Knowledge (XXG)

Hanabi (card game)

Source 📝

25: 331:. Computer programs developed for self-play fail badly when playing on ad hoc teams, since they don't know how to learn to adapt to the way other players play. Hu et al. demonstrated that learning symmetry-invariant strategies helps AI agents avoid learning uninterpretable conventions, improving their performance when matched with separately trained AI agents (scoring around 22), and with humans (scoring around 16 vs. a baseline self-play model that scored around 9). 251:: The player chooses a card from their hand and attempts to add it to the cards already played. This is successful if the card is a 1 in a suit that has not yet been played, or if it is the next number sequentially in a suit that has been played. Otherwise a fuse token is consumed and the misplayed card is discarded. Successfully playing a 5 of any suit replenishes one information token. Whether the play was successful or not, the player draws a replacement card. 239:: The player points out the cards of either a given number or a given suit in the hand of another player (examples: "This card is your only red card," "These two cards are your only 3s"). The information given must be complete and correct. (In some editions, it is allowed to indicate that a player has zero of something; other versions explicitly forbid this case.) Giving information consumes one information token. 84: 255:
The game ends immediately when either all fuse tokens are used up, resulting in a game loss, or all 5s have been played successfully, leading to a game win. Otherwise, play continues until the deck runs out, and for one full round after that. At the end of the game, the values of the highest cards in
292:
or "ad hoc team play". In self-play, multiple instances of the program play with each other on a team. They thus share a carefully honed strategy for communication and play, though of course they are not allowed to illegally share any information about each game with other instances of the program.
326:
Ad hoc team play is a far greater challenge for AI, because "Hanabi elevates reasoning about the beliefs and intentions of other agents to the foreground". Playing at human levels with ad hoc teams requires the algorithms to learn and develop communication conventions and strategies over time with
322:
In self-play mode, the challenge is to develop a program which can learn from scratch to play well with other instances of itself. Such programs achieve only about 15 points per game as of 2019, far worse than hand-coded programs. However, this gap has narrowed significantly as of 2020, with the
227:
deck contains cards in five suits (white, yellow, green, blue, and red): three 1s, two each of 2s, 3s, and 4s, and one 5. The game begins with 8 available information tokens and 3 fuse tokens. To start the game, players are dealt a hand containing five cards (four for 4 or 5 players). As in
264:
Hanabi received positive reviews. Board Game Quest awarded the game four and a half stars, praising its uniqueness, accessibility and engagement. Similarly, The Opinionated Gamers also praised the game's engagement and addictiveness. It won several awards, including the 2013
245:: The player chooses a card from their hand and adds it to the discard pile, then draws a card to replace it. The discarded card is out of the game and can no longer be played. Discarding a card replenishes one information token. 562:
Bowling, Michael; Bellemare, Marc G.; Larochelle, Hugo; Mourad, Shibl; Dunning, Iain; Hughes, Edward; Moitra, Subhodeep; Dumoulin, Vincent; Parisotto, Emilio (2019-02-01). "The Hanabi Challenge: A New Frontier for AI Research".
512:
Cox, Christopher; De Silva, Jessica; Deorsey, Philip; Kenter, Franklin H. J.; Retter, Troy; Tobin, Josh (December 2014). "How to Make the Perfect Fireworks Display: Two Strategies for Hanabi".
821: 203:
and published in 2010. Players are aware of other players' cards but not their own, and attempt to play a series of cards in a specific order to set off a simulated
207:
show. The types of information that players may give to each other is limited, as is the total amount of information that can be given during the game. In 2013,
232:, players can see each other's cards but they cannot see their own. Play proceeds around the table; each turn, a player must take one of the following actions: 289: 752: 303:
strategies. The best programs, such as WTFWThat, achieved near-perfect results in self-play with five players, with an average score of 24.9 out of 25.
481: 1158: 1173: 609: 1163: 59: 44:
Please help improve this article by looking for better, more reliable sources. Unreliable citations may be challenged and removed.
1148: 917: 745: 38: 1168: 989: 229: 1153: 973: 452: 33: 805: 738: 656:
hanabi_learning_environment is a research platform for Hanabi experiments.: deepmind/hanabi-learning-environment
334:
Deepmind released an open source code framework to facilitate research, called the Hanabi Learning Environment.
949: 467: 845: 797: 316: 1109: 1069: 893: 348: 270: 193: 24: 1117: 1101: 1093: 965: 713:"Solving Hanabi: Estimating Hands by Opponent's Actions in Cooperative Game with Incomplete Information" 358: 353: 869: 853: 692: 584: 1013: 981: 491: 428: 941: 282: 160: 1021: 933: 789: 564: 537: 712: 1085: 773: 529: 300: 186: 269:
winner and 2013 Fairplay À la carte Award winner. Hanabi also placed sixth place in the 2013
829: 761: 521: 266: 212: 1037: 877: 654: 403: 343: 885: 813: 328: 296:
In ad hoc team play, the program plays with other arbitrary programs or human players.
1142: 957: 685: 541: 200: 102: 97: 637: 1125: 1061: 378: 861: 696: 679: 189: 168: 719:. AAAI Workshops at the Twenty-Ninth AAAI Conference on Artificial Intelligence 585:"The next big challenge for Google's A.I. is a card game you've never heard of" 1077: 997: 525: 256:
each suit are summed, resulting in a total score out of a possible 25 points.
533: 315:
proposed Hanabi as an ideal game with which to establish a new benchmark for
925: 909: 837: 204: 196: 693:"Hanabi: card game with the goal to launch a spectacular firework display" 83: 1053: 781: 312: 172: 1029: 1005: 610:"A cooperative benchmark: Announcing the Hanabi Learning Environment" 569: 164: 730: 901: 486: 299:
A variety of computer programs have been developed by hand-coding
636:
Hu, Hengyuan; Lerer, Adam; Peysakhovich, Alex; Foerster, Jakob.
734: 18: 288:
Computer programs which play Hanabi can either engage in
323:
Simplified Action Decoder achieving scores around 24.
643:. International Conference on Machine Learning, 2020. 215:, an industry award for best board game of the year. 156: 148: 140: 132: 124: 116: 108: 96: 746: 468:"Spiel des Jahres official site: 2013 winner" 8: 631: 629: 112:R&R Games, Cocktail Games, Abacus Spiele 74: 753: 739: 731: 638:""Other-Play" for Zero-Shot Coordination" 568: 60:Learn how and when to remove this message 453:"Fairplay Online: À la carte prize 2013" 370: 73: 822:Sherlock Holmes: Consulting Detective 379:"Hanabi | Board Game | BoardGameGeek" 7: 557: 555: 553: 551: 691:Seagull, Jon (29 September 2014). 14: 402:Mastrangeli, Tony (2014-02-25). 281:Hanabi is a cooperative game of 199:created by French game designer 82: 23: 711:Hirotaka Osawa (1 April 2015). 319:research in cooperative play. 1: 1159:Card games introduced in 2010 1174:Game artificial intelligence 429:"SdJ Re-Reviews #35: Hanabi" 482:"PreistrĂ€ger – SPIEL Messe" 1190: 427:Wray, Chris (2015-12-29). 1164:Dedicated deck card games 768: 526:10.4169/math.mag.88.5.323 81: 1149:Cooperative board games 614:www.marcgbellemare.info 317:Artificial intelligence 32:Some of this article's 1110:MicroMacro: Crime City 659:, DeepMind, 2019-07-01 433:The Opinionated Gamers 349:7 Wonders (board game) 271:Deutscher Spiele Preis 902:The Settlers of Catan 589:www.digitaltrends.com 359:Terror in Meeple City 354:Takenoko (board game) 283:imperfect information 514:Mathematics Magazine 327:other players via a 1169:Antoine Bauza games 78: 1154:French board games 870:Drunter und DrĂŒber 854:CafĂ© International 192:, fireworks) is a 1136: 1135: 918:Mississippi Queen 774:Hare and Tortoise 591:. 9 February 2019 383:boardgamegeek.com 230:blind man's bluff 178: 177: 88:The box cover of 70: 69: 62: 1181: 830:Top Secret Spies 798:Enchanted Forest 762:Spiel des Jahres 755: 748: 741: 732: 727: 725: 724: 707: 705: 703: 667: 666: 665: 664: 651: 645: 644: 642: 633: 624: 623: 621: 620: 606: 600: 599: 597: 596: 581: 575: 574: 572: 559: 546: 545: 509: 503: 502: 500: 499: 490:. Archived from 478: 472: 471: 464: 458: 456: 449: 443: 442: 440: 439: 424: 418: 417: 415: 414: 408:Board Game Quest 399: 393: 392: 390: 389: 375: 267:Spiel des Jahres 237:Give information 213:Spiel des Jahres 86: 79: 65: 58: 54: 51: 45: 27: 19: 1189: 1188: 1184: 1183: 1182: 1180: 1179: 1178: 1139: 1138: 1137: 1132: 1038:Kingdom Builder 990:Thurn and Taxis 878:Um Reifenbreite 764: 759: 722: 720: 710: 701: 699: 690: 676: 671: 670: 662: 660: 653: 652: 648: 640: 635: 634: 627: 618: 616: 608: 607: 603: 594: 592: 583: 582: 578: 561: 560: 549: 511: 510: 506: 497: 495: 480: 479: 475: 466: 465: 461: 451: 450: 446: 437: 435: 426: 425: 421: 412: 410: 404:"Hanabi Review" 401: 400: 396: 387: 385: 377: 376: 372: 367: 344:Computer bridge 340: 309: 279: 277:Computer Hanabi 262: 221: 92: 66: 55: 49: 46: 43: 28: 17: 12: 11: 5: 1187: 1185: 1177: 1176: 1171: 1166: 1161: 1156: 1151: 1141: 1140: 1134: 1133: 1131: 1130: 1122: 1114: 1106: 1098: 1090: 1082: 1074: 1066: 1058: 1050: 1042: 1034: 1026: 1018: 1010: 1002: 994: 986: 978: 974:Ticket to Ride 970: 962: 954: 946: 938: 930: 922: 914: 906: 898: 890: 882: 874: 866: 858: 850: 842: 834: 826: 818: 814:Railway Rivals 810: 802: 794: 786: 778: 769: 766: 765: 760: 758: 757: 750: 743: 735: 729: 728: 708: 688: 675: 674:External links 672: 669: 668: 646: 625: 601: 576: 547: 520:(5): 323–336. 504: 473: 459: 444: 419: 394: 369: 368: 366: 363: 362: 361: 356: 351: 346: 339: 336: 329:theory of mind 308: 305: 278: 275: 261: 258: 253: 252: 246: 243:Discard a card 240: 220: 217: 176: 175: 158: 154: 153: 150: 146: 145: 142: 138: 137: 134: 130: 129: 126: 122: 121: 118: 114: 113: 110: 106: 105: 100: 94: 93: 87: 68: 67: 34:listed sources 31: 29: 22: 15: 13: 10: 9: 6: 4: 3: 2: 1186: 1175: 1172: 1170: 1167: 1165: 1162: 1160: 1157: 1155: 1152: 1150: 1147: 1146: 1144: 1128: 1127: 1123: 1120: 1119: 1115: 1112: 1111: 1107: 1104: 1103: 1099: 1096: 1095: 1091: 1088: 1087: 1083: 1080: 1079: 1075: 1072: 1071: 1067: 1064: 1063: 1059: 1056: 1055: 1051: 1048: 1047: 1043: 1040: 1039: 1035: 1032: 1031: 1027: 1024: 1023: 1019: 1016: 1015: 1011: 1008: 1007: 1003: 1000: 999: 995: 992: 991: 987: 984: 983: 979: 976: 975: 971: 968: 967: 963: 960: 959: 958:Villa Paletti 955: 952: 951: 947: 944: 943: 939: 936: 935: 931: 928: 927: 923: 920: 919: 915: 912: 911: 907: 904: 903: 899: 896: 895: 891: 888: 887: 883: 880: 879: 875: 872: 871: 867: 864: 863: 859: 856: 855: 851: 848: 847: 843: 840: 839: 835: 832: 831: 827: 824: 823: 819: 816: 815: 811: 808: 807: 806:Scotland Yard 803: 800: 799: 795: 792: 791: 787: 784: 783: 779: 776: 775: 771: 770: 767: 763: 756: 751: 749: 744: 742: 737: 736: 733: 718: 714: 709: 698: 694: 689: 687: 686:BoardGameGeek 683: 682: 678: 677: 673: 658: 657: 650: 647: 639: 632: 630: 626: 615: 611: 605: 602: 590: 586: 580: 577: 571: 566: 558: 556: 554: 552: 548: 543: 539: 535: 531: 527: 523: 519: 515: 508: 505: 494:on 2020-11-03 493: 489: 488: 483: 477: 474: 469: 463: 460: 454: 448: 445: 434: 430: 423: 420: 409: 405: 398: 395: 384: 380: 374: 371: 364: 360: 357: 355: 352: 350: 347: 345: 342: 341: 337: 335: 332: 330: 324: 320: 318: 314: 306: 304: 302: 297: 294: 291: 286: 284: 276: 274: 272: 268: 259: 257: 250: 247: 244: 241: 238: 235: 234: 233: 231: 226: 218: 216: 214: 210: 206: 202: 201:Antoine Bauza 198: 195: 191: 188: 184: 183: 174: 170: 166: 162: 159: 155: 151: 147: 143: 139: 136:20–30 minutes 135: 131: 127: 123: 119: 115: 111: 107: 104: 103:Antoine Bauza 101: 99: 95: 91: 85: 80: 77: 72: 64: 61: 53: 41: 40: 35: 30: 26: 21: 20: 1126:Dorfromantik 1124: 1116: 1108: 1100: 1092: 1084: 1076: 1068: 1062:Colt Express 1060: 1052: 1045: 1044: 1036: 1028: 1020: 1012: 1004: 996: 988: 980: 972: 964: 956: 948: 940: 932: 924: 916: 908: 900: 892: 884: 876: 868: 860: 852: 844: 836: 828: 820: 812: 804: 796: 788: 780: 772: 721:. Retrieved 717:www.aaai.org 716: 700:. Retrieved 680: 661:, retrieved 655: 649: 617:. Retrieved 613: 604: 593:. Retrieved 588: 579: 570:1902.00506v1 517: 513: 507: 496:. Retrieved 492:the original 485: 476: 462: 447: 436:. Retrieved 432: 422: 411:. Retrieved 407: 397: 386:. Retrieved 382: 373: 333: 325: 321: 310: 307:AI challenge 298: 295: 287: 280: 263: 254: 248: 242: 236: 224: 222: 208: 181: 180: 179: 133:Playing time 89: 75: 71: 56: 47: 36: 950:Carcassonne 886:Liar's dice 862:Hoity Toity 697:Boing Boing 457:(in German) 249:Play a card 194:cooperative 169:Cooperation 37:may not be 1143:Categories 1078:Kingdomino 998:Zooloretto 846:Barbarossa 723:2015-06-07 663:2019-07-04 619:2019-07-04 595:2019-07-04 498:2022-12-22 438:2022-02-25 413:2022-02-24 388:2016-01-24 365:References 301:rule-based 125:Setup time 109:Publishers 1070:Codenames 926:Elfenland 910:El Grande 894:Manhattan 838:Auf Achse 542:124445429 534:0025-570X 311:In 2019, 290:self-play 260:Reception 205:fireworks 197:card game 161:Deduction 149:Age range 128:5 minutes 98:Designers 50:July 2022 16:Card game 1118:Cascadia 1102:Pictures 1094:Just One 1054:Camel Up 1014:Dominion 966:Alhambra 782:Rummikub 338:See also 313:DeepMind 219:Gameplay 211:won the 187:Japanese 173:Planning 152:8 and up 39:reliable 1030:Qwirkle 982:Niagara 117:Players 1129:(2023) 1121:(2022) 1113:(2021) 1105:(2020) 1097:(2019) 1089:(2018) 1081:(2017) 1073:(2016) 1065:(2015) 1057:(2014) 1049:(2013) 1046:Hanabi 1041:(2012) 1033:(2011) 1025:(2010) 1017:(2009) 1009:(2008) 1006:Keltis 1001:(2007) 993:(2006) 985:(2005) 977:(2004) 969:(2003) 961:(2002) 953:(2001) 945:(2000) 942:Torres 937:(1999) 929:(1998) 921:(1997) 913:(1996) 905:(1995) 897:(1994) 889:(1993) 881:(1992) 873:(1991) 865:(1990) 857:(1989) 849:(1988) 841:(1987) 833:(1986) 825:(1985) 817:(1984) 809:(1983) 801:(1982) 793:(1981) 785:(1980) 777:(1979) 702:7 June 681:Hanabi 540:  532:  225:Hanabi 209:Hanabi 185:(from 182:Hanabi 165:Memory 157:Skills 144:Medium 141:Chance 120:2 to 5 90:Hanabi 76:Hanabi 1022:Dixit 934:Tikal 790:Focus 641:(PDF) 565:arXiv 538:S2CID 487:Spiel 1086:Azul 704:2015 530:ISSN 223:The 684:at 522:doi 1145:: 715:. 695:. 628:^ 612:. 587:. 550:^ 536:. 528:. 518:88 516:. 484:. 431:. 406:. 381:. 285:. 273:. 190:花火 171:, 167:, 163:, 754:e 747:t 740:v 726:. 706:. 622:. 598:. 573:. 567:: 544:. 524:: 501:. 470:. 455:. 441:. 416:. 391:. 63:) 57:( 52:) 48:( 42:.

Index


listed sources
reliable
Learn how and when to remove this message

Designers
Antoine Bauza
Deduction
Memory
Cooperation
Planning
Japanese
花火
cooperative
card game
Antoine Bauza
fireworks
Spiel des Jahres
blind man's bluff
Spiel des Jahres
Deutscher Spiele Preis
imperfect information
self-play
rule-based
DeepMind
Artificial intelligence
theory of mind
Computer bridge
7 Wonders (board game)
Takenoko (board game)

Text is available under the Creative Commons Attribution-ShareAlike License. Additional terms may apply.

↑