From 2f64f1702669f55a6265c027443aae19d53bbc04 Mon Sep 17 00:00:00 2001 From: Philipp Horstenkamp Date: Fri, 31 Mar 2023 21:58:54 +0200 Subject: [PATCH] First draft. --- main.ipynb | 885 +++++++++++++++++++++++++++++++++++++++++++++-------- 1 file changed, 757 insertions(+), 128 deletions(-) diff --git a/main.ipynb b/main.ipynb index 965eb77..3bb910a 100644 --- a/main.ipynb +++ b/main.ipynb @@ -2829,11 +2829,14 @@ { "cell_type": "markdown", "metadata": {}, - "source": [] + "source": [ + "When using the Q-Learning networks as defined above in the sections `DQLNet` and `DQLSimple` the state and the actions need to be converted into the input format accepted by the Q-Learning network.\n", + "The function below merges both input values and gets them into the proper format." + ] }, { "cell_type": "code", - "execution_count": 57, + "execution_count": 326, "metadata": { "tags": [] }, @@ -2842,8 +2845,8 @@ "name": "stdout", "output_type": "stream", "text": [ - "7.54 ms ± 1.07 ms per loop (mean ± std. dev. of 7 runs, 100 loops each)\n", - "peak memory: 446.80 MiB, increment: 0.02 MiB\n" + "37.2 ms ± 1.67 ms per loop (mean ± std. dev. of 7 runs, 10 loops each)\n", + "peak memory: 1014.62 MiB, increment: 0.23 MiB\n" ] } ], @@ -2851,7 +2854,12 @@ "def action_to_q_learning_format(\n", " board_history: np.ndarray, action_history: np.ndarray\n", ") -> np.ndarray:\n", - " \"\"\" \"\"\"\n", + " \"\"\"Fomrats the board history and the action history into Q-Learning inputs.\n", + "\n", + " Args:\n", + " board_history: A stack of board histories.\n", + " action_history: A stack of action histories.\n", + " \"\"\"\n", " q_learning_format = np.zeros(\n", " (SIMULATE_TURNS, board_history.shape[1], 2, 8, 8), dtype=float\n", " )\n", @@ -2946,7 +2954,9 @@ { "cell_type": "markdown", "metadata": {}, - "source": [] + "source": [ + "During the training of an ANN, it is necessary to monitor its progress by tracking its loss. Additional metrics are also used to measure significant results separately, since we work with a composite reward in reinforcement learning. Therefore, the final score and win ratio are tracked and can be plotted using the function below." + ] }, { "cell_type": "code", @@ -3383,7 +3393,7 @@ "metadata": {}, "source": [ "### Defining some polices\n", - "In varius stages of this project varius Policies where created. Those differ in Exploartion facotr" + "In varius stages of this project varius Policies where created. Those compose of different values for exploration, reword fractions and gamma values and different network structures." ] }, { @@ -3875,6 +3885,13 @@ " ql_policy.train(1, 10, 1000, 250, constant_metric_policies)" ] }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "Below the training pogress for the Q-Learning policy is shown." + ] + }, { "cell_type": "code", "execution_count": 78, @@ -4022,6 +4039,25 @@ " ql_policy.plot_history()" ] }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "The training consistency varies from policy to policy, with some policies being more stable and exhibiting less variation. Moreover, the highly interdependent metrics indicate that the evaluation sample size may be too small.\n", + "\n", + "Since batch size and metics have similar sizes this suggest that the batch size should also be increads. Seadly with the current Q-Learning setup this is not possible due to CPU constraints." + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "### Example simulations\n", + "\n", + "The section below shows the trained policies, playing as black against a white oponent.\n", + "The Policy can be choosen via the dropdown menu." + ] + }, { "cell_type": "code", "execution_count": 225, @@ -4056,7 +4092,7 @@ " simulation_results[0].reshape(-1, 8, 8), simulation_results[1].reshape(-1, 2)\n", " )\n", " plot_othello_boards(_unique_bords, actions=_unique_actions)\n", - " plt.suptitle(str(ql_policy))" + " plt.show()" ] }, { @@ -4773,9 +4809,9 @@ "cell_type": "markdown", "metadata": {}, "source": [ - "To reduce the bias from the matrix we build an perspectivly correct mean by subtracted its transposed version and dividing by tow. This will compensate for the easier time the black policy has.\n", - "The resulting values are not all positve. Some policies now reach a negative score in their average win sigunum. This means that a policy will win 60% of the games when a score of 0.1 is reached.\n", - "Since most of the games played against the random and the greedy policy have a positive score we conclude that some learning did hapen. But it stayed a lot behind what was expected. A simple policy weighting each stone by its position should be much more effective. Especially if the value trends to 1 at the end of the game. This seadly fits with the observations made above." + "To mitigate the bias in the matrix, we calculate a perspective-correct mean by subtracting its transposed version and dividing by two. This compensates for the black policy's relative advantage. However, the resulting values may not all be positive, as some policies may now have a negative average win signum. Specifically, a policy with a win signum of 0.1 will win 60% of the games.\n", + "\n", + "Despite this, we observe that learning did occur, as most games played against the random and greedy policies resulted in a positive score. However, the degree of learning fell far short of our expectations. A simple policy that weights each stone by its position may be much more effective, particularly if the value trends towards 1 towards the end of the game. This aligns with our earlier observations." ] }, { @@ -5398,9 +5434,599 @@ }, { "cell_type": "code", - "execution_count": null, + "execution_count": 327, "metadata": {}, - "outputs": [], + "outputs": [ + { + "data": { + "text/html": [ + "
\n", + "\n", + "\n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + "
White policyrandomgreedy_policyQL-M-G09-WW00-FSF10-DQLSimple-MSELossQL-M-G08-WW00-FSF10-DQLSimple-MSELossQL-M-G10-WW00-FSF10-DQLSimple-MSELossQL-M-G09-WW10-FSF00-DQLSimple-MSELossQL-M-G09-WW03-FSF03-DQLSimple-MSELossQL-M-G09-WW03-FSF07-DQLSimple-MSELossQL-M-G09-WW02-FSF07-DQLSimple-MSELossQL-M-G08-WW00-FSF10-DQLNet-MSELossQL-M-G10-WW00-FSF10-DQLNet-MSELossQL-M-G09-WW10-FSF00-DQLNet-MSELossQL-M-G09-WW03-FSF03-DQLNet-MSELossQL-M-G09-WW03-FSF07-DQLNet-MSELossQL-M-G09-WW02-FSF07-DQLNet-MSELoss
Black policy
random0.000-7.828-6.020-6.652-3.832-3.400-8.016-6.296-5.636-0.800-4.208-8.928-5.372-2.976-6.084
greedy_policy7.8280.0004.0283.0840.4727.428-2.200-0.3922.7524.5602.2920.2164.7481.4040.240
QL-M-G09-WW00-FSF10-DQLSimple-MSELoss6.020-4.0280.000-14.176-2.912-11.464-5.1769.160-12.1360.924-14.460-13.6081.91213.204-3.764
QL-M-G08-WW00-FSF10-DQLSimple-MSELoss6.652-3.08414.1760.0003.460-7.60010.820-20.352-9.22812.09212.940-13.20412.21212.672-16.348
QL-M-G10-WW00-FSF10-DQLSimple-MSELoss3.832-0.4722.912-3.4600.0009.660-4.916-7.84017.62811.796-1.080-17.580-15.2646.1689.144
QL-M-G09-WW10-FSF00-DQLSimple-MSELoss3.400-7.42811.4647.600-9.6600.000-1.2401.8328.0361.02814.328-8.7561.544-1.692-9.496
QL-M-G09-WW03-FSF03-DQLSimple-MSELoss8.0162.2005.176-10.8204.9161.2400.0002.34011.26813.25612.028-7.292-8.53611.360-7.336
QL-M-G09-WW03-FSF07-DQLSimple-MSELoss6.2960.392-9.16020.3527.840-1.832-2.3400.0005.0207.676-7.7724.564-9.4920.448-2.552
QL-M-G09-WW02-FSF07-DQLSimple-MSELoss5.636-2.75212.1369.228-17.628-8.036-11.268-5.0200.0007.7209.91612.032-24.1009.61218.056
QL-M-G08-WW00-FSF10-DQLNet-MSELoss0.800-4.560-0.924-12.092-11.796-1.028-13.256-7.676-7.7200.000-8.476-8.544-6.116-1.38014.544
QL-M-G10-WW00-FSF10-DQLNet-MSELoss4.208-2.29214.460-12.9401.080-14.328-12.0287.772-9.9168.4760.0001.476-1.8765.3282.228
QL-M-G09-WW10-FSF00-DQLNet-MSELoss8.928-0.21613.60813.20417.5808.7567.292-4.564-12.0328.544-1.4760.0005.76815.26818.408
QL-M-G09-WW03-FSF03-DQLNet-MSELoss5.372-4.748-1.912-12.21215.264-1.5448.5369.49224.1006.1161.876-5.7680.00027.580-14.512
QL-M-G09-WW03-FSF07-DQLNet-MSELoss2.976-1.404-13.204-12.672-6.1681.692-11.360-0.448-9.6121.380-5.328-15.268-27.5800.000-1.792
QL-M-G09-WW02-FSF07-DQLNet-MSELoss6.084-0.2403.76416.348-9.1449.4967.3362.552-18.056-14.544-2.228-18.40814.5121.7920.000
\n", + "
" + ], + "text/plain": [ + "White policy random greedy_policy \\\n", + "Black policy \n", + "random 0.000 -7.828 \n", + "greedy_policy 7.828 0.000 \n", + "QL-M-G09-WW00-FSF10-DQLSimple-MSELoss 6.020 -4.028 \n", + "QL-M-G08-WW00-FSF10-DQLSimple-MSELoss 6.652 -3.084 \n", + "QL-M-G10-WW00-FSF10-DQLSimple-MSELoss 3.832 -0.472 \n", + "QL-M-G09-WW10-FSF00-DQLSimple-MSELoss 3.400 -7.428 \n", + "QL-M-G09-WW03-FSF03-DQLSimple-MSELoss 8.016 2.200 \n", + "QL-M-G09-WW03-FSF07-DQLSimple-MSELoss 6.296 0.392 \n", + "QL-M-G09-WW02-FSF07-DQLSimple-MSELoss 5.636 -2.752 \n", + "QL-M-G08-WW00-FSF10-DQLNet-MSELoss 0.800 -4.560 \n", + "QL-M-G10-WW00-FSF10-DQLNet-MSELoss 4.208 -2.292 \n", + "QL-M-G09-WW10-FSF00-DQLNet-MSELoss 8.928 -0.216 \n", + "QL-M-G09-WW03-FSF03-DQLNet-MSELoss 5.372 -4.748 \n", + "QL-M-G09-WW03-FSF07-DQLNet-MSELoss 2.976 -1.404 \n", + "QL-M-G09-WW02-FSF07-DQLNet-MSELoss 6.084 -0.240 \n", + "\n", + "White policy QL-M-G09-WW00-FSF10-DQLSimple-MSELoss \\\n", + "Black policy \n", + "random -6.020 \n", + "greedy_policy 4.028 \n", + "QL-M-G09-WW00-FSF10-DQLSimple-MSELoss 0.000 \n", + "QL-M-G08-WW00-FSF10-DQLSimple-MSELoss 14.176 \n", + "QL-M-G10-WW00-FSF10-DQLSimple-MSELoss 2.912 \n", + "QL-M-G09-WW10-FSF00-DQLSimple-MSELoss 11.464 \n", + "QL-M-G09-WW03-FSF03-DQLSimple-MSELoss 5.176 \n", + "QL-M-G09-WW03-FSF07-DQLSimple-MSELoss -9.160 \n", + "QL-M-G09-WW02-FSF07-DQLSimple-MSELoss 12.136 \n", + "QL-M-G08-WW00-FSF10-DQLNet-MSELoss -0.924 \n", + "QL-M-G10-WW00-FSF10-DQLNet-MSELoss 14.460 \n", + "QL-M-G09-WW10-FSF00-DQLNet-MSELoss 13.608 \n", + "QL-M-G09-WW03-FSF03-DQLNet-MSELoss -1.912 \n", + "QL-M-G09-WW03-FSF07-DQLNet-MSELoss -13.204 \n", + "QL-M-G09-WW02-FSF07-DQLNet-MSELoss 3.764 \n", + "\n", + "White policy QL-M-G08-WW00-FSF10-DQLSimple-MSELoss \\\n", + "Black policy \n", + "random -6.652 \n", + "greedy_policy 3.084 \n", + "QL-M-G09-WW00-FSF10-DQLSimple-MSELoss -14.176 \n", + "QL-M-G08-WW00-FSF10-DQLSimple-MSELoss 0.000 \n", + "QL-M-G10-WW00-FSF10-DQLSimple-MSELoss -3.460 \n", + "QL-M-G09-WW10-FSF00-DQLSimple-MSELoss 7.600 \n", + "QL-M-G09-WW03-FSF03-DQLSimple-MSELoss -10.820 \n", + "QL-M-G09-WW03-FSF07-DQLSimple-MSELoss 20.352 \n", + "QL-M-G09-WW02-FSF07-DQLSimple-MSELoss 9.228 \n", + "QL-M-G08-WW00-FSF10-DQLNet-MSELoss -12.092 \n", + "QL-M-G10-WW00-FSF10-DQLNet-MSELoss -12.940 \n", + "QL-M-G09-WW10-FSF00-DQLNet-MSELoss 13.204 \n", + "QL-M-G09-WW03-FSF03-DQLNet-MSELoss -12.212 \n", + "QL-M-G09-WW03-FSF07-DQLNet-MSELoss -12.672 \n", + "QL-M-G09-WW02-FSF07-DQLNet-MSELoss 16.348 \n", + "\n", + "White policy QL-M-G10-WW00-FSF10-DQLSimple-MSELoss \\\n", + "Black policy \n", + "random -3.832 \n", + "greedy_policy 0.472 \n", + "QL-M-G09-WW00-FSF10-DQLSimple-MSELoss -2.912 \n", + "QL-M-G08-WW00-FSF10-DQLSimple-MSELoss 3.460 \n", + "QL-M-G10-WW00-FSF10-DQLSimple-MSELoss 0.000 \n", + "QL-M-G09-WW10-FSF00-DQLSimple-MSELoss -9.660 \n", + "QL-M-G09-WW03-FSF03-DQLSimple-MSELoss 4.916 \n", + "QL-M-G09-WW03-FSF07-DQLSimple-MSELoss 7.840 \n", + "QL-M-G09-WW02-FSF07-DQLSimple-MSELoss -17.628 \n", + "QL-M-G08-WW00-FSF10-DQLNet-MSELoss -11.796 \n", + "QL-M-G10-WW00-FSF10-DQLNet-MSELoss 1.080 \n", + "QL-M-G09-WW10-FSF00-DQLNet-MSELoss 17.580 \n", + "QL-M-G09-WW03-FSF03-DQLNet-MSELoss 15.264 \n", + "QL-M-G09-WW03-FSF07-DQLNet-MSELoss -6.168 \n", + "QL-M-G09-WW02-FSF07-DQLNet-MSELoss -9.144 \n", + "\n", + "White policy QL-M-G09-WW10-FSF00-DQLSimple-MSELoss \\\n", + "Black policy \n", + "random -3.400 \n", + "greedy_policy 7.428 \n", + "QL-M-G09-WW00-FSF10-DQLSimple-MSELoss -11.464 \n", + "QL-M-G08-WW00-FSF10-DQLSimple-MSELoss -7.600 \n", + "QL-M-G10-WW00-FSF10-DQLSimple-MSELoss 9.660 \n", + "QL-M-G09-WW10-FSF00-DQLSimple-MSELoss 0.000 \n", + "QL-M-G09-WW03-FSF03-DQLSimple-MSELoss 1.240 \n", + "QL-M-G09-WW03-FSF07-DQLSimple-MSELoss -1.832 \n", + "QL-M-G09-WW02-FSF07-DQLSimple-MSELoss -8.036 \n", + "QL-M-G08-WW00-FSF10-DQLNet-MSELoss -1.028 \n", + "QL-M-G10-WW00-FSF10-DQLNet-MSELoss -14.328 \n", + "QL-M-G09-WW10-FSF00-DQLNet-MSELoss 8.756 \n", + "QL-M-G09-WW03-FSF03-DQLNet-MSELoss -1.544 \n", + "QL-M-G09-WW03-FSF07-DQLNet-MSELoss 1.692 \n", + "QL-M-G09-WW02-FSF07-DQLNet-MSELoss 9.496 \n", + "\n", + "White policy QL-M-G09-WW03-FSF03-DQLSimple-MSELoss \\\n", + "Black policy \n", + "random -8.016 \n", + "greedy_policy -2.200 \n", + "QL-M-G09-WW00-FSF10-DQLSimple-MSELoss -5.176 \n", + "QL-M-G08-WW00-FSF10-DQLSimple-MSELoss 10.820 \n", + "QL-M-G10-WW00-FSF10-DQLSimple-MSELoss -4.916 \n", + "QL-M-G09-WW10-FSF00-DQLSimple-MSELoss -1.240 \n", + "QL-M-G09-WW03-FSF03-DQLSimple-MSELoss 0.000 \n", + "QL-M-G09-WW03-FSF07-DQLSimple-MSELoss -2.340 \n", + "QL-M-G09-WW02-FSF07-DQLSimple-MSELoss -11.268 \n", + "QL-M-G08-WW00-FSF10-DQLNet-MSELoss -13.256 \n", + "QL-M-G10-WW00-FSF10-DQLNet-MSELoss -12.028 \n", + "QL-M-G09-WW10-FSF00-DQLNet-MSELoss 7.292 \n", + "QL-M-G09-WW03-FSF03-DQLNet-MSELoss 8.536 \n", + "QL-M-G09-WW03-FSF07-DQLNet-MSELoss -11.360 \n", + "QL-M-G09-WW02-FSF07-DQLNet-MSELoss 7.336 \n", + "\n", + "White policy QL-M-G09-WW03-FSF07-DQLSimple-MSELoss \\\n", + "Black policy \n", + "random -6.296 \n", + "greedy_policy -0.392 \n", + "QL-M-G09-WW00-FSF10-DQLSimple-MSELoss 9.160 \n", + "QL-M-G08-WW00-FSF10-DQLSimple-MSELoss -20.352 \n", + "QL-M-G10-WW00-FSF10-DQLSimple-MSELoss -7.840 \n", + "QL-M-G09-WW10-FSF00-DQLSimple-MSELoss 1.832 \n", + "QL-M-G09-WW03-FSF03-DQLSimple-MSELoss 2.340 \n", + "QL-M-G09-WW03-FSF07-DQLSimple-MSELoss 0.000 \n", + "QL-M-G09-WW02-FSF07-DQLSimple-MSELoss -5.020 \n", + "QL-M-G08-WW00-FSF10-DQLNet-MSELoss -7.676 \n", + "QL-M-G10-WW00-FSF10-DQLNet-MSELoss 7.772 \n", + "QL-M-G09-WW10-FSF00-DQLNet-MSELoss -4.564 \n", + "QL-M-G09-WW03-FSF03-DQLNet-MSELoss 9.492 \n", + "QL-M-G09-WW03-FSF07-DQLNet-MSELoss -0.448 \n", + "QL-M-G09-WW02-FSF07-DQLNet-MSELoss 2.552 \n", + "\n", + "White policy QL-M-G09-WW02-FSF07-DQLSimple-MSELoss \\\n", + "Black policy \n", + "random -5.636 \n", + "greedy_policy 2.752 \n", + "QL-M-G09-WW00-FSF10-DQLSimple-MSELoss -12.136 \n", + "QL-M-G08-WW00-FSF10-DQLSimple-MSELoss -9.228 \n", + "QL-M-G10-WW00-FSF10-DQLSimple-MSELoss 17.628 \n", + "QL-M-G09-WW10-FSF00-DQLSimple-MSELoss 8.036 \n", + "QL-M-G09-WW03-FSF03-DQLSimple-MSELoss 11.268 \n", + "QL-M-G09-WW03-FSF07-DQLSimple-MSELoss 5.020 \n", + "QL-M-G09-WW02-FSF07-DQLSimple-MSELoss 0.000 \n", + "QL-M-G08-WW00-FSF10-DQLNet-MSELoss -7.720 \n", + "QL-M-G10-WW00-FSF10-DQLNet-MSELoss -9.916 \n", + "QL-M-G09-WW10-FSF00-DQLNet-MSELoss -12.032 \n", + "QL-M-G09-WW03-FSF03-DQLNet-MSELoss 24.100 \n", + "QL-M-G09-WW03-FSF07-DQLNet-MSELoss -9.612 \n", + "QL-M-G09-WW02-FSF07-DQLNet-MSELoss -18.056 \n", + "\n", + "White policy QL-M-G08-WW00-FSF10-DQLNet-MSELoss \\\n", + "Black policy \n", + "random -0.800 \n", + "greedy_policy 4.560 \n", + "QL-M-G09-WW00-FSF10-DQLSimple-MSELoss 0.924 \n", + "QL-M-G08-WW00-FSF10-DQLSimple-MSELoss 12.092 \n", + "QL-M-G10-WW00-FSF10-DQLSimple-MSELoss 11.796 \n", + "QL-M-G09-WW10-FSF00-DQLSimple-MSELoss 1.028 \n", + "QL-M-G09-WW03-FSF03-DQLSimple-MSELoss 13.256 \n", + "QL-M-G09-WW03-FSF07-DQLSimple-MSELoss 7.676 \n", + "QL-M-G09-WW02-FSF07-DQLSimple-MSELoss 7.720 \n", + "QL-M-G08-WW00-FSF10-DQLNet-MSELoss 0.000 \n", + "QL-M-G10-WW00-FSF10-DQLNet-MSELoss 8.476 \n", + "QL-M-G09-WW10-FSF00-DQLNet-MSELoss 8.544 \n", + "QL-M-G09-WW03-FSF03-DQLNet-MSELoss 6.116 \n", + "QL-M-G09-WW03-FSF07-DQLNet-MSELoss 1.380 \n", + "QL-M-G09-WW02-FSF07-DQLNet-MSELoss -14.544 \n", + "\n", + "White policy QL-M-G10-WW00-FSF10-DQLNet-MSELoss \\\n", + "Black policy \n", + "random -4.208 \n", + "greedy_policy 2.292 \n", + "QL-M-G09-WW00-FSF10-DQLSimple-MSELoss -14.460 \n", + "QL-M-G08-WW00-FSF10-DQLSimple-MSELoss 12.940 \n", + "QL-M-G10-WW00-FSF10-DQLSimple-MSELoss -1.080 \n", + "QL-M-G09-WW10-FSF00-DQLSimple-MSELoss 14.328 \n", + "QL-M-G09-WW03-FSF03-DQLSimple-MSELoss 12.028 \n", + "QL-M-G09-WW03-FSF07-DQLSimple-MSELoss -7.772 \n", + "QL-M-G09-WW02-FSF07-DQLSimple-MSELoss 9.916 \n", + "QL-M-G08-WW00-FSF10-DQLNet-MSELoss -8.476 \n", + "QL-M-G10-WW00-FSF10-DQLNet-MSELoss 0.000 \n", + "QL-M-G09-WW10-FSF00-DQLNet-MSELoss -1.476 \n", + "QL-M-G09-WW03-FSF03-DQLNet-MSELoss 1.876 \n", + "QL-M-G09-WW03-FSF07-DQLNet-MSELoss -5.328 \n", + "QL-M-G09-WW02-FSF07-DQLNet-MSELoss -2.228 \n", + "\n", + "White policy QL-M-G09-WW10-FSF00-DQLNet-MSELoss \\\n", + "Black policy \n", + "random -8.928 \n", + "greedy_policy 0.216 \n", + "QL-M-G09-WW00-FSF10-DQLSimple-MSELoss -13.608 \n", + "QL-M-G08-WW00-FSF10-DQLSimple-MSELoss -13.204 \n", + "QL-M-G10-WW00-FSF10-DQLSimple-MSELoss -17.580 \n", + "QL-M-G09-WW10-FSF00-DQLSimple-MSELoss -8.756 \n", + "QL-M-G09-WW03-FSF03-DQLSimple-MSELoss -7.292 \n", + "QL-M-G09-WW03-FSF07-DQLSimple-MSELoss 4.564 \n", + "QL-M-G09-WW02-FSF07-DQLSimple-MSELoss 12.032 \n", + "QL-M-G08-WW00-FSF10-DQLNet-MSELoss -8.544 \n", + "QL-M-G10-WW00-FSF10-DQLNet-MSELoss 1.476 \n", + "QL-M-G09-WW10-FSF00-DQLNet-MSELoss 0.000 \n", + "QL-M-G09-WW03-FSF03-DQLNet-MSELoss -5.768 \n", + "QL-M-G09-WW03-FSF07-DQLNet-MSELoss -15.268 \n", + "QL-M-G09-WW02-FSF07-DQLNet-MSELoss -18.408 \n", + "\n", + "White policy QL-M-G09-WW03-FSF03-DQLNet-MSELoss \\\n", + "Black policy \n", + "random -5.372 \n", + "greedy_policy 4.748 \n", + "QL-M-G09-WW00-FSF10-DQLSimple-MSELoss 1.912 \n", + "QL-M-G08-WW00-FSF10-DQLSimple-MSELoss 12.212 \n", + "QL-M-G10-WW00-FSF10-DQLSimple-MSELoss -15.264 \n", + "QL-M-G09-WW10-FSF00-DQLSimple-MSELoss 1.544 \n", + "QL-M-G09-WW03-FSF03-DQLSimple-MSELoss -8.536 \n", + "QL-M-G09-WW03-FSF07-DQLSimple-MSELoss -9.492 \n", + "QL-M-G09-WW02-FSF07-DQLSimple-MSELoss -24.100 \n", + "QL-M-G08-WW00-FSF10-DQLNet-MSELoss -6.116 \n", + "QL-M-G10-WW00-FSF10-DQLNet-MSELoss -1.876 \n", + "QL-M-G09-WW10-FSF00-DQLNet-MSELoss 5.768 \n", + "QL-M-G09-WW03-FSF03-DQLNet-MSELoss 0.000 \n", + "QL-M-G09-WW03-FSF07-DQLNet-MSELoss -27.580 \n", + "QL-M-G09-WW02-FSF07-DQLNet-MSELoss 14.512 \n", + "\n", + "White policy QL-M-G09-WW03-FSF07-DQLNet-MSELoss \\\n", + "Black policy \n", + "random -2.976 \n", + "greedy_policy 1.404 \n", + "QL-M-G09-WW00-FSF10-DQLSimple-MSELoss 13.204 \n", + "QL-M-G08-WW00-FSF10-DQLSimple-MSELoss 12.672 \n", + "QL-M-G10-WW00-FSF10-DQLSimple-MSELoss 6.168 \n", + "QL-M-G09-WW10-FSF00-DQLSimple-MSELoss -1.692 \n", + "QL-M-G09-WW03-FSF03-DQLSimple-MSELoss 11.360 \n", + "QL-M-G09-WW03-FSF07-DQLSimple-MSELoss 0.448 \n", + "QL-M-G09-WW02-FSF07-DQLSimple-MSELoss 9.612 \n", + "QL-M-G08-WW00-FSF10-DQLNet-MSELoss -1.380 \n", + "QL-M-G10-WW00-FSF10-DQLNet-MSELoss 5.328 \n", + "QL-M-G09-WW10-FSF00-DQLNet-MSELoss 15.268 \n", + "QL-M-G09-WW03-FSF03-DQLNet-MSELoss 27.580 \n", + "QL-M-G09-WW03-FSF07-DQLNet-MSELoss 0.000 \n", + "QL-M-G09-WW02-FSF07-DQLNet-MSELoss 1.792 \n", + "\n", + "White policy QL-M-G09-WW02-FSF07-DQLNet-MSELoss \n", + "Black policy \n", + "random -6.084 \n", + "greedy_policy 0.240 \n", + "QL-M-G09-WW00-FSF10-DQLSimple-MSELoss -3.764 \n", + "QL-M-G08-WW00-FSF10-DQLSimple-MSELoss -16.348 \n", + "QL-M-G10-WW00-FSF10-DQLSimple-MSELoss 9.144 \n", + "QL-M-G09-WW10-FSF00-DQLSimple-MSELoss -9.496 \n", + "QL-M-G09-WW03-FSF03-DQLSimple-MSELoss -7.336 \n", + "QL-M-G09-WW03-FSF07-DQLSimple-MSELoss -2.552 \n", + "QL-M-G09-WW02-FSF07-DQLSimple-MSELoss 18.056 \n", + "QL-M-G08-WW00-FSF10-DQLNet-MSELoss 14.544 \n", + "QL-M-G10-WW00-FSF10-DQLNet-MSELoss 2.228 \n", + "QL-M-G09-WW10-FSF00-DQLNet-MSELoss 18.408 \n", + "QL-M-G09-WW03-FSF03-DQLNet-MSELoss -14.512 \n", + "QL-M-G09-WW03-FSF07-DQLNet-MSELoss -1.792 \n", + "QL-M-G09-WW02-FSF07-DQLNet-MSELoss 0.000 " + ] + }, + "execution_count": 327, + "metadata": {}, + "output_type": "execute_result" + } + ], "source": [ "df = result_df.applymap(lambda x: x[\"final_score\"])\n", "(df - df.T) / 2 * 64" @@ -5410,7 +6036,7 @@ "cell_type": "markdown", "metadata": {}, "source": [ - "Also here there is no clear best policy to identify." + "There is no clear best policy to identify." ] }, { @@ -5419,7 +6045,7 @@ "source": [ "### Analysing the corner behavior\n", "\n", - "As notet in the analysis of the game a heat map can be used to calculate how a game was won compared to a random strategy." + "Given the importance of corner capture in Othello strategies, it is worth investigating if the Policies discover this on their own. To analyze this behavior, we selected a policy and measured the availability of corners, the probability of a corner stone being available, and the number of corners captured by the AI when playing against a random player." ] }, { @@ -5489,34 +6115,6 @@ ")" ] }, - { - "cell_type": "code", - "execution_count": 169, - "metadata": { - "tags": [] - }, - "outputs": [ - { - "data": { - "text/plain": [ - "((70, 1000, 8, 8), (70, 1000, 8, 8))" - ] - }, - "execution_count": 169, - "metadata": {}, - "output_type": "execute_result" - } - ], - "source": [ - "_l_actions_possible = get_possible_turns(_l_board_history.reshape(-1, 8, 8)).reshape(\n", - " 70, -1, 8, 8\n", - ")\n", - "_r_actions_possible = get_possible_turns(_r_board_history.reshape(-1, 8, 8)).reshape(\n", - " 70, -1, 8, 8\n", - ")\n", - "_l_actions_possible.shape, _r_actions_possible.shape" - ] - }, { "cell_type": "code", "execution_count": 170, @@ -5536,35 +6134,24 @@ } ], "source": [ + "_l_actions_possible = get_possible_turns(_l_board_history.reshape(-1, 8, 8)).reshape(\n", + " 70, -1, 8, 8\n", + ")\n", + "_r_actions_possible = get_possible_turns(_r_board_history.reshape(-1, 8, 8)).reshape(\n", + " 70, -1, 8, 8\n", + ")\n", + "\n", + "\n", "_l_mean_actions_possible = np.mean(_l_actions_possible, axis=(1))\n", - "_r_mean_actions_possible = np.mean(_r_actions_possible, axis=(1))\n", - "_l_mean_actions_possible.shape, _r_mean_actions_possible.shape" + "_r_mean_actions_possible = np.mean(_r_actions_possible, axis=(1))" ] }, { - "cell_type": "code", - "execution_count": 173, - "metadata": { - "tags": [] - }, - "outputs": [ - { - "data": { - "text/plain": [ - "((8, 8), (1000, 8, 8), (1000, 8, 8))" - ] - }, - "execution_count": 173, - "metadata": {}, - "output_type": "execute_result" - } - ], + "cell_type": "markdown", + "metadata": {}, "source": [ - "turn = 0\n", - "l_turn_possibility_on_field = _l_actions_possible[turn]\n", - "r_turn_possibility_on_field = _r_actions_possible[turn]\n", - "turn_possibility_on_field = mean_poss_turn[turn]\n", - "turn_possibility_on_field.shape, l_turn_possibility_on_field.shape, r_turn_possibility_on_field.shape" + "#### Corner stone analysis via possiblity space\n", + "As notet in the analysis of the game and its behavior over time a heat map can be used to calculate how a game was won compared to a random strategy." ] }, { @@ -5642,16 +6229,30 @@ " plt.tight_layout()" ] }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "The analysis reveals that the chance of capturing a cornerstone is significantly higher when playing against a random player compared to playing against the trained policy. This is because the trained policy understands the importance of cornerstones and tends to capture them faster, while also avoiding giving the same opportunity to the opponent." + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "Another way to demonstrate this is by visualizing the mean chance of capturing one of the four cornerstones over time, as shown in the plot below. This provides a compact representation of the trend over time." + ] + }, { "cell_type": "code", - "execution_count": 231, + "execution_count": 292, "metadata": { "tags": [] }, "outputs": [ { "data": { - "image/png": "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\n", + "image/png": "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\n", "text/plain": [ "
" ] @@ -5661,9 +6262,9 @@ } ], "source": [ - "a = np.mean(_l_mean_actions_possible[:, [1, -1]][:, :, [1, -1]], axis=(1, 2))\n", - "b = np.mean(_r_mean_actions_possible[:, [1, -1]][:, :, [1, -1]], axis=(1, 2))\n", - "c = np.mean(mean_poss_turn[:, [1, -1]][:, :, [1, -1]], axis=(1, 2))\n", + "a = np.mean(_l_mean_actions_possible[:, [0, -1]][:, :, [0, -1]], axis=(1, 2))\n", + "b = np.mean(_r_mean_actions_possible[:, [0, -1]][:, :, [0, -1]], axis=(1, 2))\n", + "c = np.mean(mean_poss_turn[:, [0, -1]][:, :, [0, -1]], axis=(1, 2))\n", "\n", "df = pd.DataFrame(\n", " [a, b, c],\n", @@ -5674,79 +6275,101 @@ { "cell_type": "markdown", "metadata": {}, - "source": [] + "source": [ + "#### Corner stone capture count\n", + "When counting how many corners where captured it is clear to see that the QL policy understood the importance of the corners somewhat. Since it is not absolut is can't be saad that they should overly priorities those and i would have expexted an outkome around $3$. The simulated $2.7$ corners as an average at the end of the game is not that high but shows a good trend." + ] }, { "cell_type": "code", - "execution_count": 277, + "execution_count": 328, "metadata": { "tags": [] }, "outputs": [ { "data": { - "text/html": [ - "
\n", - "\n", - "\n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - "
QL-Policy vs. RandomRandom vs. QL-PolicyRandom vs. Random
Average number of stones2.2522.1281.871
\n", - "
" - ], "text/plain": [ - " QL-Policy vs. Random Random vs. QL-Policy \\\n", - "Average number of stones 2.252 2.128 \n", - "\n", - " Random vs. Random \n", - "Average number of stones 1.871 " + "QL-Policy vs. Random 2.748\n", + "Random vs. QL-Policy 2.507\n", + "Random vs. Random 1.952\n", + "dtype: float64" ] }, - "execution_count": 277, + "execution_count": 328, "metadata": {}, "output_type": "execute_result" } ], "source": [ - "a = np.sum(np.mean(_l_board_history[-1][:, [1, -1]][:, :, [1, -1]], axis=(0)))\n", - "b = -np.sum(np.mean(_r_board_history[-1][:, [1, -1]][:, :, [1, -1]], axis=(0)))\n", - "c = np.sum(np.mean(_board_history[-1][:, [1, -1]][:, :, [1, -1]], axis=(0)))\n", + "a = np.sum(np.mean(_l_board_history[-1][:, [0, -1]][:, :, [0, -1]], axis=(0)))\n", + "b = -np.sum(np.mean(_r_board_history[-1][:, [0, -1]][:, :, [0, -1]], axis=(0)))\n", + "c = np.sum(np.mean(_board_history[-1][:, [0, -1]][:, :, [0, -1]], axis=(0)))\n", "\n", - "df = (\n", - " pd.DataFrame(\n", + "(\n", + " pd.Series(\n", " [a, b, c],\n", " index=[\"QL-Policy vs. Random\", \"Random vs. QL-Policy\", \"Random vs. Random\"],\n", - " columns=[\"Average number of stones\"],\n", - " ).T\n", + " )\n", " + 2\n", - ")\n", - "df" + ")" + ] + }, + { + "cell_type": "markdown", + "metadata": { + "tags": [] + }, + "source": [ + "### Time till capture\n", + "\n", + "An additional aspect to analyze is the time it takes to capture a corner when it becomes available. As capturing a corner is critical in Othello strategy, it is expected that the AI would priorites its capture. However, since the opponent is usually not able to capture the same corner at the same moment, the AI may not need to hurry. To test this hypothesis, we measured the time it took for the AI to capture a corner and compared it to the time it took for the AI to capture a non-corner cell. The results showed that the AI captured corners faster, although not by as much as expected." + ] + }, + { + "cell_type": "code", + "execution_count": 322, + "metadata": { + "tags": [] + }, + "outputs": [ + { + "data": { + "text/plain": [ + "QL-Policy vs. Random 2.727500\n", + "Random vs. QL-Policy 2.995250\n", + "Random vs. Random 3.309563\n", + "dtype: float64" + ] + }, + "execution_count": 322, + "metadata": {}, + "output_type": "execute_result" + } + ], + "source": [ + "pd.Series(\n", + " {\n", + " \"QL-Policy vs. Random\": np.mean(\n", + " np.sum(_l_actions_possible[::2], axis=0)[:, [0, -1]][:, :, [0, -1]]\n", + " ),\n", + " \"Random vs. QL-Policy\": np.mean(\n", + " np.sum(_r_actions_possible[1::2], axis=0)[:, [0, -1]][:, :, [0, -1]]\n", + " ),\n", + " \"Random vs. Random\": np.mean(\n", + " (np.sum(_poss_turns, axis=0) / 2)[:, [0, -1]][:, :, [0, -1]]\n", + " ),\n", + " }\n", + ")" + ] + }, + { + "cell_type": "markdown", + "metadata": { + "tags": [] + }, + "source": [ + "With the coner capture of this first policy I was quit happy even if it did not went as far as I expected." ] }, { @@ -6060,18 +6683,24 @@ "cell_type": "markdown", "metadata": {}, "source": [ - "The resulting R2 scores are very similar to the scores calculated before which I did not expect." + "The scores calculated above indicate that the policies exhibit almost symmetrical behavior across the two validated axes. Although the same has not been validated for the third axis, it is assumed to be similar. This demonstrates that the learning behavior is progressing in the right direction, even if the policy's score is lower than expected." ] }, { "cell_type": "markdown", "metadata": {}, "source": [ - "## Review and Outlook\n", + "## Concluseion\n", "\n", - "This project was much bigger then what I estimated. Especially the prework and simulation of the game was code whise much more complicated then expected. Working without proper testing and only with asserts did not make it easier. Jupyter notebooks are also normally used for less code than was executed here.\n", + "The implementation of Q-learning using artificial neural networks (ANNs) for the board game Othello proved to be a challenging task. Although I had planned to implement increasingly complex policies, the amount of work required for a reinforcement learning (RL) project, including Q-learning, was much larger than expected. Nonetheless, I am satisfied with the progress made, considering the limited amount of time available. Although the networks did not clearly converge, they showed promise, and I am uncertain how to define the best values for parameters such as $\\gamma$, $\\epsilon$, who_won_fraction, and final_score_fraction.\n", "\n", - "Initially I estimated to implement the policy with increasing complexities. I estimated the " + "Bug chasing was a significant challenge during the implementation, leading to multiple restarts of the training process and limiting the amount of computation. The use of memory replay and Q-learning, which takes the discounted maximum reward for the next state as a reward, would have been useful for overcoming this issue, but I did not explore this approach. The next step would have been to implement policy gradient and connect it to Q-learning to build an actor-critic, which would have improved the stability and reduced the need for validating as many actions as possible during training.\n", + "\n", + "This version of Q-learning using ANNs is more of a hashed version of Q-learning via deep Q-networks (DQN), which defines its loss as $L_{DQN} = r + \\gamma \\max_{a'} Q(s', a'; \\theta^-_i) - Q(s, a; \\theta_i)$. However, we worked with discounted values and not a discounted backpropagation through the ANN, allowing for exploration without skewing the rewards. Nonetheless, this approach needs to be balanced with other values that tend to skew the results towards the mean and may lead to mediocre results.\n", + "\n", + "In retrospect, I underestimated the effort required for this project. The pre-work and simulation of the game were more complex than expected, and working without proper testing and only using asserts added to the difficulties. Using Jupyter notebooks for such a large codebase was also not ideal. Although adding tests helped, the implementation of custom reward functions, with changes in perspective, axis, and other values, was still challenging to code and debug. Next time, I will start with testing at an early stage, even if it means creating test data, to speed up development.\n", + "\n", + "Although the ANN structure seems to work, it is limited by its environment. One possible change would be to eliminate the need for discounts, as Othello has a maximum of 60 turns. As such, it can be argued that no discount is necessary since it is possible to avoid the end of the game or an endless loop." ] }, {