What’s the Difference Between a CNN and an RNN?

blogs.nvidia.com

37 points·bcaulfield··7 comments

> An RNN is a neural network with an active data memory, known as the LSTM, that can be applied to a sequence of data to help guess what comes next.

Uh no. LSTMs are a specific formulation of a wider class of recurrent network topologies.

It's appalling to me how often this subset or instance-of relationship is missed when discussing these things.

I didn't read the whole article, but the title itself is a big giveaway that this is a bad one.

There is literally nothing similar between CNNs and RNNs apart from being types of neural networks.

Their designs are fundamentally different, to accommodate fundamentally different problems.( Temporal and spatial respectively)

This is a rather painful read and a relatively shallow article on CNN's vs. RNNs. Potentially useful for someone new to the field but it's a very grab bag take. The reality is much less well defined than CNNs = eyes, RNNs = ears and mouth. To start, CNNs have seen quite a bit of success in NLP applications.

Yeah, this article may actually be worse than nothing at all, in that it conflates a lot of concepts, and gets plenty of things flat out wrong.

Also, does it even matter what the difference between a CNN and a RNN is? Maybe in the sense of "what problems should I consider using a CNN vs. an RNN to solve" ... ?

Also, all "difference between ..." lists should be required to have a 3 column table with "criteria, item 1, item 2". That's the densest way to convey distinctions between items being compared.

The prose in this article reads like an 80s advertisement. Intentional, given the Knight Rider references?

Obviously, a Republican News Network gives a "fair and balanced" version of the news, while a CNN is just the opposite.

Reply on news.ycombinator.com