# Why Deep Learning takes off
The quality and quantity of data increases and makes [[Deep Learning]] much more performant.
Faster computation and specialized hardware also helps. There's also been huge innovation on the algorithms (e.g., switch the activation functions from [[Sigmoid Function]] to [[Rectified Linear Unit Functions (RELU)]]). Faster computation and faster training helped increase the rate at which we can try things out and iterate faster.
Back propagation is great for learning.
## Related
- [[Deep Learning]]
- [[Neural Networks and Deep Learning]]