Member-only story
Embeddings : Data to Numbers
(You can find the Korean version of the post at this link.)
This post marks the conclusion of our series on the topic of āRepresentationā. Throughout this journey, we delved into 1) Representation Learning, 2) Sequence-to-Sequence Learning, and 3) Hands-on practice with Sequence-to-Sequence models.
Whether developing artificial intelligence or other ICT software, our starting point must always be encapsulating the world in a format that computers can process. This means transforming any form of data into numbers. The data we refer to here encompasses not just text and images we encounter daily, but also chemical formulas, Excel files, weather conditions, and even stock market trends over 50 years ā virtually all types and forms of data. To encapsulate this process, we can aptly describe it as āData to Numbersā.
This method is known by several other names as well,
- Encoding
- Embedding
- Feature Extraction
- Vectorization
- ā¦
The concept of āData to Numbersā is not exclusively tied to recent technological advancements. While neural network-based techniques are frequently highlighted nowadays, various forms of technology have existed even before the advent of deep learning. In this post, we aim toā¦
