List your company

Reinforcement Learning on Atari 2600 (test) (Full Game Set)

4,704Alien ScoreDelta Net
686.56321,729.55162,772.543,815.5284Jul 2, 2018Dec 28, 2018Jun 25, 2019Dec 21, 2019Jun 17, 2020Dec 13, 2020Jun 11, 2021

Evaluation results

MethodLinks
Delta NetEnvironmental steps=50 M, Number of parameters=1.5 M2021.064,7043395,710———3216,780———————31,359—————8,707——3,338——8,095—2,116——21—6,2489,185—7.28,85325,76911,599———120,806———4,265331
Delta RNNEnvironmental steps=50 M, Number of parameters=1.6 M2021.063,4202167,503———3247,040———————36,807—————25,445——2,929.3——7,388—2,287——21—6,62613,429—75,1232,8479,523———148,759———3,903305
LTEnvironmental steps=50 M, Number of parameters=1.5 M2021.062,1353202,764———309933———————8,939—————8,197——2,425——3,722—1,647——21—4,6936,695—4.81,8122,3458,915———57,435———2,719323
FeedforwardEnvironmental steps=50 M, Number of parameters=1.1 M2021.061,9852084,658———32312,267———————36,255—————10,356——2,942——5,449—1,737——21—4,9679,932—7.146948,1509,397———185,632———4,863326
RDNEnvironmental steps=50 M, Number of parameters=1.5 M2021.061,7543688,088———3167,373———————41,726—————19,775——2,979——9,201—2,584——21—5,89718,156—7.568627,2139,737———126,140———5,285336
POP3DFrames=40M, Trials=32018.071,147.29299.552,139.152,004.431,652.48488,134.03662.2611,131.441,965.2737.9783.12143.63,056.813,269.4797,257.527,611.27—107.84-2117.76276.471,556.29413.2-4.67358.541,614.636,538.1623,253.960.141,214.095,353.14-2.4113.2487.375,852.15,260.8925,456.313.081,487.84693.2614,734.11-19.863,396.61179.9638,728.4815.8927,346.442,340.63,739.56—
PPOFrames=40M, Trials=32018.071,115.94413.462,168.932,102.11,470.46596,807.27643.949,387.771,460.5939.4178.61124.983,344.633,106.1490,169.67,180.43—321.2-27.5115.87267.731,196.2509.81-5.5394.452,199.747,195.2423,283.310.741,482.775,199.37-5.8112.8352.766,744.135,487.1724,688.078.651,120.15632.1713,643.8-21.84,410.87152.7243,208.9914.6627,549.552,743.41,813.9—
BASELINEFrames=40M, Trials=32018.07851.13295.911,159.51,884.681,477.71192,798.74859.2511,674.33,321.2533.927.9229.993,042.481,780.3869,258.319,814.42—92.59-81.915.93270.42900.74342.74-4.61380.91937.984,760.6619,637.580.22860.634,562.32-31.277.256.71,760.925,220.6420,385.9113.891,112.51552.516,288.35-21.844,718.46103.955,430.220.5723,998.092,409.94256.78—
LSTMEnvironmental steps=50 M, Number of parameters=1.6 M2021.068462337,551———3276,327———————31,447—————13,765——3,020——15,216—1,676——21—3,90513,638—7.670812,4618,337———155,847———2,737474
TRPOFrames=40M, Trials=32018.07841.08169.12971.781,342.831,760.73174,394.94831.9512,918.392,431.6330.9923.0726.563,142.221,595.8263,189.786,204.68—140.67-81.9717.33270.57875.93317.86-5.21519.01566.853,861.8718,293.120.12864.844,504.67-33.93-2.9198.791,679.034,549.2216,269.414.57848.47483.4813,341.23-21.044,544.68109.187,085.020.0323,705.392,045.171,521.28—