1326:. Feature construction is the application of a set of constructive operators to a set of existing features resulting in construction of new features. Examples of such constructive operators include checking for the equality conditions {=, â }, the arithmetic operators {+,â,Ă, /}, the array operators {max(S), min(S), average(S)} as well as other more sophisticated operators, for example count(S,C) that counts the number of features in the feature vector S satisfying some condition C or, for example, distances to other recognition classes generalized by some accepting device. Feature construction has long been considered a powerful tool for increasing both accuracy and understanding of structure, particularly in high-dimensional problems. Applications include studies of disease and
43:
1279:
in machine learning require a numerical representation of objects, since such representations facilitate processing and statistical analysis. When representing images, the feature values might correspond to the pixels of an image, while when representing texts the features might be the frequencies of
1142:
The type of feature that is used in feature engineering depends on the specific machine learning algorithm that is being used. Some machine learning algorithms, such as decision trees, can handle both numerical and categorical features. Other machine learning algorithms, such as linear regression,
1138:
are discrete values that can be grouped into categories. Examples of categorical features include gender, color, and zip code. Categorical features typically need to be converted to numerical features before they can be used in machine learning algorithms. This can be done using a variety of
1131:
Numerical features are continuous values that can be measured on a scale. Examples of numerical features include age, height, weight, and income. Numerical features can be used in machine learning algorithms directly.
1486:
Bloedorn, E., Michalski, R. Data-driven constructive induction: a methodology and its applications. IEEE Intelligent
Systems, Special issue on Feature Transformation and Subset Selection, pp. 30-37, March/April,
1226:
detection algorithms, features may include the presence or absence of certain email headers, the email structure, the language, the frequency of specific terms, the grammatical correctness of the text.
1318:
Higher-level features can be obtained from already available features and added to the feature vector; for example, for the study of diseases the feature 'Age' is useful and is defined as
986:
1348:
The initial set of raw features can be redundant and large enough that estimation and optimization is made difficult or ineffective. Therefore, a preliminary step in many applications of
1024:
1084:
is an individual measurable property or characteristic of a phenomenon. Choosing informative, discriminating and independent features is a crucial element of effective algorithms in
981:
971:
812:
1019:
976:
827:
558:
60:
1059:
862:
938:
487:
1371:. It requires the experimentation of multiple possibilities and the combination of automated techniques with the intuition and knowledge of the
1534:
996:
759:
294:
1208:
counting the number of black pixels along horizontal and vertical directions, number of internal holes, stroke detection and many others.
1413:
1014:
107:
847:
822:
771:
79:
1442:
1089:
895:
890:
543:
126:
553:
191:
86:
948:
1052:
712:
533:
64:
1105:
923:
625:
401:
93:
1566:
1171:
1101:
880:
817:
727:
705:
548:
538:
1031:
943:
928:
389:
211:
918:
75:
1167:
between the feature vector and a vector of weights, qualifying those observations whose result exceeds a threshold.
1408:
1179:
991:
668:
563:
351:
284:
244:
53:
1561:
1297:
1235:
1195:
1175:
1156:
1097:
1045:
651:
419:
289:
31:
1393:
1312:
673:
593:
516:
434:
264:
226:
221:
181:
176:
1367:
Extracting or selecting features is a combination of art and science; developing systems to do so is known as
1364:
a new and reduced set of features to facilitate learning, and to improve generalization and interpretability.
620:
469:
369:
196:
800:
776:
678:
439:
414:
374:
186:
1513:. Internet Technology and Secured Transactions Conference 2009 (ICITST-2009), London, November 9â12. IEEE
1201:
1152:
754:
576:
528:
384:
299:
171:
1474:
100:
1556:
1281:
1135:
1109:
683:
633:
1477:. In Journal of Expert Systems with Applications. Vol. 36 , Iss. 2 (March 2009), pp. 3401-3406, 2009
1398:
1368:
1353:
1327:
1260:
1093:
1085:
1077:
786:
722:
693:
598:
424:
357:
343:
329:
304:
254:
206:
166:
1361:
1343:
1272:
1212:
1183:
764:
688:
474:
269:
1530:
1438:
1357:
1339:
1289:
1223:
1117:
857:
700:
613:
409:
379:
324:
319:
274:
216:
1376:
1349:
1264:
1219:
can include noise ratios, length of sounds, relative power, filter matches and many others.
1128:
In feature engineering, two types of features are commonly used: numerical and categorical.
1073:
885:
638:
588:
498:
482:
452:
314:
309:
259:
249:
147:
1231:
913:
717:
583:
523:
1510:
1151:
A numeric feature can be conveniently described by a feature vector. One way to achieve
1511:
Syntactic learning for ESEDA.1, tool for enhanced speech emotion detection and analysis
1379:, where a machine not only uses features for learning, but learns the features itself.
1247:
933:
464:
201:
1550:
1403:
1372:
1253:
852:
781:
663:
394:
279:
1304:
1524:
1459:
1293:
1164:
658:
152:
42:
1285:
1280:
occurrence of textual terms. Feature vectors are equivalent to the vectors of
1160:
1113:
807:
503:
429:
17:
1526:
The
Elements of Statistical Learning: Data Mining, Inference, and Prediction
1388:
1276:
1205:
1139:
techniques, such as one-hot encoding, label encoding, and ordinal encoding.
966:
747:
1252:"Feature space" redirects here. For feature spaces in kernel machines, see
1475:
Iterative feature construction for improving inductive learning algorithms
1311:. In order to reduce the dimensionality of the feature space, a number of
1216:
1163:) with a feature vector as input. The method consists of calculating the
742:
493:
737:
732:
459:
1523:
Hastie, Trevor; Tibshirani, Robert; Friedman, Jerome H. (2009).
1096:. Features are usually numeric, but structural features such as
36:
1170:
Algorithms for classification from a feature vector include
1300:
that is used to determine a score for making a prediction.
1292:. Feature vectors are often combined with weights using a
1460:
Feature
Selection for Knowledge Discovery and Data Mining
1025:
List of datasets in computer vision and image processing
1496:
Breiman, L. Friedman, T., Olshen, R., Stone, C. (1984)
1275:
of numerical features that represent some object. Many
1464:, Kluwer Academic Publishers. Norwell, MA, USA. 1998.
67:. Unsourced material may be challenged and removed.
1307:associated with these vectors is often called the
1108:. The concept of "feature" is related to that of
1020:List of datasets for machine-learning research
1053:
8:
1320:Age = 'Year of death' minus 'Year of birth'
1060:
1046:
138:
127:Learn how and when to remove this message
1435:Pattern recognition and machine learning
1425:
1234:, there are a large number of possible
146:
27:Measurable property or characteristic
7:
1143:can only handle numerical features.
65:adding citations to reliable sources
1498:Classification and regression trees
1414:Explainable artificial intelligence
1015:Glossary of artificial intelligence
25:
1322:. This process is referred to as
41:
1172:nearest neighbor classification
76:"Feature" machine learning
52:needs additional citations for
435:Relevance vector machine (RVM)
1:
1375:. Automating this process is
1238:, such as edges and objects.
1106:syntactic pattern recognition
924:Computational learning theory
488:Expectationâmaximization (EM)
1473:Piramuthu, S., Sikora R. T.
1433:Bishop, Christopher (2006).
1315:techniques can be employed.
881:Coefficient of determination
728:Convolutional neural network
440:Support vector machine (SVM)
1215:, features for recognizing
1032:Outline of machine learning
929:Empirical risk minimization
1583:
1457:Liu, H., Motoda H. (1998)
1409:Statistical classification
1337:
1251:
1245:
1193:
669:Feedforward neural network
420:Artificial neural networks
29:
1360:a subset of features, or
1298:linear predictor function
1196:Feature (computer vision)
1157:linear predictor function
652:Artificial neural network
32:Feature (computer vision)
1394:Dimensionality reduction
1334:Selection and extraction
1313:dimensionality reduction
1296:in order to construct a
961:Journals and conferences
908:Mathematical foundations
818:Temporal difference (TD)
674:Recurrent neural network
594:Conditional random field
517:Dimensionality reduction
265:Dimensionality reduction
227:Quantum machine learning
222:Neuromorphic engineering
182:Self-supervised learning
177:Semi-supervised learning
30:Not to be confused with
1509:Sidorova, J., Badia T.
1204:, features may include
370:Apprenticeship learning
1180:statistical techniques
919:Biasâvariance tradeoff
801:Reinforcement learning
777:Spiking neural network
187:Reinforcement learning
1282:explanatory variables
1202:character recognition
1153:binary classification
755:Neural radiance field
577:Structured prediction
300:Structured prediction
172:Unsupervised learning
1437:. Berlin: Springer.
1324:feature construction
1271:is an n-dimensional
1136:Categorical features
1110:explanatory variable
944:Statistical learning
842:Learning with humans
634:Local outlier factor
61:improve this article
1567:Pattern recognition
1399:Feature engineering
1369:feature engineering
1354:pattern recognition
1328:emotion recognition
1288:procedures such as
1261:pattern recognition
1184:Bayesian approaches
1116:techniques such as
1086:pattern recognition
1078:pattern recognition
787:Electrochemical RAM
694:reservoir computing
425:Logistic regression
344:Supervised learning
330:Multimodal learning
305:Feature engineering
250:Generative modeling
212:Rule-based learning
207:Curriculum learning
167:Supervised learning
142:Part of a series on
1344:Feature extraction
1213:speech recognition
355: •
270:Density estimation
1536:978-0-387-84884-6
1340:Feature selection
1290:linear regression
1118:linear regression
1070:
1069:
875:Model diagnostics
858:Human-in-the-loop
701:Boltzmann machine
614:Anomaly detection
410:Linear regression
325:Ontology learning
320:Grammar induction
295:Semantic analysis
290:Association rules
275:Anomaly detection
217:Neuro-symbolic AI
137:
136:
129:
111:
16:(Redirected from
1574:
1562:Machine learning
1541:
1540:
1520:
1514:
1507:
1501:
1494:
1488:
1484:
1478:
1471:
1465:
1455:
1449:
1448:
1430:
1377:feature learning
1350:machine learning
1265:machine learning
1159:(related to the
1074:machine learning
1062:
1055:
1048:
1009:Related articles
886:Confusion matrix
639:Isolation forest
584:Graphical models
363:
362:
315:Learning to rank
310:Feature learning
148:Machine learning
139:
132:
125:
121:
118:
112:
110:
69:
45:
37:
21:
1582:
1581:
1577:
1576:
1575:
1573:
1572:
1571:
1547:
1546:
1545:
1544:
1537:
1522:
1521:
1517:
1508:
1504:
1495:
1491:
1485:
1481:
1472:
1468:
1456:
1452:
1445:
1432:
1431:
1427:
1422:
1385:
1346:
1338:Main articles:
1336:
1257:
1250:
1244:
1242:Feature vectors
1232:computer vision
1198:
1192:
1176:neural networks
1149:
1126:
1066:
1037:
1036:
1010:
1002:
1001:
962:
954:
953:
914:Kernel machines
909:
901:
900:
876:
868:
867:
848:Active learning
843:
835:
834:
803:
793:
792:
718:Diffusion model
654:
644:
643:
616:
606:
605:
579:
569:
568:
524:Factor analysis
519:
509:
508:
492:
455:
445:
444:
365:
364:
348:
347:
346:
335:
334:
240:
232:
231:
197:Online learning
162:
150:
133:
122:
116:
113:
70:
68:
58:
46:
35:
28:
23:
22:
15:
12:
11:
5:
1580:
1578:
1570:
1569:
1564:
1559:
1549:
1548:
1543:
1542:
1535:
1515:
1502:
1489:
1479:
1466:
1450:
1443:
1424:
1423:
1421:
1418:
1417:
1416:
1411:
1406:
1401:
1396:
1391:
1384:
1381:
1335:
1332:
1269:feature vector
1248:Word embedding
1243:
1240:
1191:
1188:
1165:scalar product
1148:
1147:Classification
1145:
1125:
1122:
1090:classification
1068:
1067:
1065:
1064:
1057:
1050:
1042:
1039:
1038:
1035:
1034:
1029:
1028:
1027:
1017:
1011:
1008:
1007:
1004:
1003:
1000:
999:
994:
989:
984:
979:
974:
969:
963:
960:
959:
956:
955:
952:
951:
946:
941:
936:
934:Occam learning
931:
926:
921:
916:
910:
907:
906:
903:
902:
899:
898:
893:
891:Learning curve
888:
883:
877:
874:
873:
870:
869:
866:
865:
860:
855:
850:
844:
841:
840:
837:
836:
833:
832:
831:
830:
820:
815:
810:
804:
799:
798:
795:
794:
791:
790:
784:
779:
774:
769:
768:
767:
757:
752:
751:
750:
745:
740:
735:
725:
720:
715:
710:
709:
708:
698:
697:
696:
691:
686:
681:
671:
666:
661:
655:
650:
649:
646:
645:
642:
641:
636:
631:
623:
617:
612:
611:
608:
607:
604:
603:
602:
601:
596:
591:
580:
575:
574:
571:
570:
567:
566:
561:
556:
551:
546:
541:
536:
531:
526:
520:
515:
514:
511:
510:
507:
506:
501:
496:
490:
485:
480:
472:
467:
462:
456:
451:
450:
447:
446:
443:
442:
437:
432:
427:
422:
417:
412:
407:
399:
398:
397:
392:
387:
377:
375:Decision trees
372:
366:
352:classification
342:
341:
340:
337:
336:
333:
332:
327:
322:
317:
312:
307:
302:
297:
292:
287:
282:
277:
272:
267:
262:
257:
252:
247:
245:Classification
241:
238:
237:
234:
233:
230:
229:
224:
219:
214:
209:
204:
202:Batch learning
199:
194:
189:
184:
179:
174:
169:
163:
160:
159:
156:
155:
144:
143:
135:
134:
49:
47:
40:
26:
24:
18:Feature vector
14:
13:
10:
9:
6:
4:
3:
2:
1579:
1568:
1565:
1563:
1560:
1558:
1555:
1554:
1552:
1538:
1532:
1528:
1527:
1519:
1516:
1512:
1506:
1503:
1499:
1493:
1490:
1483:
1480:
1476:
1470:
1467:
1463:
1461:
1454:
1451:
1446:
1444:0-387-31073-8
1440:
1436:
1429:
1426:
1419:
1415:
1412:
1410:
1407:
1405:
1404:Hashing trick
1402:
1400:
1397:
1395:
1392:
1390:
1387:
1386:
1382:
1380:
1378:
1374:
1373:domain expert
1370:
1365:
1363:
1359:
1355:
1351:
1345:
1341:
1333:
1331:
1330:from speech.
1329:
1325:
1321:
1316:
1314:
1310:
1309:feature space
1306:
1301:
1299:
1295:
1291:
1287:
1283:
1278:
1274:
1270:
1266:
1262:
1255:
1254:Kernel method
1249:
1241:
1239:
1237:
1233:
1228:
1225:
1220:
1218:
1214:
1209:
1207:
1203:
1197:
1189:
1187:
1185:
1181:
1177:
1173:
1168:
1166:
1162:
1158:
1154:
1146:
1144:
1140:
1137:
1133:
1129:
1124:Feature types
1123:
1121:
1119:
1115:
1111:
1107:
1103:
1099:
1095:
1091:
1087:
1083:
1079:
1075:
1063:
1058:
1056:
1051:
1049:
1044:
1043:
1041:
1040:
1033:
1030:
1026:
1023:
1022:
1021:
1018:
1016:
1013:
1012:
1006:
1005:
998:
995:
993:
990:
988:
985:
983:
980:
978:
975:
973:
970:
968:
965:
964:
958:
957:
950:
947:
945:
942:
940:
937:
935:
932:
930:
927:
925:
922:
920:
917:
915:
912:
911:
905:
904:
897:
894:
892:
889:
887:
884:
882:
879:
878:
872:
871:
864:
861:
859:
856:
854:
853:Crowdsourcing
851:
849:
846:
845:
839:
838:
829:
826:
825:
824:
821:
819:
816:
814:
811:
809:
806:
805:
802:
797:
796:
788:
785:
783:
782:Memtransistor
780:
778:
775:
773:
770:
766:
763:
762:
761:
758:
756:
753:
749:
746:
744:
741:
739:
736:
734:
731:
730:
729:
726:
724:
721:
719:
716:
714:
711:
707:
704:
703:
702:
699:
695:
692:
690:
687:
685:
682:
680:
677:
676:
675:
672:
670:
667:
665:
664:Deep learning
662:
660:
657:
656:
653:
648:
647:
640:
637:
635:
632:
630:
628:
624:
622:
619:
618:
615:
610:
609:
600:
599:Hidden Markov
597:
595:
592:
590:
587:
586:
585:
582:
581:
578:
573:
572:
565:
562:
560:
557:
555:
552:
550:
547:
545:
542:
540:
537:
535:
532:
530:
527:
525:
522:
521:
518:
513:
512:
505:
502:
500:
497:
495:
491:
489:
486:
484:
481:
479:
477:
473:
471:
468:
466:
463:
461:
458:
457:
454:
449:
448:
441:
438:
436:
433:
431:
428:
426:
423:
421:
418:
416:
413:
411:
408:
406:
404:
400:
396:
395:Random forest
393:
391:
388:
386:
383:
382:
381:
378:
376:
373:
371:
368:
367:
360:
359:
354:
353:
345:
339:
338:
331:
328:
326:
323:
321:
318:
316:
313:
311:
308:
306:
303:
301:
298:
296:
293:
291:
288:
286:
283:
281:
280:Data cleaning
278:
276:
273:
271:
268:
266:
263:
261:
258:
256:
253:
251:
248:
246:
243:
242:
236:
235:
228:
225:
223:
220:
218:
215:
213:
210:
208:
205:
203:
200:
198:
195:
193:
192:Meta-learning
190:
188:
185:
183:
180:
178:
175:
173:
170:
168:
165:
164:
158:
157:
154:
149:
145:
141:
140:
131:
128:
120:
117:December 2014
109:
106:
102:
99:
95:
92:
88:
85:
81:
78: â
77:
73:
72:Find sources:
66:
62:
56:
55:
50:This article
48:
44:
39:
38:
33:
19:
1529:. Springer.
1525:
1518:
1505:
1497:
1492:
1482:
1469:
1458:
1453:
1434:
1428:
1366:
1362:constructing
1356:consists of
1347:
1323:
1319:
1317:
1308:
1305:vector space
1302:
1268:
1258:
1229:
1221:
1210:
1199:
1169:
1150:
1141:
1134:
1130:
1127:
1104:are used in
1081:
1071:
939:PAC learning
626:
475:
470:Hierarchical
402:
356:
350:
123:
114:
104:
97:
90:
83:
71:
59:Please help
54:verification
51:
1557:Data mining
1500:, Wadsworth
1294:dot product
1286:statistical
1155:is using a
1114:statistical
823:Multi-agent
760:Transformer
659:Autoencoder
415:Naive Bayes
153:data mining
1551:Categories
1420:References
1277:algorithms
1246:See also:
1206:histograms
1194:See also:
1161:perceptron
1094:regression
808:Q-learning
706:Restricted
504:Mean shift
453:Clustering
430:Perceptron
358:regression
260:Clustering
255:Regression
87:newspapers
1389:Covariate
1358:selecting
967:ECML PKDD
949:VC theory
896:ROC curve
828:Self-play
748:DeepDream
589:Bayes net
380:Ensembles
161:Paradigms
1383:See also
1284:used in
1236:features
1217:phonemes
1190:Examples
1182:such as
1112:used in
390:Boosting
239:Problems
1098:strings
1082:feature
972:NeurIPS
789:(ECRAM)
743:AlexNet
385:Bagging
101:scholar
1533:
1441:
1273:vector
1178:, and
1102:graphs
765:Vision
621:RANSAC
499:OPTICS
494:DBSCAN
478:-means
285:AutoML
103:
96:
89:
82:
74:
987:IJCAI
813:SARSA
772:Mamba
738:LeNet
733:U-Net
559:t-SNE
483:Fuzzy
460:BIRCH
108:JSTOR
94:books
1531:ISBN
1487:1998
1439:ISBN
1352:and
1342:and
1303:The
1267:, a
1263:and
1224:spam
1100:and
1092:and
1080:, a
1076:and
997:JMLR
982:ICLR
977:ICML
863:RLHF
679:LSTM
465:CURE
151:and
80:news
1259:In
1230:In
1222:In
1211:In
1200:In
1088:,
1072:In
723:SOM
713:GAN
689:ESN
684:GRU
629:-NN
564:SDL
554:PGD
549:PCA
544:NMF
539:LDA
534:ICA
529:CCA
405:-NN
63:by
1553::
1186:.
1174:,
1120:.
992:ML
1539:.
1462:.
1447:.
1256:.
1061:e
1054:t
1047:v
627:k
476:k
403:k
361:)
349:(
130:)
124:(
119:)
115:(
105:¡
98:¡
91:¡
84:¡
57:.
34:.
20:)
Text is available under the Creative Commons Attribution-ShareAlike License. Additional terms may apply.