I wanna make a machine learning model, I know the basics like adjusting the weights to better find out if the image is a one or a zero, I need a large dataset of 1s and 0s and tell the model if it is a one or a zero then test and adjust, But how do I execute that in scratch.

I think you need tools to pull off machine learning that don't really exist in Scratch. For instance, you'll need to create an algorithm that can edit the code of other algorithms. Of course, this isn't possible in Scratch, as scripts can't edit other scripts.

I'm not an expert in these sorts of things, though. So take what I'm saying here with a grain of salt.

Ok, clearly This can't go anywhere. I'm sorry If I'm being a bummer. @king_of_fish made a good point. How do you make something that can write itself? Configuring lists to random broadcasts is so much spaghetti that I already tried 3 times and it still 1. Performs horrible 2. Doesn't do anything useful. I can suggest you make an algorithm. An algorithm is different form an ML model. It is a hard coded way to automate a task. If you consider the horrible bots in my game Zenophiles as algorithmic, that is an example. I highly suggest you try something else other than these two. It is hard to do something that is useful for many projects. If you are looking for something advanced to do, try 3d.

Ok

bump

Tems_student wrote:

Ok, clearly This can't go anywhere. I'm sorry If I'm being a bummer. @king_of_fish made a good point. How do you make something that can write itself? Configuring lists to random broadcasts is so much spaghetti that I already tried 3 times and it still 1. Performs horrible 2. Doesn't do anything useful. I can suggest you make an algorithm. An algorithm is different form an ML model. It is a hard coded way to automate a task. If you consider the horrible bots in my game Zenophiles as algorithmic, that is an example. I highly suggest you try something else other than these two. It is hard to do something that is useful for many projects. If you are looking for something advanced to do, try 3d.

It doesn't right itself its already written its just trained to tell the difference between 1s and 0s, also its not hard coded, heres and example:https://scratch.mit.edu/projects/238988807/

Tems_student wrote:

Ok, clearly This can't go anywhere. I'm sorry If I'm being a bummer. @king_of_fish made a good point. How do you make something that can write itself? Configuring lists to random broadcasts is so much spaghetti that I already tried 3 times and it still 1. Performs horrible 2. Doesn't do anything useful. I can suggest you make an algorithm. An algorithm is different form an ML model. It is a hard coded way to automate a task. If you consider the horrible bots in my game Zenophiles as algorithmic, that is an example. I highly suggest you try something else other than these two. It is hard to do something that is useful for many projects. If you are looking for something advanced to do, try 3d.

Its not going to be hard coded, It learns if a number is either a 1 or 0, machine learning does not mean coding itself.
If you want to see and example its here: https://scratch.mit.edu/projects/238988807/ tho i dont understand it

Sorry

jarscratch1111 wrote:

Tems_student wrote:

Ok, clearly This can't go anywhere. I'm sorry If I'm being a bummer. @king_of_fish made a good point. How do you make something that can write itself? Configuring lists to random broadcasts is so much spaghetti that I already tried 3 times and it still 1. Performs horrible 2. Doesn't do anything useful. I can suggest you make an algorithm. An algorithm is different form an ML model. It is a hard coded way to automate a task. If you consider the horrible bots in my game Zenophiles as algorithmic, that is an example. I highly suggest you try something else other than these two. It is hard to do something that is useful for many projects. If you are looking for something advanced to do, try 3d.

Its not going to be hard coded, It learns if a number is either a 1 or 0, machine learning does not mean coding itself.
If you want to see and example its here: https://scratch.mit.edu/projects/238988807/ tho i dont understand it

Sorry scratch broke so i didnt see my post

bump

Bump

2 words

if <> then 

  

else

end

And a lot of them
Then put every code block in it and have it choose what one to run depending on the ai level and stuff

cookieclickerer33 wrote:

2 words

if <> then 

  

else

end

And a lot of them
Then put every code block in it and have it choose what one to run depending on the ai level and stuff

I'll look into that but not helping that much tho

bump

jarscratch1111 wrote:

bump

Please don't bump topics unless they're on the second page…

Ok

groggs wrote:

jarscratch1111 wrote:

bump

Please don't bump topics unless they're on the second page…

Alright

Not an expert but everyone before me has been wrong, machine learning is definitely possible, I would use lists to store info and train it. It jsut takes a simple search through the projects to find a lot of examples.

hcps-holstejs wrote:

Not an expert but everyone before me has been wrong, machine learning is definitely possible, I would use lists to store info and train it.

Ok

What if you made a sort of coding language in the project that can write itself, would that be possible?

BUMP

cool_at_coding wrote:

What if you made a sort of coding language in the project that can write itself, would that be possible?

So an ML algorithm can work many ways, the first for example is is it makes a configuration of connections that will change each connection slightly to make it more accurate

second is there are a bunch of values you can change and they will alter what will happen I don't know much about it though