In this chapter, we will discuss and define imbalanced datasets, explaining how they differ from other types of datasets. The ubiquity of imbalanced data will be demonstrated with examples of common problems and scenarios. We will also go through the basics of machine learning and cover the essentials, such as loss functions, regularization, and feature engineering. We will also learn about common evaluation metrics, particularly those that can be very helpful for imbalanced datasets. We will then introduce the imbalanced-learn library.

In this chapter, we will introduce the concept of oversampling, discuss when to use it, and the various techniques to perform it. We will also demonstrate how to utilize these techniques through the imbalanced-learn library APIs and compare their performance using some classical machine learning models. Finally, we will conclude with some practical advice on which techniques tend to work best under specific real-world conditions.

In this chapter, you will learn about the concept of undersampling, including when to use it and the various techniques to perform it. You will also see how to use these techniques via the imbalancedlearn library APIs and compare their performance with some classical machine learning models.

The problem with traditional ensemble methods is that they use classifiers that assume balanced data. Thus, they may not work very well with imbalanced datasets. So, we combine the popular machine learning ensembling methods with the techniques for dealing with imbalanced data that we studied in previous chapters. We are going to discuss those combinations in this chapter.

Cost-sensitive learning is an e!ective strategy to tackle imbalanced data. We will go through this technique and learn why it can be useful. This will help us understand some of the details of cost functions and how machine learning models are not designed to deal with imbalanced datasets by default. While machine learning models aren’t equipped to handle imbalanced datasets, we will see how modern libraries enable this..

Class imbalanced data is a common issue for deep learning models. When one or more classes have significantly fewer samples, the performance of deep learning models can su!er as they tend to prioritize learning from the majority class, resulting in poor generalization for the minority class(es). We will cover the following topics in this chapter: A brief introduction to deep learning, Data imbalance in deep learning, Overview of deep learning techniques to handle data imbalance, and Multi-label classification

In this chapter, we’ll explore how to apply familiar sampling methods to deep learning models. Deep learning o!ers unique opportunities to enhance these methods further. We’ll delve into elegant techniques to combine deep learning with oversampling and undersampling. Additionally, we’ll learn how to implement various sampling methods with a basic neural network. We’ll also cover dynamic sampling, which involves adjusting the data sample across multiple training iterations, using varying balancing ratios for each iteration. Then, we will learn to use some data augmentation techniques for both images and text. We’ll end the chapter by highlighting key takeaways from a variety of other data-level techniques.

This chapter will be on the same lines as Chapter 5, Cost-Sensitive Learning, extending the ideas to deep learning models. We will look at algorithm-level deep learning techniques to handle the imbalance in data. Generally, these techniques do not modify the training data and o#en require no pre-processing steps, o!ering the bene"t of no increased training times or additional runtime hardware costs.

In this chapter, we will talk about some of the hybrid deep learning techniques that combine the data-level and algorithm-level methods in some ways. $is chapter contains some recent and more advanced techniques that can be challenging to implement, so it is recommended to have a good understanding of the previous chapters. We will continue to explore strategies to tackle class imbalance in deep learning, examining techniques that manipulate data distribution and prioritize challenging examples. We will also go over techniques called hard example mining and minority class incremental recti"cation, which focus on improving model performance through prioritization of di%cult instances and iterative enhancement of minority class representation, respectively.

In this chapter, we will see the need to do some post-processing of the prediction scores that we get from the trained models. This can be helpful either during the real-time prediction from the model or during the o&ine training time evaluation of the model. We will also understand some ways of measuring how calibrated the model is and how imbalanced datasets make the model calibration inevitable.