Is Deep Learning actually hard to learn for someone with a non-math background?

All answers to this question.

It’s definitely a steep learning curve, but it’s not impossible! You don't need to be a mathematician to start. Modern libraries like Keras and Fast.ai are designed to let you build models with just a few lines of code. However, you will eventually need a basic grasp of linear algebra (for data shapes) and calculus (to understand how models improve) if you want to troubleshoot or optimize your work. My advice is to start with a "top-down" approach: build something first to see it work, and then slowly dive into the math as you encounter concepts you don't understand.

   Answered 2025-01-07 by Barbara Nelson

Do you think it's better to start with a library like PyTorch or TensorFlow for a beginner who wants to understand the underlying logic without getting buried in the math?

   Answered 2025-01-09 by Gary Collins

• Gary, I'd suggest PyTorch. It feels more like standard Python and is much more intuitive for debugging. TensorFlow has a steeper learning curve for beginners, though it's very powerful for production. PyTorch allows you to see what's happening at each step more clearly, which helps bridge that gap between code and theory.

     Commented 2025-01-11 by Ronald Perry

It's more about persistence than genius. If you can handle Python logic, you can eventually wrap your head around the concepts of neural networks.

   Answered 2025-01-12 by Larry Simmons

• Agreed, Larry. Most of the "hard" part is just getting used to the specific terminology. Once the vocabulary clicks, the rest starts to follow naturally with practice.

     Commented 2025-01-13 by Donna Higgins