Summary  
This chapter covers implementing complete forward propagation in a multi-layer perceptron by iteratively passing inputs through each layer’s forward method to produce the final outputs.

General domain of usage  
Machine learning model inference

You have already implemented forward propagation for a single layer in the previous chapter. Now, the goal is to implement **complete forward propagation**, from inputs to outputs. 

To implement the entire forward propagation process, you need to define the `forward()` method in the `Perceptron` class. This method performs forward propagation **layer by layer** by calling the respective method for each layer:

```python
class Perceptron:
    def __init__(self, layers):
        self.layers = layers

    def forward(self, inputs):
        x = inputs
        for layer in ...:
            # Pass x layer by layer
            x = ...

        return ...
```

The inputs pass through the first hidden layer, with **each layer's outputs serving as inputs for the next**, until reaching the final layer to produce the final output.

import unittest
import importlib
import numpy as np


def _dynamic_test(test_case, condition, success_msg, failure_msg):
    if condition:
        test_case._testMethodName = success_msg
        test_case.assertTrue(True, success_msg)
    else:
        test_case._testMethodName = failure_msg
        test_case.fail(failure_msg)


class TestPerceptronForward(unittest.TestCase):

    @classmethod
    def setUpClass(cls):
        # Import user code
        cls.user_code = importlib.import_module("user_code")
        np.random.seed(10)

        # Build perceptron as in the task
        cls.perceptron = cls.user_code.perceptron
        cls.inputs = [1, 0]

        # Compute user output
        cls.user_output = cls.perceptron.forward(cls.inputs)

        # Reference forward computation (deterministic)
        # Rebuild layers with same seed and logic from layers.py
        from layers import hidden_1, hidden_2, output_layer

        def forward_reference(inputs):
            x = inputs
            for layer in [hidden_1, hidden_2, output_layer]:
                x = layer.forward(x)
            return x

        cls.ref_output = forward_reference(cls.inputs)

    def test_output_type(self):
        _dynamic_test(
            self,
            isinstance(self.user_output, np.ndarray),
            "Output type is NumPy array",
            "Output type is incorrect (must be NumPy array)",
        )

    def test_output_shape(self):
        _dynamic_test(
            self,
            self.user_output.shape == (1, 1),
            "Output shape is correct (1, 1)",
            f"Output shape is incorrect: got {self.user_output.shape}",
        )

    def test_forward_result_value(self):
        _dynamic_test(
            self,
            np.allclose(self.user_output, self.ref_output, atol=1e-6),
            "Forward propagation output is correct",
            f"Forward propagation output is incorrect: got {self.user_output}, expected {self.ref_output}",
        )

    def test_output_range(self):
        val = float(self.user_output[0, 0])
        _dynamic_test(
            self,
            0 <= val <= 1,
            "Output is within (0, 1) as expected for sigmoid activation",
            f"Output value {val} is outside expected sigmoid range (0,1)",
        )

    def test_determinism(self):
        # Recompute again to ensure reproducibility
        np.random.seed(10)
        out2 = self.user_code.perceptron.forward(self.inputs)
        _dynamic_test(
            self,
            np.allclose(out2, self.user_output, atol=1e-6),
            "Forward pass is deterministic",
            "Forward pass produces inconsistent results between runs",
        )


if __name__ == "__main__":
    unittest.main(argv=["first-arg-is-ignored"], exit=False)


test_code.py

Neural networks are powerful algorithms inspired by the structure of the human brain that are used to solve complex machine learning problems. You will build your own Neural Network from scratch to understand how it works. After this course, you will be able to create neural networks for solving classification and regression problems using the scikit-learn library.

First, we will discuss what a neural network is and how it works. And also consider the scope of its application.

Next, we will try to build our own neural network and see how efficiently it copes with learning. We will also consider a ready-made solution from the scikit-learn library.

Finally, we will give you some additional useful information on how to understand which model to use and what types of neural networks there are. To complete the course, you will be tested on your acquired knowledge.

Forward Propagation

解答