Experiment No - 01

Title: Linear regression using Deep Neural network: Implement Boston housing price
prediction problem by Linear regression using Deep Neural network. Use the Boston House
price prediction dataset.

Objectives: Students should be able to perform Linear regression by using a Deep Neural
network on the Boston House Dataset.

Software Requirement:

Operating system: Windows, Linux, macOS.

Programming Languages: Python, C++, Java.

Theoretical Concept:

What is Linear Regression? Linear regression is a statistical approach that is commonly used
to model the relationship between a dependent variable and one or more independent variables.
It assumes a linear relationship between variables and uses mathematical methods to estimate
the coefficients that best fit the data. Deep neural networks are a type of machine learning
algorithm that is modeled after the structure and function of the human brain. They consist of
multiple layers of interconnected neurons that process data and learn from it to make
predictions or classifications. Linear regression using deep neural networks combines the
principles of linear regression with the power of deep learning algorithms. In this approach, the
input features are passed through one or more layers of neurons to extract features and then a
linear regression model is applied to the output of the last layer to make predictions. The
weights and biases of the neural network are adjusted during training to optimize the
performance of the model. This approach can be used for a variety of tasks, including predicting
numerical values, such as stock prices or housing prices, and classifying data, such as detecting
whether an image contains a particular object or not. It is often used in fields such as finance,
healthcare, and image recognition.

1|Page
Example Of Linear Regression:

A suitable example of linear regression using a deep neural network would be predicting the
price of a house based on various features such as the size of the house, the number of
bedrooms, the location, and the age of the house. In this example, the input features would be
fed into a deep neural network, consisting of multiple layers of interconnected neurons. The
first few layers of the network would learn to extract features from the input data, such as
identifying patterns and correlations between the input features. The output of the last layer

would then be passed through a linear regression model, which would use the learned features
to predict the price of the house. During training, the weights and biases of the neural network
would be adjusted to minimize the difference between the predicted price and the actual price
of the house. This process is known as gradient descent, and it involves iteratively adjusting
the model's parameters until the optimal values are reached. Once the model is trained, it can
be used to predict the price of a new house based on its features. This approach can be used in
the real estate industry to provide accurate and reliable estimates of house prices, which can
help both buyers and sellers make informed decisions.

A deep neural network is a type of machine learning algorithm that is modeled after the
structure and function of the human brain. It consists of multiple layers of interconnected
nodes, or artificial neurons, that process data and learn from it to make predictions or
classifications. Each layer of the network performs a specific type of processing on the data,
such as identifying patterns or correlations between features, and passes the results to the next
layer. The layers closest to the input are known as the "input layer", while the layers closest to
the output are known as the "output layer". The intermediate layers between the input and
output are known as "hidden layers". These layers are responsible for extracting increasingly
complex features from the input data and can be deep (i.e., containing many hidden layers) or
shallow (i.e., containing only a few hidden layers). Deep neural networks are trained using a
process known as backpropagation, which involves adjusting the weights and biases of the
nodes based on the error between the predicted output and the actual output. This process is
repeated for multiple iterations until the model reaches an optimal level of accuracy. Deep
neural networks are used in a variety of applications, such as image and speech recognition,
natural language processing, and recommendation systems. They are capable of learning from
vast amounts of data. They can automatically extract features from raw data, making them a
powerful tool for solving complex problems in a wide range of domains.

2|Page
How Deep Neural Networks Work

Boston House Price Prediction is a common example used to illustrate how a deep neural
network can work for regression tasks. The goal of this task is to predict the price of a house
in Boston based on various features such as the number of rooms, crime rate, and accessibility
to public transportation. Here's how a deep neural network can work for Boston House Price
Prediction:

1. Data Pre-processing: The first step is preprocessing. This involves normalizing the input
features to have a mean of 0 and a standard deviation of 1, which helps the network learn more
efficiently. The dataset is then split into training and testing sets.

2. Model Architecture: A deep neural network is then defined with multiple layers. The first
layer is the input layer, which takes in the normalized features. This is followed by several
hidden layers, which can be deep or shallow. The last layer is the output layer, which predicts
the house price.

3. Model Training: The model is then trained using the training set. During training, the
weights and biases of the nodes are adjusted based on the error between the predicted output
and the actual output. This is done using an optimization algorithm such as stochastic gradient
descent.

4. Model evaluation: Once the model is trained, it is evaluated using the testing set. The
performance of the model is measured using metrics such as mean squared error or mean
absolute error.

5. Model prediction: Finally, the trained model can be used to make predictions on new data,
such as predicting the price of a new house in Boston based on its features.

By using a deep neural network for Boston House Price Prediction, we can obtain accurate
predictions based on a large set of input features. This approach is scalable and can be used for
other regression tasks as well.

Boston House Price Prediction Dataset

Boston House Price Prediction is a well-known dataset in machine learning and is often used
to demonstrate regression analysis techniques. The dataset contains information about 506
houses in Boston, Massachusetts, USA. The goal is to predict the median value of owner-
occupied homes in thousands of dollars.

3|Page
Implementation:

4|Page
5|Page
Conclusion: In this way, we can Predict the Boston House Price using a Deep Neural
Network.

6|Page
 Experiment No - 02
Title: Binary classification using deep neural networks example: classify movie reviews into
positive" reviews and "negative" reviews, just based on the text content of the reviews. Use the
IMDB dataset.

Objectives: The objective of this experiment is to develop a deep neural network model that
can accurately classify movie reviews as either "positive" or "negative" based solely on the
textual content of the reviews. The model will be trained using the IMDB dataset, which
consists of movie reviews along with their corresponding sentiment labels.

Problem Statement: The problem is to create a deep neural network model that can
accurately classify movie reviews as positive or negative based solely on the textual content of
the reviews. The model will be trained using the IMDb dataset, consisting of labeled movie
reviews. The challenge is to design a model architecture and optimize its performance through
experimentation with hyperparameters. The trained model will be evaluated on a test set to
measure its accuracy and generalization ability. The objective is to develop a reliable and
efficient system for automating sentiment analysis of movie reviews using deep learning
techniques.

Software Requirement:

1. Python: The experiment will be implemented using the Python programming language, so
ensure that Python is installed on your system.

2. TensorFlow/Keras: These deep learning frameworks provide a high-level API for building
and training neural networks. Install TensorFlow or Keras to facilitate model development.

3. NumPy: This library is used for numerical computations and array operations. Install
NumPy to handle numerical data efficiently.

4. IMDB Dataset: Download the IMDB dataset, which contains movie reviews and their
sentiment labels, to train and evaluate the model.

5. Text Preprocessing Libraries: Depending on the specific preprocessing steps required, you
may need to install libraries such as NLTK or spaCy for text cleaning, tokenization, and other
natural language processing tasks.

7|Page
Hardware Requirement:

1. CPU/GPU: A reasonably powerful CPU or GPU is recommended for training deep neural
networks. While a CPU can handle smaller models, training larger models may benefit from a
GPU, which can significantly speed up computations.

2. Memory (RAM): Sufficient memory is required to load and process the dataset, especially
if the dataset is large. The exact memory requirement depends on the dataset size and model
complexity.

3. Storage: Adequate storage space is necessary to store the IMDb dataset and any intermediate
or final model checkpoints, logs, and results.

4. Internet Connection: An internet connection is required to download the IMDb dataset and
any necessary software libraries or dependencies.

5. Operating System: The experiment can be conducted on various operating systems such as
Windows, macOS, or Linux, depending on your preference and the compatibility of the
required software.

Theoretical Concept:

The theoretical concepts underlying the classification of movie reviews using deep neural
networks are natural language processing (NLP) and sentiment analysis. Natural Language
Processing (NLP) is a branch of artificial intelligence that focuses on understanding and
processing human language. It involves techniques for text analysis, tokenization, syntactic
and semantic analysis, and sentiment analysis. Sentiment analysis, also known as opinion
mining, is a specific task within NLP that aims to determine the sentiment or subjective
information expressed in a piece of text. In the context of movie reviews, sentiment analysis
involves classifying the reviews into positive or negative sentiments based on the expressed
opinions and emotions.

Deep neural networks (DNNs) are a class of machine learning models inspired by the structure
and functioning of the human brain. DNNs are particularly effective for NLP tasks as they can
learn complex patterns and representations from large amounts of text data. In the case of movie
review classification, a deep neural network model is trained using labeled data, where the
movie reviews are the input and the sentiment labels (positive or negative) are the target
outputs. The model learns to extract meaningful features from the textual content and make

8|Page
predictions based on those features. Common architectures for text classification include
Convolutional Neural Networks (CNNs), Recurrent Neural Networks (RNNs) such as Long
Short-Term Memory (LSTM) or Gated Recurrent Unit (GRU), and combinations of these
architectures. These models are trained using gradient-based optimization algorithms and loss
functions specific to binary classification tasks, such as binary cross-entropy. During training,
the model adjusts its internal parameters based on the training data, optimizing the model's
ability to classify movie reviews accurately.

The performance of the model is evaluated using metrics such as accuracy, precision, recall,
and F1 score on a separate test set. By leveraging deep neural networks and sentiment analysis
techniques, it is possible to build powerful models that can automatically classify movie
reviews, enabling applications such as sentiment analysis in recommendation systems, market
research, and social media monitoring.

How Binary Classification Works:

The process of classifying movie reviews using deep neural networks involves several steps.
Here is an overview of how it works:

1. Data Preprocessing

• The IMDb dataset, which contains movie reviews labeled as positive or negative, is
loaded.
• The text data undergoes preprocessing steps such as removing HTML tags, and
punctuation marks and converting the text to lowercase.
• The reviews are often transformed into a numerical representation suitable for input to
the neural network, such as using word embeddings or one-hot encoding.

2. Model Architecture

• A deep neural network model architecture is designed, typically using layers such as
embedding layers, convolutional layers, recurrent layers (such as LSTM or GRU), and
dense layers.
• The architecture is chosen to effectively capture the semantic meaning and sentiment
expressed in the textual content of the reviews.
• The model's architecture and hyperparameters, such as the number of layers, number
of neurons, and activation functions, are determined based on experimentation and
validation performance.

9|Page
3. Training

• The prepared data is divided into training and validation sets. The model is trained on
the training set, which consists of input movie reviews and their corresponding
sentiment labels.
• During training, the model's parameters (weights and biases) are adjusted using
optimization algorithms, such as stochastic gradient descent (SGD) or Adam, to
minimize a defined loss function (e.g., binary cross-entropy).
• The model iteratively processes batches of reviews, updating the parameters based on
the gradient of the loss function, until convergence or a specified number of epochs.
• The performance of the model on the validation set is monitored to ensure it is not
overfitting, and techniques like early stopping can be applied to prevent excessive
training.

4. Evaluation

• The trained model is evaluated on a separate test set that was not seen during training
or validation.
• Performance metrics such as accuracy, precision, recall, and F1 score are calculated to
assess the model's classification performance.
• Additional evaluation measures like the confusion matrix and ROC curve may also be
analyzed to gain insights into the model's predictions.

5. Prediction and Deployment

• Once the model has been trained and evaluated, it can be used to predict the sentiment
of new, unseen movie reviews.
• The model is deployed in a production environment, where it takes input in the form of
text and outputs the predicted sentiment (positive or negative).
• The deployment can be through an API or an interface that allows users to interact with
the model and obtain sentiment predictions for movie reviews.

By iterative training on labeled data, adjusting parameters, and optimizing the model's ability
to recognize patterns and sentiment in text, deep neural networks can effectively classify movie
reviews into positive or negative categories.

10 | P a g e
Implementation:

11 | P a g e
12 | P a g e
13 | P a g e
Conclusion:

The classification of movie reviews using deep neural networks and sentiment analysis
techniques is a powerful approach for automating sentiment analysis tasks. By leveraging the
capabilities of deep learning models and natural language processing techniques, we can
develop models that accurately classify movie reviews as positive or negative based solely on
their textual content.

Overall, the combination of deep neural networks, sentiment analysis techniques, and natural
language processing enables efficient and accurate classification of movie reviews. This
approach revolutionizes the way we can analyze and understand sentiment in textual data,
opening doors to more effective decision-making and personalized user experiences in various
domains.

14 | P a g e
 Experiment No - 03
Title: Convolutional neural network: Use any dataset of plant disease and design a plant
disease detection system using CNN.

Objective: Software Requirement: Students should be able to implement a Convolutional

Neural Network. Design a plant disease detection system using CNN.

Software Requirement:

Operating system: Windows, Linux, MacOS

Programming Language: Python, C++, Java.

Theoretical Concept:

What Is a Convolutional Neural Network?

A Convolutional Neural Network (CNN) is a type of deep learning algorithm that is primarily
used for image recognition and processing, but it can also be applied to other types of data such
as sound and text. CNNs are inspired by the biological processes in the visual cortex of the
human brain.

The key components of a CNN include:

Convolutional Layers: These layers apply convolution operations to the input data using
filters or kernels. Convolution involves sliding a filter over the input image, performing
element-wise multiplication with the pixels in the filter and corresponding region of the input,
and then summing up the results to produce a feature map.

Pooling Layers: Pooling layers downsample the feature maps produced by the convolutional
layers. Common pooling operations include max pooling and average pooling, which reduce
the dimensionality of the feature maps while retaining important information.

Activation Functions: Activation functions introduce non-linearity to the network, allowing

it to learn complex patterns in the data. Common activation functions used in CNNs include
ReLU (Rectified Linear Unit), sigmoid, and tanh.

15 | P a g e
Fully Connected Layers: After several convolutional and pooling layers, the high-level
reasoning in the neural network is done via fully connected layers. These layers connect every
neuron in one layer to every neuron in the next layer, allowing for complex relationships to be
learned.

Output Layer: The output layer produces the final prediction or classification. The number of
neurons in this layer depends on the problem at hand. For example, in image classification
tasks, the number of neurons in the output layer corresponds to the number of classes.

Figure: Simple Architecture of CNN

Plant Disease Detection System

A plant disease detection system using Convolutional Neural Networks (CNNs) typically
involves the following steps:

Data Collection: Gather a dataset consisting of images of healthy plants as well as plants
affected by various diseases. These images should cover different plant species, diseases, and
stages of disease progression.

Data Preprocessing: Preprocess the images to ensure uniformity in size, color, and orientation.
Common preprocessing steps include resizing, normalization, and augmentation to increase the
diversity of the dataset.

16 | P a g e
Model Training: Train a CNN model using the preprocessed dataset. The CNN architecture
will consist of convolutional layers, pooling layers, activation functions, and fully connected
layers. The goal is to teach the model to distinguish between healthy and diseased plants based
on the visual patterns present in the images.

Validation: Validate the trained model using a separate dataset or through techniques like
cross-validation to assess its performance and generalization ability.

Deployment: Deploy the trained model into a system or application where it can be used for
disease detection. This could be a mobile app, a web-based tool, or an embedded system that
analyzes images captured from a camera.

Real-time Detection: Implement the system to allow real-time detection of plant diseases. This
may involve capturing images of plants using cameras or smartphones, preprocessing the
images, feeding them into the trained model, and then providing feedback on the presence of
diseases.

Feedback Loop: Continuously improve the system by collecting feedback on its performance
and updating the model accordingly. This could involve retraining the model with new data,
fine-tuning hyperparameters, or adjusting the architecture to improve accuracy and robustness.

Implementation:

17 | P a g e
18 | P a g e
19 | P a g e
20 | P a g e
21 | P a g e
22 | P a g e
23 | P a g e
Conclusion: In this way, we used a Convolutional neural network to design a plant disease
detection system.

24 | P a g e