Deep Learning Specialization Cert

AI Discussions
1

I am currently on the 3rd course for the Specialization but my interests are in Physics Informed Applications of ML for engineering optimization and design. I see the last two courses are about CNNs and Sequence models. From my understanding that largely relates to LLMs and image processing. Should I take the time to finish those courses too or pursue other courses like IBMs courses on ML frameworks and libraries?

2

I think that’s taking too limited a view of what is possible with Convolutional Nets (CNNs) and Sequence Models (RNNs). RNNs are a general mechanism that is useful for dealing with (modeling) phenomena that are serial where subsequent events are influenced (but not completely determined) by previous events in the sequence. Language processing is one great application of RNNs, but serial processes are also very common in physics, aren’t they?

Convolutional Nets are good at applying filters across the geometry of inputs and detecting patterns that may occur in multiple places. As you say, image processing and computer vision are very powerful applications of CNNs, but aren’t there cases in physics in which you might have phenomena that occur in parallel at multiple points in the inputs in a similar fashion?

Note that Courses 4 and 5 of DLS are essentially independent in terms of the fundamentals of CNNs and RNNs, but C4 is the first place where we really start to see serious applications of the power of TensorFlow (we just barely got introduced to it in C2). Then in C5, the material assumes you are reasonably well versed in building complex TF models because you’ve been through C4. Of course if you already have previous solid experience using TF, then that frees you more from any need to take C4 before C5.

1 Like
3

Ah ok makes sense every time I tried to look up what the uses for those were they all generally pointed to LLMS and imagery which made it hard for me to know if the sources were more attuned to recency bias with all these LLMS starting to appear. Thanks for the adivce I’ll continue towards finishing the course!

1 Like