966:
meaningful training signals. SSL tasks are designed so that solving it requires capturing essential features or relationships in the data. The input data is typically augmented or transformed in a way that creates pairs of related samples. One sample serves as the input, and the other is used to formulate the supervisory signal. This augmentation can involve introducing noise, cropping, rotation, or other transformations. Self-supervised learning more closely imitates the way humans learn to classify objects.
977:. The model learns in two steps. First, the task is solved based on an auxiliary or pretext classification task using pseudo-labels which help to initialize the model parameters. Second, the actual task is performed with supervised or unsupervised learning. Other auxiliary tasks involve pattern completion from masked input patterns (silent pauses in speech or image portions masked in black).
1566:
1042:
Non-contrastive self-supervised learning (NCSSL) uses only positive examples. Counterintuitively, NCSSL converges on a useful local minimum rather than reaching a trivial solution, with zero loss. For the example of binary classification, it would trivially learn to classify each example as positive.
1016:
The training process involves presenting the model with input data and requiring it to reconstruct the same data as closely as possible. The loss function used during training typically penalizes the difference between the original input and the reconstructed output. By minimizing this reconstruction
1029:
can be divided into positive examples and negative examples. Positive examples are those that match the target. For example, if you're learning to identify birds, the positive training data are those pictures that contain birds. Negative examples are those that do not. Contrastive self-supervised
1083:
intrinsically constitutes a self-supervised process, because the output pattern needs to become an optimal reconstruction of the input pattern itself. However, in current jargon, the term 'self-supervised' has become associated with classification tasks that are based on a pretext-task training
1790:
Grill, Jean-Bastien; Strub, Florian; Altché, Florent; Tallec, Corentin; Richemond, Pierre H.; Buchatskaya, Elena; Doersch, Carl; Pires, Bernardo Avila; Guo, Zhaohan Daniel; Azar, Mohammad
Gheshlaghi; Piot, Bilal (10 September 2020). "Bootstrap your own latent: A new approach to self-supervised
965:
where a model is trained on a task using the data itself to generate supervisory signals, rather than relying on external labels provided by humans. In the context of neural networks, self-supervised learning aims to leverage inherent structures or relationships within the input data to create
1005:
Autoassociative self-supervised learning is a specific category of self-supervised learning where a neural network is trained to reproduce or reconstruct its own input data. In other words, the model is tasked with learning a representation of the data that captures its essential features or
1875:
Balestriero, Randall; Ibrahim, Mark; Sobal, Vlad; Morcos, Ari; Shekhar, Shashank; Goldstein, Tom; Bordes, Florian; Bardes, Adrien; Mialon, Gregoire; Tian, Yuandong; Schwarzschild, Avi; Wilson, Andrew Gordon; Geiping, Jonas; Garrido, Quentin; Fernandez, Pierre (24 April 2023). "A Cookbook of
1013:, which are a type of neural network architecture used for representation learning. Autoencoders consist of an encoder network that maps the input data to a lower-dimensional representation (latent space), and a decoder network that reconstructs the input data from this representation.
1051:
SSL belongs to supervised learning methods insofar as the goal is to generate a classified output from the input. At the same time, however, it does not require the explicit use of labeled input-output pairs. Instead, correlations, metadata embedded in the data, or
1059:
SSL is similar to unsupervised learning in that it does not require labels in the sample data. Unlike unsupervised learning, however, learning is not done using inherent data structures.
1091:, self-supervising learning from a combination of losses can create abstract representations where only the most important information about the state are kept in a compressed way.
1195:
868:
906:
1056:
present in the input are implicitly and autonomously extracted from the data. These supervisory signals, generated from the data, can then be used for training.
863:
1812:
Gündüz, Hüseyin Anil; Binder, Martin; To, Xiao-Yin; Mreches, René; Bischl, Bernd; McHardy, Alice C.; Münch, Philipp C.; Rezaei, Mina (11 September 2023).
853:
694:
1009:
The term "autoassociative" comes from the fact that the model is essentially associating the input data with itself. This is often achieved using
1694:
Francois-Lavet, Vincent; Bengio, Yoshua; Precup, Doina; Pineau, Joelle (2019). "Combined
Reinforcement Learning via Abstract Representations".
1614:
901:
858:
709:
2089:
440:
941:
744:
1537:
820:
1084:
setup. This involves the (human) design of such pretext task(s), unlike the case of fully self-contained autoencoder training.
369:
1978:
1929:
1670:
1510:
1400:
1345:
1287:
878:
641:
176:
896:
729:
704:
653:
777:
772:
425:
1135:
that can be used in language processing. It can be used to translate texts or answer questions, among other things.
435:
73:
1034:
minimizes the distance between positive sample pairs while maximizing the distance between negative sample pairs.
1043:
Effective NCSSL requires an extra predictor on the online side that does not back-propagate on the target side.
980:
Self-supervised learning has produced promising results in recent years and has found practical application in
934:
830:
594:
415:
1108:
805:
507:
283:
1719:
762:
699:
609:
587:
430:
420:
1747:"Structural Supervision Improves Few-Shot Learning and Syntactic Generalization in Neural Language Models"
1153:
913:
825:
810:
271:
93:
1479:
Gidaris, Spyros; Bursuc, Andrei; Komodakis, Nikos; Perez, Patrick Perez; Cord, Matthieu (October 2019).
1745:
Wilcox, Ethan; Qian, Peng; Futrell, Richard; Kohita, Ryosuke; Levy, Roger; Ballesteros, Miguel (2020).
873:
800:
550:
445:
233:
166:
126:
1065:
combines supervised and unsupervised learning, requiring only a small portion of the learning data be
2084:
1157:
970:
927:
533:
301:
171:
1062:
555:
475:
398:
316:
146:
108:
103:
58:
981:
502:
351:
251:
78:
1088:
682:
658:
560:
321:
296:
256:
68:
1751:
Proceedings of the 2020 Conference on
Empirical Methods in Natural Language Processing (EMNLP)
1118:
636:
458:
410:
266:
181:
53:
1581:
1017:
error, the autoencoder learns a meaningful representation of the data in its latent space.
565:
515:
1167:
DirectPred is a NCSSL that directly sets the predictor weights instead of learning it via
8:
668:
604:
575:
480:
306:
239:
225:
211:
186:
136:
88:
48:
1848:
1813:
1585:
2051:
Proceedings of the 33rd Annual
Meeting of the Association for Computational Linguistics
2033:
1984:
1956:
1935:
1907:
1877:
1792:
1772:
1754:
1753:. Stroudsburg, PA, USA: Association for Computational Linguistics. pp. 4640–4652.
1699:
1676:
1648:
1516:
1488:
1461:
1406:
1378:
1351:
1323:
1293:
1265:
1226:
Proceedings of the 33rd Annual
Meeting of the Association for Computational Linguistics
1149:
1099:
Self-supervised learning is particularly suitable for speech recognition. For example,
989:
646:
570:
356:
151:
1768:
1196:"What is Self-Supervised Learning? | Will machines ever be able to learn like humans?"
2025:
2017:
1998:"Fast and robust segmentation of white blood cell images by self-supervised learning"
1974:
1925:
1853:
1835:
1776:
1746:
1666:
1520:
1506:
1453:
1445:
1426:"Fast and robust segmentation of white blood cell images by self-supervised learning"
1396:
1355:
1341:
1283:
1073:
739:
582:
495:
291:
261:
206:
201:
156:
98:
2054:
2037:
2009:
1997:
1988:
1966:
1917:
1843:
1825:
1764:
1720:"Open Sourcing BERT: State-of-the-Art Pre-training for Natural Language Processing"
1680:
1658:
1589:
1498:
1465:
1437:
1410:
1388:
1333:
1314:
Beyer, Lucas; Zhai, Xiaohua; Oliver, Avital; Kolesnikov, Alexander (October 2019).
1275:
1229:
1168:
1053:
962:
767:
520:
470:
380:
364:
334:
196:
191:
141:
131:
29:
1939:
1425:
1297:
2013:
1615:"Demystifying a key self-supervised learning technique: Non-contrastive learning"
1441:
795:
599:
465:
405:
1814:"A self-supervised deep learning method for data-efficient training in genomics"
1948:
1899:
1830:
1132:
815:
346:
83:
1567:"Nonlinear principal component analysis using autoassociative neural networks"
2078:
2021:
1839:
1480:
1449:
1315:
1031:
1026:
974:
734:
663:
545:
276:
161:
1502:
1337:
1107:, a self-supervised algorithm, to perform speech recognition using two deep
2029:
1857:
1640:
1457:
1370:
1257:
1066:
1010:
2059:
1970:
1921:
1662:
1593:
1392:
1279:
1234:
1030:
learning uses both positive and negative examples. Contrastive learning's
1121:(BERT) model is used to better understand the context of search queries.
1080:
540:
34:
1538:"Wav2vec: State-of-the-art speech recognition through self-supervision"
689:
385:
311:
1645:
2016 IEEE Conference on
Computer Vision and Pattern Recognition (CVPR)
1174:
Self-GenomeNet is an example of self-supervised learning in genomics.
1156:. From a small number of labeled examples, it learns to predict which
2053:. Cambridge, MA: Association for Computational Linguistics: 189–196.
2046:
1228:. Cambridge, MA: Association for Computational Linguistics: 189–196.
1221:
848:
629:
2047:"Unsupervised Word Sense Disambiguation Rivaling Supervised Methods"
1222:"Unsupervised Word Sense Disambiguation Rivaling Supervised Methods"
1996:
Zheng, Xin; Wang, Yong; Wang, Guoyou; Liu, Jianguo (1 April 2018).
1961:
1949:"Unsupervised Visual Representation Learning by Context Prediction"
1912:
1882:
1797:
1759:
1704:
1653:
1493:
1383:
1371:"Unsupervised Visual Representation Learning by Context Prediction"
1328:
1270:
1161:
1142:
1100:
985:
1046:
1947:
Doersch, Carl; Gupta, Abhinav; Efros, Alexei A. (December 2015).
1424:
Zheng, Xin; Wang, Yong; Wang, Guoyou; Liu, Jianguo (April 2018).
1369:
Doersch, Carl; Gupta, Abhinav; Efros, Alexei A. (December 2015).
624:
1693:
1485:
2019 IEEE/CVF International
Conference on Computer Vision (ICCV)
1320:
2019 IEEE/CVF International
Conference on Computer Vision (ICCV)
1124:
1114:
375:
1874:
1696:
Proceedings of the AAAI Conference on
Artificial Intelligence
1128:
1076:
a model designed for one task is reused on a different task.
619:
614:
341:
1953:
2015 IEEE International
Conference on Computer Vision (ICCV)
1904:
2017 IEEE International Conference on Computer Vision (ICCV)
1375:
2015 IEEE International Conference on Computer Vision (ICCV)
1262:
2017 IEEE International Conference on Computer Vision (ICCV)
1478:
1037:
1000:
1481:"Boosting Few-Shot Visual Learning with Self-Supervision"
1313:
1006:
structure, allowing it to regenerate the original input.
907:
List of datasets in computer vision and image processing
1744:
1119:
Bidirectional Encoder Representations from Transformers
1020:
1789:
1811:
1141:(BYOL) is a NCSSL that produced excellent results on
1641:"The Multiverse Loss for Robust Transfer Learning"
1946:
1898:Doersch, Carl; Zisserman, Andrew (October 2017).
1368:
1256:Doersch, Carl; Zisserman, Andrew (October 2017).
2076:
1995:
1897:
1423:
1255:
1145:and on transfer and semi-supervised benchmarks.
1316:"S4L: Self-Supervised Semi-Supervised Learning"
1047:Comparison with other forms of machine learning
902:List of datasets for machine-learning research
1164:word is being used at a given point in text.
1152:is an example of self-supervised learning in
935:
1900:"Multi-task Self-Supervised Visual Learning"
1258:"Multi-task Self-Supervised Visual Learning"
1638:
942:
928:
2058:
1960:
1911:
1881:
1847:
1829:
1796:
1758:
1703:
1652:
1492:
1382:
1327:
1269:
1233:
2044:
1219:
1193:
1038:Non-contrastive self-supervised learning
1001:Autoassociative self-supervised learning
1639:Littwin, Etai; Wolf, Lior (June 2016).
1309:
1307:
1189:
1187:
2077:
1564:
1532:
1530:
969:The typical SSL method is based on an
1609:
1607:
1605:
1603:
1560:
1558:
1304:
1194:Bouchard, Louis (25 November 2020).
1184:
1021:Contrastive self-supervised learning
1527:
897:Glossary of artificial intelligence
13:
2090:Generative artificial intelligence
1868:
1600:
1555:
1025:For a binary classification task,
14:
2101:
1891:
1805:
1783:
1769:10.18653/v1/2020.emnlp-main.375
1738:
1712:
1687:
1632:
1472:
1417:
1362:
1249:
1213:
317:Relevance vector machine (RVM)
16:A paradigm in machine learning
1:
1177:
1109:convolutional neural networks
806:Computational learning theory
370:Expectation–maximization (EM)
2014:10.1016/j.micron.2018.01.010
1647:. IEEE. pp. 3957–3966.
1487:. IEEE. pp. 8058–8067.
1442:10.1016/j.micron.2018.01.010
1377:. IEEE. pp. 1422–1430.
1322:. IEEE. pp. 1476–1485.
1264:. IEEE. pp. 2070–2079.
763:Coefficient of determination
610:Convolutional neural network
322:Support vector machine (SVM)
7:
1876:Self-Supervised Learning".
1154:natural language processing
1094:
914:Outline of machine learning
811:Empirical risk minimization
10:
2106:
1831:10.1038/s42003-023-05310-2
1111:that build on each other.
551:Feedforward neural network
302:Artificial neural networks
1139:Bootstrap Your Own Latent
973:or other model such as a
971:artificial neural network
534:Artificial neural network
2045:Yarowsky, David (1995).
1565:Kramer, Mark A. (1991).
1220:Yarowsky, David (1995).
1063:Semi-supervised learning
995:
955:Self-supervised learning
843:Journals and conferences
790:Mathematical foundations
700:Temporal difference (TD)
556:Recurrent neural network
476:Conditional random field
399:Dimensionality reduction
147:Dimensionality reduction
109:Quantum machine learning
104:Neuromorphic engineering
64:Self-supervised learning
59:Semi-supervised learning
1503:10.1109/iccv.2019.00815
1338:10.1109/iccv.2019.00156
252:Apprenticeship learning
1955:. pp. 1422–1430.
1906:. pp. 2070–2079.
1818:Communications Biology
1089:reinforcement learning
801:Bias–variance tradeoff
683:Reinforcement learning
659:Spiking neural network
69:Reinforcement learning
2060:10.3115/981658.981684
1971:10.1109/ICCV.2015.167
1922:10.1109/ICCV.2017.226
1663:10.1109/cvpr.2016.429
1594:10.1002/aic.690370209
1393:10.1109/iccv.2015.167
1280:10.1109/iccv.2017.226
1235:10.3115/981658.981684
1131:is an autoregressive
984:and is being used by
637:Neural radiance field
459:Structured prediction
182:Structured prediction
54:Unsupervised learning
826:Statistical learning
724:Learning with humans
516:Local outlier factor
1586:1991AIChE..37..233K
961:) is a paradigm in
669:Electrochemical RAM
576:reservoir computing
307:Logistic regression
226:Supervised learning
212:Multimodal learning
187:Feature engineering
132:Generative modeling
94:Rule-based learning
89:Curriculum learning
49:Supervised learning
24:Part of a series on
1150:Yarowsky algorithm
990:speech recognition
237: •
152:Density estimation
1980:978-1-4673-8391-2
1931:978-1-5386-1032-9
1726:. 2 November 2018
1672:978-1-4673-8851-1
1512:978-1-7281-4803-8
1402:978-1-4673-8391-2
1347:978-1-7281-4803-8
1289:978-1-5386-1032-9
1074:transfer learning
952:
951:
757:Model diagnostics
740:Human-in-the-loop
583:Boltzmann machine
496:Anomaly detection
292:Linear regression
207:Ontology learning
202:Grammar induction
177:Semantic analysis
172:Association rules
157:Anomaly detection
99:Neuro-symbolic AI
2097:
2085:Machine learning
2071:
2069:
2067:
2062:
2041:
1992:
1964:
1943:
1915:
1887:
1885:
1862:
1861:
1851:
1833:
1809:
1803:
1802:
1800:
1787:
1781:
1780:
1762:
1742:
1736:
1735:
1733:
1731:
1716:
1710:
1709:
1707:
1691:
1685:
1684:
1656:
1636:
1630:
1629:
1627:
1625:
1611:
1598:
1597:
1571:
1562:
1553:
1552:
1550:
1548:
1534:
1525:
1524:
1496:
1476:
1470:
1469:
1421:
1415:
1414:
1386:
1366:
1360:
1359:
1331:
1311:
1302:
1301:
1273:
1253:
1247:
1246:
1244:
1242:
1237:
1217:
1211:
1210:
1208:
1206:
1191:
1054:domain knowledge
982:audio processing
963:machine learning
944:
937:
930:
891:Related articles
768:Confusion matrix
521:Isolation forest
466:Graphical models
245:
244:
197:Learning to rank
192:Feature learning
30:Machine learning
21:
20:
2105:
2104:
2100:
2099:
2098:
2096:
2095:
2094:
2075:
2074:
2065:
2063:
1981:
1932:
1894:
1871:
1869:Further reading
1866:
1865:
1810:
1806:
1788:
1784:
1743:
1739:
1729:
1727:
1718:
1717:
1713:
1692:
1688:
1673:
1637:
1633:
1623:
1621:
1619:ai.facebook.com
1613:
1612:
1601:
1569:
1563:
1556:
1546:
1544:
1542:ai.facebook.com
1536:
1535:
1528:
1513:
1477:
1473:
1422:
1418:
1403:
1367:
1363:
1348:
1312:
1305:
1290:
1254:
1250:
1240:
1238:
1218:
1214:
1204:
1202:
1192:
1185:
1180:
1169:gradient update
1097:
1049:
1040:
1023:
1003:
998:
988:and others for
948:
919:
918:
892:
884:
883:
844:
836:
835:
796:Kernel machines
791:
783:
782:
758:
750:
749:
730:Active learning
725:
717:
716:
685:
675:
674:
600:Diffusion model
536:
526:
525:
498:
488:
487:
461:
451:
450:
406:Factor analysis
401:
391:
390:
374:
337:
327:
326:
247:
246:
230:
229:
228:
217:
216:
122:
114:
113:
79:Online learning
44:
32:
17:
12:
11:
5:
2103:
2093:
2092:
2087:
2073:
2072:
2042:
1993:
1979:
1944:
1930:
1893:
1892:External links
1890:
1889:
1888:
1870:
1867:
1864:
1863:
1804:
1782:
1737:
1724:Google AI Blog
1711:
1686:
1671:
1631:
1599:
1580:(2): 233–243.
1554:
1526:
1511:
1471:
1416:
1401:
1361:
1346:
1303:
1288:
1248:
1212:
1182:
1181:
1179:
1176:
1133:language model
1096:
1093:
1048:
1045:
1039:
1036:
1022:
1019:
1002:
999:
997:
994:
950:
949:
947:
946:
939:
932:
924:
921:
920:
917:
916:
911:
910:
909:
899:
893:
890:
889:
886:
885:
882:
881:
876:
871:
866:
861:
856:
851:
845:
842:
841:
838:
837:
834:
833:
828:
823:
818:
816:Occam learning
813:
808:
803:
798:
792:
789:
788:
785:
784:
781:
780:
775:
773:Learning curve
770:
765:
759:
756:
755:
752:
751:
748:
747:
742:
737:
732:
726:
723:
722:
719:
718:
715:
714:
713:
712:
702:
697:
692:
686:
681:
680:
677:
676:
673:
672:
666:
661:
656:
651:
650:
649:
639:
634:
633:
632:
627:
622:
617:
607:
602:
597:
592:
591:
590:
580:
579:
578:
573:
568:
563:
553:
548:
543:
537:
532:
531:
528:
527:
524:
523:
518:
513:
505:
499:
494:
493:
490:
489:
486:
485:
484:
483:
478:
473:
462:
457:
456:
453:
452:
449:
448:
443:
438:
433:
428:
423:
418:
413:
408:
402:
397:
396:
393:
392:
389:
388:
383:
378:
372:
367:
362:
354:
349:
344:
338:
333:
332:
329:
328:
325:
324:
319:
314:
309:
304:
299:
294:
289:
281:
280:
279:
274:
269:
259:
257:Decision trees
254:
248:
234:classification
224:
223:
222:
219:
218:
215:
214:
209:
204:
199:
194:
189:
184:
179:
174:
169:
164:
159:
154:
149:
144:
139:
134:
129:
127:Classification
123:
120:
119:
116:
115:
112:
111:
106:
101:
96:
91:
86:
84:Batch learning
81:
76:
71:
66:
61:
56:
51:
45:
42:
41:
38:
37:
26:
25:
15:
9:
6:
4:
3:
2:
2102:
2091:
2088:
2086:
2083:
2082:
2080:
2061:
2056:
2052:
2048:
2043:
2039:
2035:
2031:
2027:
2023:
2019:
2015:
2011:
2007:
2003:
1999:
1994:
1990:
1986:
1982:
1976:
1972:
1968:
1963:
1958:
1954:
1950:
1945:
1941:
1937:
1933:
1927:
1923:
1919:
1914:
1909:
1905:
1901:
1896:
1895:
1884:
1879:
1873:
1872:
1859:
1855:
1850:
1845:
1841:
1837:
1832:
1827:
1823:
1819:
1815:
1808:
1799:
1794:
1786:
1778:
1774:
1770:
1766:
1761:
1756:
1752:
1748:
1741:
1725:
1721:
1715:
1706:
1701:
1697:
1690:
1682:
1678:
1674:
1668:
1664:
1660:
1655:
1650:
1646:
1642:
1635:
1620:
1616:
1610:
1608:
1606:
1604:
1595:
1591:
1587:
1583:
1579:
1575:
1574:AIChE Journal
1568:
1561:
1559:
1543:
1539:
1533:
1531:
1522:
1518:
1514:
1508:
1504:
1500:
1495:
1490:
1486:
1482:
1475:
1467:
1463:
1459:
1455:
1451:
1447:
1443:
1439:
1435:
1431:
1427:
1420:
1412:
1408:
1404:
1398:
1394:
1390:
1385:
1380:
1376:
1372:
1365:
1357:
1353:
1349:
1343:
1339:
1335:
1330:
1325:
1321:
1317:
1310:
1308:
1299:
1295:
1291:
1285:
1281:
1277:
1272:
1267:
1263:
1259:
1252:
1236:
1231:
1227:
1223:
1216:
1201:
1197:
1190:
1188:
1183:
1175:
1172:
1170:
1165:
1163:
1159:
1155:
1151:
1146:
1144:
1140:
1136:
1134:
1130:
1126:
1122:
1120:
1116:
1112:
1110:
1106:
1102:
1092:
1090:
1085:
1082:
1077:
1075:
1070:
1068:
1064:
1060:
1057:
1055:
1044:
1035:
1033:
1032:loss function
1028:
1027:training data
1018:
1014:
1012:
1007:
993:
991:
987:
983:
978:
976:
975:decision list
972:
967:
964:
960:
956:
945:
940:
938:
933:
931:
926:
925:
923:
922:
915:
912:
908:
905:
904:
903:
900:
898:
895:
894:
888:
887:
880:
877:
875:
872:
870:
867:
865:
862:
860:
857:
855:
852:
850:
847:
846:
840:
839:
832:
829:
827:
824:
822:
819:
817:
814:
812:
809:
807:
804:
802:
799:
797:
794:
793:
787:
786:
779:
776:
774:
771:
769:
766:
764:
761:
760:
754:
753:
746:
743:
741:
738:
736:
735:Crowdsourcing
733:
731:
728:
727:
721:
720:
711:
708:
707:
706:
703:
701:
698:
696:
693:
691:
688:
687:
684:
679:
678:
670:
667:
665:
664:Memtransistor
662:
660:
657:
655:
652:
648:
645:
644:
643:
640:
638:
635:
631:
628:
626:
623:
621:
618:
616:
613:
612:
611:
608:
606:
603:
601:
598:
596:
593:
589:
586:
585:
584:
581:
577:
574:
572:
569:
567:
564:
562:
559:
558:
557:
554:
552:
549:
547:
546:Deep learning
544:
542:
539:
538:
535:
530:
529:
522:
519:
517:
514:
512:
510:
506:
504:
501:
500:
497:
492:
491:
482:
481:Hidden Markov
479:
477:
474:
472:
469:
468:
467:
464:
463:
460:
455:
454:
447:
444:
442:
439:
437:
434:
432:
429:
427:
424:
422:
419:
417:
414:
412:
409:
407:
404:
403:
400:
395:
394:
387:
384:
382:
379:
377:
373:
371:
368:
366:
363:
361:
359:
355:
353:
350:
348:
345:
343:
340:
339:
336:
331:
330:
323:
320:
318:
315:
313:
310:
308:
305:
303:
300:
298:
295:
293:
290:
288:
286:
282:
278:
277:Random forest
275:
273:
270:
268:
265:
264:
263:
260:
258:
255:
253:
250:
249:
242:
241:
236:
235:
227:
221:
220:
213:
210:
208:
205:
203:
200:
198:
195:
193:
190:
188:
185:
183:
180:
178:
175:
173:
170:
168:
165:
163:
162:Data cleaning
160:
158:
155:
153:
150:
148:
145:
143:
140:
138:
135:
133:
130:
128:
125:
124:
118:
117:
110:
107:
105:
102:
100:
97:
95:
92:
90:
87:
85:
82:
80:
77:
75:
74:Meta-learning
72:
70:
67:
65:
62:
60:
57:
55:
52:
50:
47:
46:
40:
39:
36:
31:
28:
27:
23:
22:
19:
2064:. Retrieved
2050:
2005:
2001:
1952:
1903:
1821:
1817:
1807:
1785:
1750:
1740:
1728:. Retrieved
1723:
1714:
1695:
1689:
1644:
1634:
1622:. Retrieved
1618:
1577:
1573:
1545:. Retrieved
1541:
1484:
1474:
1433:
1429:
1419:
1374:
1364:
1319:
1261:
1251:
1239:. Retrieved
1225:
1215:
1203:. Retrieved
1199:
1173:
1166:
1147:
1138:
1137:
1123:
1113:
1104:
1098:
1086:
1079:Training an
1078:
1071:
1061:
1058:
1050:
1041:
1024:
1015:
1011:autoencoders
1008:
1004:
979:
968:
958:
954:
953:
821:PAC learning
508:
357:
352:Hierarchical
284:
238:
232:
63:
18:
1791:Learning".
1081:autoencoder
705:Multi-agent
642:Transformer
541:Autoencoder
297:Naive Bayes
35:data mining
2079:Categories
2066:1 November
1962:1505.05192
1913:1708.07860
1883:2304.12210
1824:(1): 928.
1798:2006.07733
1760:2010.05725
1705:1809.04506
1654:1511.09033
1494:1906.05186
1384:1505.05192
1329:1905.03670
1271:1708.07860
1241:1 November
1178:References
1162:polysemous
1158:word sense
1103:developed
690:Q-learning
588:Restricted
386:Mean shift
335:Clustering
312:Perceptron
240:regression
142:Clustering
137:Regression
2022:0968-4328
2008:: 55–71.
1840:2399-3642
1777:222291675
1624:5 October
1521:186206588
1450:0968-4328
1436:: 55–71.
1356:167209887
849:ECML PKDD
831:VC theory
778:ROC curve
710:Self-play
630:DeepDream
471:Bayes net
262:Ensembles
43:Paradigms
2030:29425969
1858:37696966
1849:10495322
1458:29425969
1143:ImageNet
1101:Facebook
1095:Examples
986:Facebook
272:Boosting
121:Problems
2038:3796689
1989:9062671
1681:6517610
1582:Bibcode
1466:3796689
1411:9062671
1105:wav2vec
1067:labeled
854:NeurIPS
671:(ECRAM)
625:AlexNet
267:Bagging
2036:
2028:
2020:
2002:Micron
1987:
1977:
1940:473729
1938:
1928:
1856:
1846:
1838:
1775:
1730:9 June
1679:
1669:
1547:9 June
1519:
1509:
1464:
1456:
1448:
1430:Micron
1409:
1399:
1354:
1344:
1298:473729
1296:
1286:
1205:9 June
1200:Medium
1125:OpenAI
1115:Google
647:Vision
503:RANSAC
381:OPTICS
376:DBSCAN
360:-means
167:AutoML
2034:S2CID
1985:S2CID
1957:arXiv
1936:S2CID
1908:arXiv
1878:arXiv
1793:arXiv
1773:S2CID
1755:arXiv
1700:arXiv
1677:S2CID
1649:arXiv
1570:(PDF)
1517:S2CID
1489:arXiv
1462:S2CID
1407:S2CID
1379:arXiv
1352:S2CID
1324:arXiv
1294:S2CID
1266:arXiv
1160:of a
1129:GPT-3
996:Types
869:IJCAI
695:SARSA
654:Mamba
620:LeNet
615:U-Net
441:t-SNE
365:Fuzzy
342:BIRCH
2068:2022
2026:PMID
2018:ISSN
1975:ISBN
1926:ISBN
1854:PMID
1836:ISSN
1732:2021
1667:ISBN
1626:2021
1549:2021
1507:ISBN
1454:PMID
1446:ISSN
1397:ISBN
1342:ISBN
1284:ISBN
1243:2022
1207:2021
1148:The
879:JMLR
864:ICLR
859:ICML
745:RLHF
561:LSTM
347:CURE
33:and
2055:doi
2010:doi
2006:107
1967:doi
1918:doi
1844:PMC
1826:doi
1765:doi
1659:doi
1590:doi
1499:doi
1438:doi
1434:107
1389:doi
1334:doi
1276:doi
1230:doi
1127:'s
1117:'s
1087:In
1072:In
992:.
959:SSL
605:SOM
595:GAN
571:ESN
566:GRU
511:-NN
446:SDL
436:PGD
431:PCA
426:NMF
421:LDA
416:ICA
411:CCA
287:-NN
2081::
2049:.
2032:.
2024:.
2016:.
2004:.
2000:.
1983:.
1973:.
1965:.
1951:.
1934:.
1924:.
1916:.
1902:.
1852:.
1842:.
1834:.
1820:.
1816:.
1771:.
1763:.
1749:.
1722:.
1698:.
1675:.
1665:.
1657:.
1643:.
1617:.
1602:^
1588:.
1578:37
1576:.
1572:.
1557:^
1540:.
1529:^
1515:.
1505:.
1497:.
1483:.
1460:.
1452:.
1444:.
1432:.
1428:.
1405:.
1395:.
1387:.
1373:.
1350:.
1340:.
1332:.
1318:.
1306:^
1292:.
1282:.
1274:.
1260:.
1224:.
1198:.
1186:^
1171:.
1069:.
874:ML
2070:.
2057::
2040:.
2012::
1991:.
1969::
1959::
1942:.
1920::
1910::
1886:.
1880::
1860:.
1828::
1822:6
1801:.
1795::
1779:.
1767::
1757::
1734:.
1708:.
1702::
1683:.
1661::
1651::
1628:.
1596:.
1592::
1584::
1551:.
1523:.
1501::
1491::
1468:.
1440::
1413:.
1391::
1381::
1358:.
1336::
1326::
1300:.
1278::
1268::
1245:.
1232::
1209:.
957:(
943:e
936:t
929:v
509:k
358:k
285:k
243:)
231:(
Text is available under the Creative Commons Attribution-ShareAlike License. Additional terms may apply.