA,sha256=BZ_-V95KQ1SHkVVJBWlzTj3oaDXFM-lGnozIxy3dkv8,7511
-pyerualjetwork-4.3.13.dev0.dist-info/WHEEL,sha256=2wepM1nk4DS4eFpYrW1TTqPcoGNfHhhO_i5m4cOimbo,92
-pyerualjetwork-4.3.13.dev0.dist-info/top_level.txt,sha256=BRyt62U_r3ZmJpj-wXNOoA345Bzamrj6RbaWsyW4tRg,15
-pyerualjetwork-4.3.13.dev0.dist-info/RECORD,,
+pyerualjetwork-4.5.dist-info/METADATA,sha256=njuvH-FdU7-KudHvxUEJvBeyOvDWzyWi52ZIpRjR9K4,7503
+pyerualjetwork-4.5.dist-info/WHEEL,sha256=2wepM1nk4DS4eFpYrW1TTqPcoGNfHhhO_i5m4cOimbo,92
+pyerualjetwork-4.5.dist-info/top_level.txt,sha256=BRyt62U_r3ZmJpj-wXNOoA345Bzamrj6RbaWsyW4tRg,15
+pyerualjetwork-4.5.dist-info/RECORD,,