When it comes to understanding deep learning, we often hear that math is essential. But does it really help build intuition? I’ve been pondering this question as I weigh the pros and cons of two master’s programs: UvA’s MSc in AI and ETHz’s MSc in DS. The main difference between the two is that UvA focuses on concepts, while ETHz delves deeper into the math. ETHz is undoubtedly more challenging and time-consuming, but does that extra effort really pay off in terms of developing a deeper understanding of neural networks?
For me, the ultimate goal is to gain a genuine understanding of how and why neural nets learn. I’m not interested in just memorizing formulas or going through the motions. I want to be able to look at a neural network and intuitively understand what’s going on. So, the question remains: does the extra math and proofs in ETHz’s program actually help build that intuition, or is it just a necessary evil that doesn’t really contribute to a deeper understanding?
In my experience, math can sometimes feel like a low-level complexity that obscures the bigger picture. It’s easy to get bogged down in derivations and proofs, but do they really help us understand the underlying mechanisms of deep learning? I’d love to hear from others who have walked this path before. Does the math really matter, or is it just a means to an end?