I'm a noob at Deep Learning. What should be my first project on it? Something I could publish as a paper. Ideas are welcomed :)

>first project
>paper

What

@RememberMe Gotta add that sweet line on your resume!

MLP based MNIST digit classification. Just don't expect the paper to be published anywhere...

@RememberMe I meant research paper. I hope it makes it clearer now.

@devAstated yes, how is a first project fit for a paper?

@RememberMe I am noob as in I haven't build something substantial on Deep Learning yet. I have trained a few neural networks, though. I want to build something that has some value, you know. If I could publish it then it will increase the credibility of the project. Otherwise it would be basic just like how most of the college undergrads do it.

@devAstated You should absolutely start basic, like how most of the undergrads do it, if it's your first project - hence my suggestion.

Sure, please do write up a research paper on it - it's good practice and allows you to cement your understanding. Just know that you're doing it for practice, not to get it published in a journal somewhere. That's never going to fly unless you have way more experience.

start basic

Not hotdog

I suggest you ditch all python / R libraries and do something really easy and simple from scratch, to force yourself to understand perfectly. You don't have to do deep learning you can implement an older algorithm from scratch also and learn a ton.

If you're a fan of John Carmack, I can't say he shares my opinion but he did do what I'm suggesting as part of his process
https://twitter.com/ID_AA_Carmack/...

@ArtOfBBQ Good advice, but it's for the university summer project and I have to present it in September.

@devAstated arf gotcha

predict mobile app profitability based on their screenshots

@ArtOfBBQ That's something different I've heard so far. Seems interesting. I started working on an American Sign Language interpretation model. I'll take it up after I finish this.

Hehe I just blurted something out. You seem to really like to dive in the deep end it's definitely not optimal but if it makes you excited & motivated that might be worth a lot also