实现自定义Dense层_牛客题霸_牛客网

实现一个自定义的Dense（全连接）层类。该类需要继承基础Layer类，并实现神经网络中全连接层的所有基本功能。

输入描述:

Dense类需要实现以下方法：1. `__init__(self, n_units, input_shape=None)`：- n_units：输出神经元数量- input_shape：输入形状（可选）2. `initialize(self, optimizer)`：- 初始化权重W（使用1sqrt(input_shape[0])作为范围进行正负区间的均匀分布采样）- 初始化偏置w0为零- 设置优化器3. `parameters(self)`：- 返回层中可训练参数总数4. `forward_pass(self, X, training=True)`：- 执行前向传播计算- 返回 5. `backward_pass(self, accum_grad)`：- 计算并返回梯度- 使用优化器更新参数6. `output_shape(self)`：- 返回输出形状(n_units,)

输出描述:

每个方法都需要返回特定的值：- initialize：无返回值- parameters：返回整数- forward_pass：返回numpy数组- backward_pass：返回numpy数组- output_shape：返回元组

示例1

输入

[[1, 2]]

输出

Forward pass output: [[ 0.10162127 -0.33551992 -0.64490545]]
Backward pass output: [[ 0.20816524 -0.22928937]]

备注:

1.对应的输入、输出已给出，您只用实现核心功能函数即可。2.支持numpy、scipy、pandas、scikit-learn库。

加载中...

import numpy as np
import copy
import math

# DO NOT CHANGE SEED
np.random.seed(42)

# DO NOT CHANGE LAYER CLASS
class Layer(object):

def set_input_shape(self, shape):
    
		self.input_shape = shape

def layer_name(self):
		return self.__class__.__name__

def parameters(self):
		return 0

def forward_pass(self, X, training):
		raise NotImplementedError()

def backward_pass(self, accum_grad):
		raise NotImplementedError()

def output_shape(self):
		raise NotImplementedError()

# Your task is to implement the Dense class based on the above structure
class Dense(Layer):
	def __init__(self, n_units, input_shape=None):
		self.layer_input = None
		self.input_shape = input_shape
		self.n_units = n_units
		self.trainable = True
		self.W = None
		self.w0 = None

def initialize(self, optimizer):
		pass

def forward_pass():
		pass

def backward_pass(self, accum_grad):
		pass

def number_of_parameters():
		pass

if __name__ == "__main__":
    # Initialize a Dense layer with 3 neurons and input shape (2,)
    dense_layer = Dense(n_units=3, input_shape=(2,))

# Define a mock optimizer with a simple update rule
    class MockOptimizer:
        def update(self, weights, grad):
            return weights - 0.01 * grad

optimizer = MockOptimizer()

# Initialize the Dense layer with the mock optimizer
    dense_layer.initialize(optimizer)

# Perform a forward pass with sample input data
    X = np.array(eval(input()))
    output = dense_layer.forward_pass(X)
    print("Forward pass output:", output)

accum_grad = np.array([[0.1, 0.2, 0.3]]) 
    back_output = dense_layer.backward_pass(accum_grad) 
    print("Backward pass output:", back_output)