What is Data Augmentation? Use Cases & Examples

The performance of most ML models, and deep learning models in particular, depends on the quality, quantity and relevancy of training data. However, insufficient data is one of the most common challenges in implementing machine learning in the enterprise. This is because collecting such data can be costly and time-consuming in many cases.

Companies can leverage data augmentation to reduce reliance on training data collection and preparation and to build more accurate machine learning models faster.

What is data augmentation?

Data augmentation is a set of techniques to artificially increase the amount of data by generating new data points from existing data. This is achieved by making small changes to data or using deep learning models to generate new data points.

Why is it important now?

Machine learning applications especially in the deep learning domain continue to diversify and increase rapidly. Data-centric approaches to model development such as data augmentation techniques can be a good tool against challenges that the artificial intelligence world faces.

Data augmentation is useful to improve the performance and outcomes of machine learning models by forming new and different examples to train datasets. If the dataset in a machine learning model is rich and sufficient, the model performs better and more accurately.

For machine learning models, collecting and labeling data can be exhausting and costly processes. Transformations in datasets by using data augmentation techniques allow companies to reduce these operational costs.

One of the steps in a data model is cleaning data which is necessary for high-accuracy models. However, if cleaning reduces the representability of data, then the model cannot provide good predictions for real-world inputs. Data augmentation techniques can enable machine learning models to be more robust by creating variations that the model may see in the real world.

You can also check our article on the benefits of data augmentation for deep learning models.

How does it work?

For image classification and segmentation

Simple alterations on visual data can be completed with data augmentation techniques. In addition, generative adversarial networks (GANs) are used to create new synthetic data. Classic image processing activities for data augmentation are:

• padding
• random rotating
• re-scaling,
• vertical and horizontal flipping
• translation ( image is moved along X, Y direction)
• cropping
• zooming
• darkening & brightening/color modification
• grayscaling
• changing contrast
• adding noise
• random erasing

Advanced models for data augmentation are

• Adversarial training/Adversarial machine learning: It generates adversarial examples that disrupt a machine learning model and inject them into a dataset to train.
• Generative adversarial networks (GANs): GAN algorithms can learn patterns from input datasets and automatically create new examples that resemble training data.
• Neural style transfer: Neural style transfer models can blend content image and style image and separate style from content.
• Reinforcement learning: Reinforcement learning models train software agents to attain their goals and make decisions in a virtual environment.

Popular open source Python packages for data augmentation in computer vision are Keras ImageDataGenerator, Skimage, and OpeCV.

For natural language processing (NLP)

Data augmentation is not as popular in the NLP domain as in the computer vision domain. Augmenting text data is difficult, due to the complexity of a language. Common methods for data augmentation in NLP are:

• Easy Data Augmentation (EDA) operations: synonym replacement, word insertion, word swap and word deletion
• Back translation: re-translating text from the target language back to its original language
• Contextualized word embeddings

What are the use cases in data augmentation?

Self-driving cars

Self-driving cars require 3D object identification. Most Object Detection models require a significant amount of data to be trained. Gathering and processing this data is a costly and time-consuming operation, thus the information extracted from each sample must be properly used.

This is possible with data augmentation since it develops strong machine-learning models for automotive applications like self-driving cars.

For example, generative adversarial networks, GANs, and variational autoencoders, VAEs, (generative models for generating new data samples) are used in data training to alter weather conditions and add or remove automobiles from street scenes.

Figure: Example results of winter to summer translation. Top row: raw winter images. Bottom row: synthesized summer images. 

Source: SemanticScholar¹

Note that, the applications of augmented data as a simulation are limitless, as it may be utilized to create real-world scenarios.

Automated speech recognition

By enriching existing data, data augmentation greatly enhances the number and diversity of training samples, resulting in better generalization and performance of speech recognition models. This allows models to properly manage various acoustic settings and speaker characteristics.

Common data augmentation techniques in speech recognition:

• Time stretching: Altering the speed of the audio without changing its pitch.
• Pitch Shifting: Modifying the pitch of the audio while maintaining the same speed.
• Adding noise: Introducing background noise to simulate real-world environments.
• Shifting time frames: Moving the audio slightly forward or backward in time.

Figure: Noise injection

Source: Medium²

Text augmentation

Data augmentation solutions are critical in domains such as computer vision and natural language processing (NLP), where data scarcity and limited variation present issues.

While creating augmented visuals is relatively simple, NLP is complicated owing to the hidden structure of language. Unlike images, we cannot replace every word with a synonym, and even when replacement is possible, maintaining context becomes a significant challenge.

Rule-based approaches: The fundamental function of data augmentation is to improve model performance by increasing the number of training data.

To do this, neural networks can be used to create new text samples from input data. This involves relatively simple find-and-replace techniques, such as:

• Random insertion: Inserting random words from a sentence into new positions to create diverse sentence patterns.
• Random deletion: Randomly removing words from a sentence to generate simplified variations.

Source: IBM³

Back translation approaches: Translating text into another language and back to the original language to generate paraphrased data.

A back-translation neural approach could convert input data into a target language and then back into the original input language. This helps generate semantic variances in a single-language dataset for augmentation.

Source: IBM⁴

Image augmentation

Data augmentation is widely used in computer vision applications, including image categorization and object recognition.

This technique covers techniques that affect the space, layout, or color of the original image, such as:

• Rotation: Rotating the image by a random angle (e.g., 15 degrees) to simulate different orientations of objects.
• Flipping: Horizontally or vertically flipping the image to create mirrored versions.
• Blurring: Applying blur to an image to simulate out-of-focus conditions.
• Cropping: Randomly cropping a portion of the image and resizing it to the original dimensions.
• Zooming: Zooming in or out of the image to simulate different object sizes.
• Color jittering: Randomly changing brightness, contrast, saturation, or hue of an image.

Source: IBM⁵

Medical imaging

Data augmentation is a helpful tool in medical imaging since it aids in developing diagnostic models that identify, and diagnose diseases using images.

The reasons for data augmentation interest in healthcare are:

• Small dataset for medical images
• Sharing data is not easy due to patient data privacy regulations
• There are only a few patients whose data can be used as training data in the diagnosis of rare diseases

Example studies in this field include:

• Brain tumor segmentation
• Differential data augmentation for medical imaging
• An automated data augmentation method for synthesizing labeled medical images
• Semi-supervised task-driven data augmentation for medical image segmentation

Source: ScienceDirect⁶

How is it different from synthetic data?

Generating synthetic data is one way to augment data. There are other approaches (e.g. making minimal changes to existing data to create new data) for data augmentation as outlined above.

Check our article on synthetic data for computer vision.

What are the benefits of data augmentation?

Benefits of data augmentation include:

• Improving model prediction accuracy
  • adding more training data to the models
  • preventing data scarcity for better models
  • reducing data overfitting ( i.e. an error in statistics, it means a function corresponds too closely to a limited set of data points) and creating variability in data
  • increasing the generalization ability of the models
  • helping resolve class imbalance issues in classification
• Reducing costs of collecting and labeling data
• Enables rare event prediction
• Prevents data privacy problems

What are the challenges of data augmentation?

• Companies need to build evaluation systems for the quality of augmented datasets. As the use of data augmentation methods increases, an assessment of the quality of their output will be required.
• Data augmentation domain needs to develop new research and studies to create new/synthetic data with advanced applications. For example, the generation of high-resolution images by using GANs can be challenging.
• If a real dataset contains biases, data augmented from it will contain biases, too. So, the identification of an optimal data augmentation strategy is important.

This article was drafted by former AIMultiple industry analyst Ayşegül Takımoğlu.

External Links

• 1. ”GAN Based Method for Labeled Image Augmentation in Autonomous Driving“. SemanticScholar. 2019 Retrieved October 23, 2024.
• 2. Ma, Eward, “Data augmentation for medical imaging: A systematic literature review“. Medium. 2019, Retrieved October 23, 2024.
• 3. “What is data augmentation?“. IBM. 2024, Retrieved October 23, 2024.
• 4. “What is data augmentation?“. IBM. 2024, Retrieved October 23, 2024.
• 5. “What is data augmentation?“. IBM. 2024, Retrieved October 23, 2024.
• 6. ”Data augmentation for medical imaging: A systematic literature review“. ScienceDirect. 2023, Retrieved October 23, 2024.