Create main.py

Author: NNEngine

Weight Initialization/main.py

@@ -0,0 +1,127 @@
+import random
+import math
+
+while True:
+	try:
+		rows:int = int(input("Enter number of rows: "))
+		break
+	except ValueError:
+		print("Enter an integer Value")
+
+
+while True:
+	try:
+		cols:int = int(input("Enter number of columns: "))
+		break
+	except ValueError:
+		print("Enter an integer Value")
+
+def zero_initialization(rows:int, cols:int)->list:
+	'''
+		This functions generates a matrix whose all values are zero for weight initialization.
+	'''
+
+	matrix:List = []
+
+	for i in range(rows):
+		row:List = []
+		for j in range(cols):
+			row.append(0.0)
+		matrix.append(row)
+	return matrix
+
+print(zero_initialization(rows, cols))
+
+def random_uniform_initialization(rows:int, cols:int)->list:
+	'''
+		This function generates a matrix whose elements are drawn from unifrom distribution.
+	'''
+	matrix:List = []
+
+	for i in range(rows):
+		row:List = []
+		for j in range(cols):
+			row.append(random.uniform(-1.0, 1.0))
+		matrix.append(row)
+	return matrix
+
+print(random_uniform_initialization(rows, cols))
+
+
+def random_normal_initialization(rows:int, cols:int)->list:
+	'''
+		This function generates a matrix whose elements are drawn from normal distribution.
+	'''
+	matrix:List = []
+
+	for i in range(rows):
+		row:List = []
+		for j in range(cols):
+			row.append(random.gauss(0.0, 1.0))
+		matrix.append(row)
+	return matrix
+
+print(random_normal_initialization(rows, cols))
+
+
+def xavier_uniform_initialization(rows:int, cols:int)->list:
+	'''
+		This function generates a matrix whose elements are drawn from unifrom distribution.
+	'''
+	matrix:List = []
+
+	for i in range(rows):
+		row:List = []
+		for j in range(cols):
+			row.append(random.uniform(-math.sqrt(6 / (rows + cols)), math.sqrt(6 / (rows + cols))))
+		matrix.append(row)
+	return matrix
+
+print(xavier_uniform_initialization(rows, cols))
+
+
+def xavier_normal_initialization(rows:int, cols:int)->list:
+	'''
+		This function generates a matrix whose elements are drawn from normal distribution.
+	'''
+	matrix:List = []
+
+	for i in range(rows):
+		row:List = []
+		for j in range(cols):
+			row.append(random.gauss(0.0, math.sqrt(2 / (rows + cols))))
+		matrix.append(row)
+	return matrix
+
+print(xavier_normal_initialization(rows, cols))
+
+def he_uniform_initialization(rows:int, cols:int)->list:
+	'''
+		This function generates a matrix whose elements are drawn from unifrom distribution.
+	'''
+	matrix:List = []
+
+	for i in range(rows):
+		row:List = []
+		for j in range(cols):
+			row.append(random.uniform(-math.sqrt(2 / (rows + cols)), math.sqrt(2 / (rows + cols))))
+		matrix.append(row)
+	return matrix
+
+
+print(he_uniform_initialization(rows, cols))
+
+def he_normal_initialization(rows:int, cols:int)->list:
+	'''
+		This function generates a matrix whose elements are drawn from normal distribution.
+	'''
+	matrix:List = []
+
+	for i in range(rows):
+		row:List = []
+		for j in range(cols):
+			row.append(random.gauss(0.0, math.sqrt(2 / (rows + cols))))
+		matrix.append(row)
+	return matrix
+
+print(he_normal_initialization(rows, cols))