Soft Computing

Course Description

Soft Computing is a collection of computational techniques designed to solve complex problems where exact mathematical models may be difficult to obtain.

This course covers:

Artificial Neural Networks
Backpropagation
Fuzzy Logic
Fuzzy Controllers
Genetic Algorithms

Unit I: Neural Networks – I

1. Biological Neuron

A biological neuron consists of:

Dendrites
Cell body
Axon
Synapse

The neuron receives signals through dendrites, processes them in the cell body, and transmits signals through the axon.

2. Artificial Neuron

An artificial neuron calculates a weighted sum of its inputs:

z = \sum_{i=1}^{n} w_i x_i + b

The output is then calculated using an activation function:

y = f(z)

Where:

Symbol	Meaning
$x_i$	Input
$w_i$	Weight
$b$	Bias
$z$	Weighted sum
$f$	Activation function
$y$	Output

3. Activation Functions

Step Function

f(x) = \begin{cases} 1, & x \geq 0 \\ 0, & x < 0 \end{cases}

Sigmoid

\sigma(x) = \frac{1}{1 + e^{-x}}

ReLU

ReLU(x) = \max(0,x)

Tanh

\tanh(x) = \frac{e^x-e^{-x}} {e^x+e^{-x}}

Unit II: Backpropagation Networks

1. Forward Propagation

For a neural network layer:

z = Wx+b

The activation is:

a=f(z)

The output of one layer becomes the input of the next layer.

2. Loss Function

For Mean Squared Error:

MSE = \frac{1}{n} \sum_{i=1}^{n} (y_i-\hat{y}_i)^2

3. Backpropagation

Backpropagation calculates gradients using the chain rule.

\frac{\partial L}{\partial w} = \frac{\partial L}{\partial \hat{y}} \frac{\partial \hat{y}}{\partial z} \frac{\partial z}{\partial w}

The weight is updated using gradient descent:

w_{new} = w_{old} - \eta \frac{\partial L}{\partial w}

4. PyTorch Implementation

import torch
import torch.nn as nn

model = nn.Linear(3, 1)

x = torch.tensor([[2., 3., 4.]])
target = torch.tensor([[10.]])

prediction = model(x)

loss_fn = nn.MSELoss()
loss = loss_fn(prediction, target)

loss.backward()

print("Prediction:", prediction)
print("Loss:", loss)
print("Weight gradient:", model.weight.grad)
print("Bias gradient:", model.bias.grad)

The PyTorch implementation demonstrates the same mathematical process:

x \rightarrow Wx+b \rightarrow \hat{y} \rightarrow Loss \rightarrow \frac{\partial L}{\partial W}

Unit III: Fuzzy Logic – I

Fuzzy Sets

In a crisp set:

\mu_A(x) \in {0,1}

In a fuzzy set:

0 \leq \mu_A(x) \leq 1

Fuzzy Union

\mu_{A\cup B}(x) ================ \max(\mu_A(x),\mu_B(x))

Fuzzy Intersection

\mu_{A\cap B}(x) ================ \min(\mu_A(x),\mu_B(x))

Fuzzy Complement

\mu_{\bar A}(x) =============== 1-\mu_A(x)

Unit IV: Fuzzy Logic – II

Membership Functions

A membership function determines the degree to which an element belongs to a fuzzy set.

\mu_A(x):X\rightarrow[0,1]

Common membership functions include:

Triangular
Trapezoidal
Gaussian
Sigmoid

Gaussian Membership Function

\mu(x) ====== e^{-\frac{(x-c)^2}{2\sigma^2}}

Fuzzy Rules

A fuzzy rule has the form:

IF temperature is HIGH THEN fan speed is FAST.

The general structure is:

IF\ A\ THEN\ B

Fuzzification

Fuzzification converts a crisp input into fuzzy membership values.

Defuzzification

The centroid method is:

x^* === \frac{ \sum_i x_i\mu(x_i) }{ \sum_i\mu(x_i) }

Unit V: Genetic Algorithms

Genetic Algorithm

A Genetic Algorithm is an optimization technique inspired by natural selection and biological evolution.

The main components are:

Population
Chromosome
Gene
Fitness
Selection
Crossover
Mutation

Selection

Selection chooses suitable chromosomes for reproduction.

For fitness-proportional selection:

P_i = \frac{Fitness_i} {\sum_j Fitness_j}

Crossover

Crossover combines genetic information from two parent chromosomes.

Example:

Parent 1: 101|101
Parent 2: 011|010

Child 1: 101|010
Child 2: 011|101

Mutation

Mutation randomly changes one or more genes.

Before: 101101
After:  101001

Genetic Algorithm Flow

Initialize Population
        ↓
Calculate Fitness
        ↓
Selection
        ↓
Crossover
        ↓
Mutation
        ↓
New Population
        ↓
Termination Check
        ↓
Repeat

Soft Computing: Neural Networks vs Fuzzy Logic vs GA

Technique	Main Purpose	Inspiration
Neural Networks	Learning	Biological brain
Fuzzy Logic	Reasoning	Human reasoning
Genetic Algorithms	Optimization	Natural evolution

Course Outcome

After completing this course, students will be able to:

Explain artificial neural networks.
Understand perceptron learning.
Understand activation functions.
Implement basic neural networks using PyTorch.
Explain backpropagation mathematically.
Understand fuzzy sets and fuzzy relations.
Design fuzzy membership functions.
Apply fuzzification and defuzzification.
Understand fuzzy controllers.
Explain Genetic Algorithm operations.
Implement selection, crossover and mutation.
Apply Soft Computing techniques to real-world problems.


### One important point

For your renderer, **don't use the HTML `<article>` inside `category.description`**. Store the course content as Markdown in your database and let `ReactMarkdown` render it.

Your current combination:

```jsx
remarkPlugins={[remarkGfm, remarkMath]}
rehypePlugins={[rehypeKatex]}

is particularly suitable for this course because the syllabus contains a lot of mathematical notation.

For the final Tech3Space version, I recommend making each unit a separate Markdown section, with formulas + worked numerical examples + PyTorch/Python practicals + exam questions. This will make the page much more useful than a purely theoretical course.

Soft Computing

Description

Soft Computing

Course Description

Unit I: Neural Networks – I

1. Biological Neuron

2. Artificial Neuron

3. Activation Functions

Step Function

Sigmoid

ReLU

Tanh

Unit II: Backpropagation Networks

1. Forward Propagation

2. Loss Function

3. Backpropagation

4. PyTorch Implementation

Unit III: Fuzzy Logic – I

Fuzzy Sets

Fuzzy Union

Fuzzy Intersection

Fuzzy Complement

Unit IV: Fuzzy Logic – II

Membership Functions

Gaussian Membership Function

Fuzzy Rules

Fuzzification

Defuzzification

Unit V: Genetic Algorithms

Genetic Algorithm

Selection

Crossover

Mutation

Genetic Algorithm Flow

Soft Computing: Neural Networks vs Fuzzy Logic vs GA

Course Outcome

Instructor

SEO Information

Formulas

Transformers deeply