Introduction to Deep Learning and Neural Networks

Business Scenario

Welcome!

You are an AI/ML Engineer on the SmartCart AI team at NextCart Technologies. The catalog team currently uses a simple Logistic Regression model to classify products into three categories: Electronics, Grocery and Apparel

However, the model has reached a limit in its accuracy because the product attributes often overlap, making the categories difficult to separate using simple decision boundaries.

Your main task is to Build a Perceptron and a basic Artificial Neural Network (ANN) from scratch to classify SmartCart products and understand“Why was a neural network needed instead of a traditional ML model?”

Pre-Lab Preparation

Topic: Deep Learning and Neural Networks

1) What is Deep Learning, and how does it differ from traditional Machine Learning

2) The Perceptron

3) Artificial Neural Network (ANN) architecture

4) Activation functions

Git Pull

git pull origin branchName

Deep Learning vs. Traditional Machine Learning

Deep Learning is a subfield of Machine Learning that uses Artificial Neural Networks with multiple layers to automatically learn patterns and representations directly from data, rather than relying on manually engineered features.

Traditional ML vs. Deep Learning

Setup

1

Task 1: Data Loading & Pre-processing

import pandas as pd
import numpy as np
import matplotlib.pyplot as plt
import seaborn as sns

Load and Inspect the SmartCart Dataset

2

Dataset : 

df = pd.read_csv('smartcart_dataset.csv')

print("Shape:", df.shape)
print(df["category"].value_counts())
df.head()

Output

Exploratory Visualization

3

sns.scatterplot(data=df, x='price', y='weight_kg', hue='category')

Preprocessing

4

from sklearn.model_selection import train_test_split
from sklearn.preprocessing import StandardScaler, LabelEncoder

FEATURES = ['price', 'weight_kg', 'rating', 'discount_pct', 'description_length']
TARGET = 'category'

X = df[FEATURES]
y_raw = df[TARGET]
le = LabelEncoder()
y = le.fit_transform(y_raw)
class_names = le.classes_
print(dict(zip(range(len(class_names)), class_names)))
X_train, X_test, y_train, y_test = train_test_split(
    X, y, test_size=0.2, random_state=42, stratify=y
)

scaler = StandardScaler()
X_train = scaler.fit_transform(X_train)
X_test = scaler.transform(X_test)

X_train.shape, X_test.shape

Output :

Output :

Perceptron

Perceptron — the simplest unit of a neural network.

It takes one or more numerical inputs, multiplies each by a weight, adds a bias and passes the result through an activation function to produce a binary output. It is the neural equivalent of a simple linear classifier.

Import Keras and Build the Model

1

Task 2 : Build a Perceptron

from tensorflow import keras
from keras import Sequential
from keras.layers import Dense

perceptron = Sequential()
perceptron.add(Dense(3, activation='softmax', input_dim=X_train.shape[1]))

perceptron.summary()

Output

Compile and Train

2

perceptron.compile(optimizer='adam',
                    loss='sparse_categorical_crossentropy',
                    metrics=['accuracy'])
history_p = perceptron.fit(X_train, y_train, epochs=100,
                            validation_split=0.2, verbose=0)                    
weights, bias = perceptron.layers[0].get_weights()
print('weights:\n', weights)
print('bias:\n', bias)

Inspect the Learned Weights and Bias

3

Output

Evaluate and Visualize

4

perceptron_loss, perceptron_acc = perceptron.evaluate(X_test, y_test, verbose=0)
print(f'Perceptron test accuracy: {perceptron_acc:.3f}')

Output :

plt.plot(history_p.history['loss'], label='Training Loss')
plt.plot(history_p.history['val_loss'], label='Validation Loss')
plt.title('Perceptron training loss')
plt.legend()
plt.show()

Aritificial Neural Network

ANN Architecture — an Artificial Neural Network is built by arranging many Perceptron-like units (“neurons”) into layers, stacked so the output of one layer feeds into the next.

Build the Model

1

Task 3 : Build a basic ANN and Test it on the SmartCart dataset

ann = Sequential()
ann.add(Dense(32, activation='relu', input_dim=X_train.shape[1]))
ann.add(Dense(16, activation='relu'))
ann.add(Dense(3, activation='softmax'))

ann.summary()

Output

Compile, Train and Evaluate

2

ann.compile(optimizer='adam',
            loss='sparse_categorical_crossentropy',
            metrics=['accuracy'])
history_ann = ann.fit(X_train, y_train, epochs=100,
                       validation_split=0.2, verbose=0)
ann_loss, ann_acc = ann.evaluate(X_test, y_test, verbose=0)
print(f'ANN test accuracy: {ann_acc:.3f}')
plt.plot(history_ann.history['loss'], 
label='Training Loss')
plt.plot(history_ann.history['val_loss'], 
label='Validation Loss')
plt.title('ANN training loss')
plt.legend()
plt.show()

Output :

Plot Training Loss

3

Task 4 : Why This Problem Needed a Neural Network

You've built a Perceptron and an ANN. Now use Logistic Regression, the traditional ML model used in the original scenario, to confirm that it also has a linear decision-boundary limitation.

Run Logistic Regression on the SmartCart Dataset

1

log_reg = LogisticRegression(max_iter=2000)
log_reg.fit(X_train_scaled, y_train)
y_pred_logreg = log_reg.predict(X_test_scaled)
logreg_acc = accuracy_score(y_test, y_pred_logreg)

print(f"Logistic Regression test accuracy: {logreg_acc:.3f}\n")
print(classification_report(y_test, y_pred_logreg, target_names=class_names))

Confusion Matrix for All Three Models 

2

fig, axes = plt.subplots(1, 3, figsize=(16, 5))

for ax, y_pred, title, cmap in zip(
        axes,
        [y_pred_perceptron, y_pred_ann, y_pred_logreg],
        ['Perceptron -- Confusion Matrix', 'ANN -- Confusion Matrix', 
        'Logistic Regression -- Confusion Matrix'],
        ['Blues', 'Greens', 'Oranges']):
    cm = confusion_matrix(y_test, y_pred)
    im = ax.imshow(cm, cmap=cmap)
    ax.set_xticks(range(len(class_names))); ax.set_xticklabels(class_names)
    ax.set_yticks(range(len(class_names))); ax.set_yticklabels(class_names)
    ax.set_xlabel('Predicted'); ax.set_ylabel('Actual')
    ax.set_title(title)
    for i in range(len(class_names)):
        for j in range(len(class_names)):
            ax.text(j, i, cm[i, j], ha='center', va='center')

plt.tight_layout()
plt.show()

Output :

Compare all 3 Models

3

comparison = pd.DataFrame({
    'Model': ['Perceptron (Keras)', 'Logistic Regression (sklearn)',
              'ANN (Keras)'],
    'Test Accuracy': [perceptron_acc,logreg_acc ,ann_acc ]})
comparison

 

Great job!

You have successfully compared Deep Learning with traditional Machine Learning, built and evaluated a Perceptron and a basic ANN in Keras using SmartCart data, and explored the key building blocks of neural networks. You compared both models with a Logistic Regression baseline and investigated why this problem needed a neural network, using accuracy scores, confusion matrices, and decision-boundary plots as evidence.

Checkpoint

   Git Push

git push origin branchName

Next-Lab Preparation

Topic :  Backpropagation and Gradient Descent

1) Loss Function
2) Gradients, Gradient Descent
3) Backpropagation Flow
4) Epochs & Convergence

GenAI - 1

By Content ITV

GenAI - 1

  • 113

More from Content ITV

  • Copy of Copy of Copy of Vertical infinite scroll OG - Template
    Content ITV
    11
  • Demonstrate Understanding of SIEM Architecture Through Real Mapping and Visualization
    Content ITV
    61
  • Perform Real-Time Security Monitoring and Detect Suspicious Activities
    Content ITV
    60
  • SOC Analyst Responsibilities by Simulating Incident Detection and Response (v2)
    Content ITV
    65