Deep Learning (DL)
In short
A subfield of Machine Learning that uses complex architectures to learn patterns that simpler ML techniques couldn’t handle — at the cost of needing much more Data.
If Machine Learning is learning to recognize dogs from a few key features (has fur, four legs, tail), Deep Learning is learning to recognize specific breeds from thousands of subtle details you couldn’t even list.
Deep Learning came up as a method through which models could learn more complex patterns and behaviors that were otherwise hard to represent with simple Machine Learning techniques. The trade-off is that it requires significantly more data for training. This is the foundation that eventually led to LLMs and the current AI revolution.
Related
- Machine Learning - DL is a subfield of ML
- LLMs - built on deep learning
- Neural Network - the building block of DL
- Data - DL is hungry for data