A hands-on guide to building and deploying deep learning models with Python
Key Features
● Acquire the skills to perform exploratory data analysis, uncover insights, and preprocess data for deep learning tasks.
● Build and train various types of neural networks, including Convolutional Neural Networks (CNNs) and Recurrent Neural Networks (RNNs).
● Gain hands-on experience by working on practical projects and applying deep learning techniques to real-world problems.
Description
“Deep Learning for Data Architects” is a comprehensive guide that bridges the gap between data architecture and deep learning. It provides a solid foundation in Python for data science and serves as a launchpad into the world of AI and deep learning.
The book begins by addressing the challenges of transforming raw data into actionable insights. It provides a practical understanding of data handling and covers the construction of neural network-based predictive models. The book then explores specialized networks such as convolutional neural networks (CNNs), recurrent neural networks (RNNs), and generative adversarial networks (GANs). The book delves into the theory and practical aspects of these networks and offers Python code implementations for each. The final chapter of the book introduces Transformers, a revolutionary model that has had a significant impact on natural language processing (NLP). This chapter provides you with a thorough understanding of how Transformers work and includes Python code implementations.
By the end of the book, you will be able to use deep learning to solve real-world problems.
What you will learn
● Develop a comprehensive understanding of neural networks' key concepts and principles.
● Gain proficiency in Python as you code and implement major deep-learning algorithms from scratch.
● Build and implement predictive models using various neural networks
● Learn how to use Transformers for complex NLP tasks
● Explore techniques to enhance the performance of your deep learning models.
Who this book is for
This book is for anyone who is interested in a career in emerging technologies, such as artificial intelligence (AI), data analytics, machine learning, deep learning, and data science. It is a comprehensive guide that covers the fundamentals of these technologies, as well as the skills and knowledge that you need to succeed in this field.
Author(s): Shekhar Khandelwal
Publisher: BPB Publications
Year: 2023
Language: English
Pages: 544
Table of Contents
1. Python for Data Science
Structure
Objectives
Setting up the development environment
Installing Anaconda
Advance Python libraries for data science
Numpy
Pandas
Reading and writing data to and from different file formats
Format - csv
Format - Excel
Format - JSON
Format - clipboard
Format - HTML tables
Format - PDF
Format - Web scraping
Improving efficiency with the pandas read_csv method
Parameter - dtype
Parameter - usecols
Parameter - chunksize
Conclusion
Questions
Answers
2. Real-World Challenges for Data Professionals in Converting Data Into Insights
Structure
Objectives
Pandas profiling
Analyzing Pandas profile report
Saving the Pandas profile report to a HTML file
Creating a Jupyter Notebook widget
Pandas profile report for big datasets
Sweetviz
Installing and getting started with Sweetviz
Analyzing Sweetviz report
Generating a report to compare two DataFrames using Sweetviz
AutoViz
Installing and getting started with AutoViz
Analyzing AutoViz report
Lux
Installing and getting started with Lux
Analyzing Lux report
Generating Lux visualizations based on intent
Saving Lux report to a HTML file
Advanced features in Lux reports
Lazy Predict
Analyzing Lazy Predict experimentation results
PyCaret
Installing and getting started with PyCaret
Analyzing PyCaret experimentation results
Advanced features of PyCaret
Conclusion
Questions
Answers
3. Build a Neural Network-Based Predictive Model
Structure
Objectives
Artificial neural network and its components
Neurons
Feed forward
Activation functions
Loss function
Backward propagation
Epoch
Batch
Iteration
Optimizer
Learning rate
Building a classification model using neural network
Problem statement
Dataset
Implementation
Load Python libraries
Load data
Descriptive analytics
Data pre-processing
Modeling
Experiment 1 - Hidden layer -1, epoch-100 – shallow neural network
Experiment 2 - hidden layer -2, epoch-100 – deep neural network
Building a regression model using neural network
Problem statement
Dataset
Data pre-processing
Modeling
Model evaluation
Conclusion
Questions
Answers
4. Convolutional Neural Networks
Structure
Objectives
Convolutional neural networks components
Load required libraries
Digital image as a numpy array
Kernels/filters and convolution process
Stride
Padding
Convolution on RGB image
Convolution operation with multiple filters
One convolution layer
Pooling
Flattening
Dense layers
Image classification using CNN
Problem statement
Dataset - MNIST
Implementation
Data pre-processing
Modeling
Plot confusion matrix
Create a confusion matrix and plot
Hyperparameters tuning using KerasTuner
Dataset – Fashion MNIST
Implementation
Install KerasTuner
Conclusion
Questions
Answers
5. Optical Character Recognition
Introduction
Structure
Objectives
Optical character recognition
OCR Python libraries and their implementation
Tesseract OCR
Tesseract OCR demo
keras-ocr
keras-ocr demo
EasyOCR
EasyOCR demo
TrOCR
TrOCR demo
Conclusion
Questions
Answers
6. Object Detection
Structure
Objectives
Object localization and detection
Object detection algorithms and their comparison
Single shot detector Python implementation
YOLO v3 Python implementation
Experiment 2
Experiment 3
Experiment 4
Experiment 5
Conclusion
Questions
Answers
7. Image Segmentation
Structure
Objectives
Difference between image classification, detection and segmentation
Image segmentation architectures
U-Net Python implementation
FCN-8 Python implementation
Mask R-CNN Python implementation
Conclusion
Questions
Answers
8. Recurrent Neural Networks
Structure
Objectives
Algorithms for RNN implementation
RNN implementation
Long short-term memory implementation
Gated Recurrent Unit implementation
Conclusion
Multiple choice questions
Answers
9. Generative Adversarial Networks
Structure
Objectives
Types of GAN
Vanilla GAN Python implementation
Key difference between Vanilla GAN and DCGAN
DCGAN Python implementation
StyleGAN Python implementation
Setup environment
Conclusion
Questions
Answers
10. Transformers
Structure
Objectives
Introduction to Transformers in deep learning
Various transformers architectures
Difference between contextual and non-contextual embeddings
BERT Python implementation
GPT Python implementation
Conclusion
Questions
Answers
Index
