Create a javascript code using ML5.js or brain.js for reinforcement learning with a reward function
Budget: $30 – $250 USD
Only bids that start with the word "Blackjack" are going to be read. If you don't know any AI library in javascript don't waste my time or yours. I want it in javascript and will not change to any other language.
I need a Class code in javascript that I can give how many inputs will be received as also how many outputs will have. I don't want it to have training material as I want it to learn will testing. Like generations. The reward function will be given to know which generation will move. Very similar to how the game of snake is used as an example for ai (ej. https://www.youtube.com/watch?v=zIkBYwdkuTk)
The code will iterate through a multidimensional array on which I can give N numbers of inputs and will give me -1, 0, or 1 as the output but I want it to be able to have another variation of the code with only 2 outputs 1 or 0. I have already on each of the loop indexes what was the correct answer.
for the reward function:
If the output by the ai was "any" and the correct answer is 0 no point was given.
If the output by the ai was "1" and the correct answer is 1, a point is given else if the correct answer is -1 a point is subtracted.
If the output by the ai was "-1" and the correct answer is -1, a point is given else if the correct answer is 1 a point is subtracted.
Any modification or suggestions is accepted as long as the end product works for what we want. We don't want something too elaborate.
I need a Class code in javascript that I can give how many inputs will be received as also how many outputs will have. I don't want it to have training material as I want it to learn will testing. Like generations. The reward function will be given to know which generation will move. Very similar to how the game of snake is used as an example for ai (ej. https://www.youtube.com/watch?v=zIkBYwdkuTk)
The code will iterate through a multidimensional array on which I can give N numbers of inputs and will give me -1, 0, or 1 as the output but I want it to be able to have another variation of the code with only 2 outputs 1 or 0. I have already on each of the loop indexes what was the correct answer.
for the reward function:
If the output by the ai was "any" and the correct answer is 0 no point was given.
If the output by the ai was "1" and the correct answer is 1, a point is given else if the correct answer is -1 a point is subtracted.
If the output by the ai was "-1" and the correct answer is -1, a point is given else if the correct answer is 1 a point is subtracted.
Any modification or suggestions is accepted as long as the end product works for what we want. We don't want something too elaborate.