neural network based rating

[Deleted member 12217]

I recently started to learn about AI and neural networks in my university and figured it would be the perfect technique to apply for rating people looks.

I'm totally new to this type of project tho so im asking fellow high iq maxxers to help me out with the planning. it will take a while (if I ever finish it or even start it )

I thought about splitting this in 2 phases

1. phase would be to gather pictures and ratings. The goal would be to have alarge enough data set of pictures and ratings to train a neural network with it .
I ve never done anything like that so I'm.open for suggestions on how to do that. my idea was to host a
small server for everyone to vote pictures of totally random dudes and Chads and save the results in a database .

if that step.is done phase 2 would simply to train the neural network with the set

as I said I've never done anything like ththaththat so I won't promise anything. but I think it would be a super beneficial project for everyone and even if I it's just gathering the data and failing at the rest. someone else could probably continue with it

help and tips for databases and how to store pictures for free if possible greatly appreciated

 

[Deleted member 10551]

Wish I could help. I’d be interested in learning.

 

[Deleted member 2214]

Rate me and my little brother bro:

 

[Deleted member 11748]

Why we didn't think of this before? Actually saves so much time on the rating section.

User won't get doxxed too. But it kinda takes the fun away...

 

[Deleted member 6873]

Why we didn't think of this before? Actually saves so much time on the rating section.

love the cat avi ngl

 

[Deleted member 11748]

love the cat avi ngl

its pretty cool ik

 

[Deleted member 1862]

1. phase would be to gather pictures and ratings. The goal would be to have alarge enough data set of pictures and ratings to train a neural network with it .
I ve never done anything like that so I'm.open for suggestions on how to do that. my idea was to host a
small server for everyone to vote pictures of totally random dudes and Chads and save the results in a database .

if that step.is done phase 2 would simply to train the neural network with the set

1. There are plenty of public datasets

2. This wouldn't rly be that good bc you'd still have to manually label each picture & all it takes is one look at the ratings section to see that there is some subjectivity in ratings

When you learn more about neural nets & how they work, you'll understand why some ideas that appear good/interesting on paper are actually just boring & needlessly difficult

 

[Deleted member 12217]

1. There are plenty of public datasets

2. This wouldn't rly be that good bc you'd still have to manually label each picture & all it takes is one look at the ratings section to see that there is some subjectivity in ratings

When you learn more about neural nets & how they work, you'll understand why some ideas that appear good/interesting on paper are actually just boring & needlessly difficult

can you link a public dataset ? I thought gathering the data was the hard part ngl .

thought i could simply apply a feed forward correct backwards which spits out a classification. just like neural networks that classificate plants but with numbers and faces instead of plant type and plants

 

[Deleted member 1862]

can you link a public dataset ? I thought gathering the data was the hard part ngl .

thought i could simply apply a feed forward correct backwards which spits out a classification. just like neural networks that classificate plants but with numbers and faces instead of plant type and plants

just check kaggle for datasets bro

wdym feedforward backwards, you realise backprop is how neural nets 'learn' right?

 

[Deleted member 12217]

just check kaggle for datasets bro

wdym feedforward backwards, you realise backprop is how neural nets 'learn' right?

yeah it's probably smarter to do a test run with kaggle data first . and yeah I was referring to a basic neural network. afaik there are a lot of different ways for example evolution ones and so on. as I said I'm just beginning. I will have to code one as an exam anyways so I will at least give it a try