Bukkit

1.0.6
Added multi-Scenario-Reinforcement, which should help for non-linear/binary responses
CHANGED WAY NN LEARN. A "learn" method is now called when learning.
CHANGED WAY startLearning.... works. You now need to use startingRepeatinglyLearning.... to repeat the learning.
Added ability to rely on last value for isTriggered. This should reduce the amount of operations done when checking if a neuron is triggered. However, this means you can't change the input on the same tick. Instead, you will need to tick every time you change the input.
Added ability to turn off Thresholds. Best if used for non-binary options of multi-layer NNs
Added activation read-outs for output neurons
Adjusted TriggerStength==0 in grapher so there is an easier distinction between non-triggered neurons and neurons with 0 strength
Added color changes to output neuron text that signifies if the neuron is active, positive, or negitive
Added backPropConnections. In case you want to add new neurons to an existing NN, use this method to connect the neurons from the previous layer to the new neuron.
Added ability to disable negative numbers for neurons, including BiasNeurons and OutputNeurons. Output neurons that do not have negative numbers will have their activation points raised to 0.5

1.0.4
Added osB to reduce the config saves even more.

1.0.3
Added code-reduction for neuron classes. Config files storing NNs should be 1/3rd the size after this update.

1.0.2
Slight change to DRutil. Should increase preformace slightly.

1.0.1
Fixed inputletterneuron.

1.0.0
init