Synthetic data types
There are various synthetic data types, such as textual, imagery, point cloud, and tabular. Based on the ML problem and task, different types of data are required. In this section, we will discuss the main types of synthetic data in more detail.
![Figure 4.5 – A sample of synthetic data types](https://static.packt-cdn.com/products/9781803245409/graphics/image/Figure_04_05_B18494.jpg)
Figure 4.5 – A sample of synthetic data types
- Text: Wikipedia, digital books, lexicons, and text corpora are examples of textual data. ML models can be trained on large-scale textual datasets to learn the structure of the text that we generate or write as humans. Then, these models can be leveraged to answer questions, summarize texts, or translate from one language to another. These models, such as ChatGPT, ChatSonic (https://writesonic.com), and Jasper Chat (https://www.jasper.ai), work by generating synthetic texts based on making predictions on what word should come next.
- Video, image, and audio: ML models can learn the patterns in a video, image, or audio, and then they...