I need someone to help me build a neural network using a free software package that is callable either from C/C++ (visual studio 2010) or python - which will model the following problem: Let's say we have a game with 3 cards types A,B,C (we have more than one of each) Two players play a number of 1000 rounds of this game. Each round each player is dealt one of the cards face up. Cards have the following odds to win the round: A vs B: 10% A vs C: 60% B vs C: 30% A vs A: 50% B vs B: 50% C vs C: 50% so for the dealt cards X vs Y for a round we have P% odds for player with card X to win it. A random number 0..100 is generated and if the number if <= P then playerX wins the round, otherwise playerY wins it. So the training file will consist of the following 0/1 binary variables: X_is_A X_is_B X_is_C Y_is_A Y_is_B Y_is_C X_wins The neural network will have as input a number of 1000 such entries and will have a floating point output 0..1 that for a given entry should in the end approximate odds of cardX vs cardY. So it's like building a neural network for a more complex coinflip based on observed data. I will provide input data and I need someone to show me how to build this thing!