Still, I give in to passion. For example, if I have two moves ago was good to win, but something happened, bad dice fell, and I began to be ill. I'm
just out of inertia continue in his brain to use pessimistic risk
function, evaluating the position, which, of course, not to do. The
program is also faster and faster switches realizes that things are not
going so well as it really is, and the program is switched, for
example, from pessimistic to optimistic risk function is switched much
faster than I am. AR Here,
perhaps, is even noted that the program for the game of backgammon,
which implements these algorithms, but which are not matched numerical
coefficients (when switched to what strategy is, in fact, static
estimate the position, good or bad), this program It is not working. To make it work, it is necessary to teach the game. the game program Education happens when she plays with herself in a
game of backgammon, then there is, in fact, fit the parameters so as to
maximize the quality of the game, to maximize the probability of
winning.
It is better, perhaps, to remember one more thing, which is just beginning to be integrated into the program for the game. In the classical theory Adelson-Belsky program for the game, when he thinks of the enemy is thinking the same way as for themselves. That is, in place of the enemy puts herself. Another technique that we used - to put on the enemy not place itself, but something else, something more complex, something stronger. Because we really have something crowd (it's a jargon - a crowd of players in backgammon), the crowd of objects to self. This applies, I repeat, in other discrete optimization problems. And you can always take the one in the lead in the game, as the imaginary enemy, you have a program, playing a game of backgammon, as the imaginary enemy takes a leader.
What else is possible? It's only started work in that direction, so that the program was trying until the game the enemy thinks, to start thinking of the enemy and the enemy was trying to pick their performance criteria. That is trying to present itself in place of the enemy, and if it is obtained, it is in place of the virtual opponent inserts itself with its coefficients. Here it would be a very interesting topic. But it is, as I said, just started, and say how it really applies to programs that I have not yet mogu.A.G. Well, I have left to wish you good luck in 2017. And if you win in backgammon, then come tell us how it was. BM Thank you!
It is better, perhaps, to remember one more thing, which is just beginning to be integrated into the program for the game. In the classical theory Adelson-Belsky program for the game, when he thinks of the enemy is thinking the same way as for themselves. That is, in place of the enemy puts herself. Another technique that we used - to put on the enemy not place itself, but something else, something more complex, something stronger. Because we really have something crowd (it's a jargon - a crowd of players in backgammon), the crowd of objects to self. This applies, I repeat, in other discrete optimization problems. And you can always take the one in the lead in the game, as the imaginary enemy, you have a program, playing a game of backgammon, as the imaginary enemy takes a leader.
What else is possible? It's only started work in that direction, so that the program was trying until the game the enemy thinks, to start thinking of the enemy and the enemy was trying to pick their performance criteria. That is trying to present itself in place of the enemy, and if it is obtained, it is in place of the virtual opponent inserts itself with its coefficients. Here it would be a very interesting topic. But it is, as I said, just started, and say how it really applies to programs that I have not yet mogu.A.G. Well, I have left to wish you good luck in 2017. And if you win in backgammon, then come tell us how it was. BM Thank you!
No comments:
Post a Comment