Welcome to Line or Not, a simple 64-8-1 neural network coded in Scratch made to identify horizontal and vertical lines (not diagonal) 6-8 pixels long. Test it with different drawings and see how well the neural network can tell a line from a non-line! -- Vocabulary -- Epoch - One pass of every image in the training dataset to the model. More epochs = the model sees every image more. Learning Rate - How big of a "step" the model takes every time. Higher learning rates = model trains faster but less accurate.
-- Notes -- Try not to use the low intensity training mode, it greatly slows down training speeds and unless if you are in the editor while training and have a low # of epochs, I would not recommend. Due to the low amount of training data, it may not be the most accurate when identifying lines/non-lines. -- Music -- Morning Routine by Ghostrifter Official | https://soundcloud.com/ghostrifter-official Music promoted by https://www.chosic.com/free-music/all/ Creative Commons CC BY-SA 3.0 https://creativecommons.org/licenses/by-sa/3.0/ -- Datasets -- Big help from Chat GPT with the .csv sheets it created housing all of the training data, saved me hours of work.
