Expert system-The first node unlocked in gameplay, used to sort colors, one color goes somewhere, the rest go somewhere else. Used to sort neat, but slow. works very simple.
Decision tree-The second Node unlocked in gameplay, used to sort colors, two colors are selected and the 3rd color is randomized. not so neat but pretty fast, Works a bit more complicated.
Sift-The final node unlocked in first epoch and first shape node in gameplay, used to sort shapes. works a lot like the expert system, but a bit slower. Works very simple.

Trashcan - Deletes nodes. The most simple node along with the two training nodes. (used later in gameplay.) Adding trashcan database has an unlimited amount of these nodes. (only counts as 1 node)

Mark-1 perceptron - First node unlocked in 2nd epoch, used to sort shapes. you can have all 3 shapes organized, even though it might not be accurate without training, later in gameplay, requires decent to train. works like the color perceptron, a bit complicated to use accurately.

#Sift - Used in RL tasks. sorts objects like the decision tree and Sift. It's full name is Sift: object detection. used to tune your car away from blockades such as cars and traffic barriers. works just like the decision tree. Pretty complicated in RL. (since there is no trashcan.)
Balancer - Sorts out data evenly, Simpler then expert system, a bit more complicated then trashcan and training nodes. used for any type of data. Requires server power

Genetic solver - Requires training and training node. (later in gameplay) Sorts color. may make mistakes depending on training. sorts all 3 colors into different places. Pretty complicated to train and use accurately. Requires evolving.
##Sift - Rl node.used to sort blockades. Car or traffic barrier. works like #Sift. it's full name is Sift: classifacation of objects
Decision tree: shape - Works like decision tree but used to sort shapes. same style as rl nodes. Simple as the decision tree is.

Gradient descent - Used to train nodes. just hook em and use em'! this node does not work in test run
Perceptron shape - A bit more accurate and faster then mark-1. used the same way though.
Perceptron color- A Relative to perceptron shape but uses color.

Rnn layer - Sorts letters so the good ones go last. That's how the accuracy builds. Requires memory connection and training.
Arma - Sorts numbers like Rnn. it also requires connection and training with training nodes.

Random forest - Like both the decision trees combined. allows up to 3 types of data sorted
Isolation forest - Like the expert system and sift combined. secret: can be used liked the expert system! just use the color you want and the sqare.
Gradient# - Double the value of each node and uses a small error fix value.