Unlock the power of data augmentation for AI and Generative AI with real-world datasets. Improve your model’s accuracy and extend images, texts, audio, and tabular using 150+ fully functional OO methods and open-source libraries.
Key Features
Practical Data augmentation techniques for images, texts, audio, and tabular data using real-world datasets
Beautiful, customized charts and infographics in full color for image, text, audio, and tabular data
Fully functional object-oriented code using open-source libraries on the Python Notebook for each chapter
Book Description
Data is paramount in an AI project, especially for Deep Learning and Generative AI. The forecasting accuracy relies on robust input datasets. The traditional method of acquiring additional data is difficult, expensive, and impractical. The only option to extend the dataset economically is data augmentation.
You will learn 20+ Geometric, Photometric, and Random erasing augmentation methods using seven real-world datasets for image classification and segmentation. In addition, we will review eight image augmentation open-source libraries, write OOP wrapper functions on the Python Notebooks, view color image augmentation effects, analyze the safe level and biases, and extend the chapter with Fun facts and Fun challenges.
You will discover 22+ character and word techniques for text augmentation using two real-world datasets and excerpts from four classic books. The advanced text augmentation chapter uses Machine Learning to extend the text dataset, such as Transformer, Word2vec, BERT, GPT-2, and others.
Similarly, the audio and tabular data chapters have real-world data, open-source libraries, amazing custom plots, Python Notebook, Fun facts, and Fun challenges.
By the end of the book, you will be proficient in image, text, audio, and tabular data augmentation techniques.
What you will learn
Write OOP Python code for image, text, audio, and tabular data
Access over 150,000 real-world datasets from the Kaggle websites
Analyze biases and safe parameters for each augmentation method
Visualize data using standard and exotics plots in color
Explore 32 advanced open-source augmentation libraries
Discover Machine Learning models, such as BERT and Transformer
Meet Pluto, an imaginary digital coding companion
Extend your learning with Fun facts and Fun challenges
Who This Book Is For
The book is for AI, Data scientists, and students interested in the AI discipline. You don’t need advanced AI or Deep Learning skills, but Python programming and familiarity with Jupyter Notebooks are required.
Author(s): Duc Haba
Edition: 1
Publisher: Packt Publishing
Year: 2023
Language: English
Pages: 394
Cover
Title Page
Copyright
Dedication
Foreword
Contributors
Table of Contents
Preface
Part 1: Data Augmentation
Chapter 1: Data Augmentation Made Easy
Data augmentation role
Data input types
Image definition
Text definition
Audio definition
Tabular data definition
Python Notebook
Google Colab
Additional Python Notebook options
Installing Python Notebook
Programming styles
Source control
The PacktDataAug class
Naming convention
Extend base class
Referencing a library
Exporting Python code
Pluto
Summary
Chapter 2: Biases in Data Augmentation
Computational biases
Human biases
Systemic biases
Python Notebook
Python Notebook
GitHub
Pluto
Verifying Pluto
Kaggle ID
Image biases
State Farm distracted drivers detection
Nike shoes
Grapevine leaves
Text biases
Netflix
Amazon reviews
Summary
Part 2: Image Augmentation
Chapter 3: Image Augmentation for Classification
Geometric transformations
Flipping
Cropping
Resizing
Padding
Rotating
Translation
Noise injection
Photometric transformations
Basic and classic
Advanced and exotic
Random erasing
Combining
Reinforcing your learning through Python code
Pluto and the Python Notebook
Real-world image datasets
Image augmentation library
Geometric transformation filters
Photographic transformations
Random erasing
Combining
Summary
Chapter 4: Image Augmentation for Segmentation
Geometric and photometric transformations
Real-world segmentation datasets
Python Notebook and Pluto
Real-world data
Pandas
Viewing data images
Reinforcing your learning
Horizontal flip
Vertical flip
Rotating
Resizing and cropping
Transpose
Lighting
FancyPCA
Combining
Summary
Part 3: Text Augmentation
Chapter 5: Text Augmentation
Character augmenting
Word augmenting
Sentence augmentation
Text augmentation libraries
Real-world text datasets
The Python Notebook and Pluto
Real-world NLP datasets
Pandas
Visualizing NLP data
Reinforcing learning through Python Notebook
Character augmentation
Word augmenting
Summary
Chapter 6: Text Augmentation with Machine Learning
Machine learning models
Word augmenting
Sentence augmenting
Real-world NLP datasets
Python Notebook and Pluto
Verify
Real-world NLP data
Pandas
Viewing the text
Reinforcing your learning through the Python Notebook
Word2Vec word augmenting
BERT
RoBERTa
Back translation
Sentence augmentation
Summary
Part 4: Audio Data Augmentation
Chapter 7: Audio Data Augmentation
Standard audio augmentation techniques
Time stretching
Time shifting
Pitch shifting
Polarity inversion
Noise injection
Filters
Low-pass filter
High-pass filter
Band-pass filter
Low-shelf filter
High-shelf filter
Band-stop filter
Peak filter
Audio augmentation libraries
Real-world audio datasets
Python Notebook and Pluto
Real-world data and pandas
Listening and viewing
Reinforcing your learning
Time shifting
Time stretching
Pitch scaling
Noise injection
Polarity inversion
Low-pass filter
Band-pass filter
High-pass and other filters
Summary
Chapter 8: Audio Data Augmentation with Spectrogram
Initializing and downloading
Audio Spectrogram
Various Spectrogram formats
Mel-spectrogram and Chroma STFT plots
Spectrogram augmentation
Spectrogram images
Summary
Part 5: Tabular Data Augmentation
Chapter 9: Tabular Data Augmentation
Tabular augmentation libraries
Augmentation categories
Real-world tabular datasets
Exploring and visualizing tabular data
Data structure
First graph view
Checksum
Specialized plots
Exploring the World Series data
Transforming augmentation
Robust scaler
Standard scaler
Capping
Interaction augmentation
Regression augmentation
Operator augmentation
Mapping augmentation
Extraction augmentation
Summary
Index
About Packt
Other Books You May Enjoy
