Effective Prevention of Semantic Drift in Continual Deep Learning
Lifelong machine learning or continual learning models attempt to learn incrementally by accumulating knowledge across a sequence of tasks. Therefore, these models learn better and faster. They are used in various intelligent systems that have to interact with humans or any dynamic environment. Dyna...
Saved in:
Published in | Intelligent Data Engineering and Automated Learning – IDEAL 2022 pp. 456 - 464 |
---|---|
Main Authors | , |
Format | Book Chapter |
Language | English |
Published |
Cham
Springer International Publishing
|
Series | Lecture Notes in Computer Science |
Subjects | |
Online Access | Get full text |
Cover
Loading…
Abstract | Lifelong machine learning or continual learning models attempt to learn incrementally by accumulating knowledge across a sequence of tasks. Therefore, these models learn better and faster. They are used in various intelligent systems that have to interact with humans or any dynamic environment. Dynamically expandable networks are continual deep learning models that allow its architecture to expand with a sequence of tasks. The model retains knowledge from the previous tasks that results in high performance on newer tasks. The existing models use Minkowski distance measures to separate nodes of the current network, resulting in higher catastrophic forgetting. These measures are susceptible to high dimensional sparse vectors, resulting in sub-optimum performance. We propose ang-DEN, as a dynamically expanding continual learning architecture that use angular distance metric. It addresses semantic drift through better separation of nodes achieving 97% average accuracy with an improvement of 1.3% across all tasks on MNIST variant datasets. |
---|---|
AbstractList | Lifelong machine learning or continual learning models attempt to learn incrementally by accumulating knowledge across a sequence of tasks. Therefore, these models learn better and faster. They are used in various intelligent systems that have to interact with humans or any dynamic environment. Dynamically expandable networks are continual deep learning models that allow its architecture to expand with a sequence of tasks. The model retains knowledge from the previous tasks that results in high performance on newer tasks. The existing models use Minkowski distance measures to separate nodes of the current network, resulting in higher catastrophic forgetting. These measures are susceptible to high dimensional sparse vectors, resulting in sub-optimum performance. We propose ang-DEN, as a dynamically expanding continual learning architecture that use angular distance metric. It addresses semantic drift through better separation of nodes achieving 97% average accuracy with an improvement of 1.3% across all tasks on MNIST variant datasets. |
Author | Taimoor Khan, Muhammad Saadi, Khouloud |
Author_xml | – sequence: 1 givenname: Khouloud surname: Saadi fullname: Saadi, Khouloud – sequence: 2 givenname: Muhammad orcidid: 0000-0002-6542-9217 surname: Taimoor Khan fullname: Taimoor Khan, Muhammad email: taimoor.muhammad@gmail.com |
BookMark | eNpVkE1OwzAQRg0Uibb0BixyAcPY49T2EqXlR4oEErC2HHeCAsWpktDz9Cw9GW5hw2o075Nm9L0JG8U2EmNXAq4FgL6x2nDkgIJLoXPkwil1wmYJY4JHJk7ZWMyF4IjKnv3LpBqxMSBIbrXCCzbp-w8AkNrKMSuWdU1haLaUPXe0pTg0bczaer97oS-ftpAtuqYesibud0WbQPz262xBtMlK8l1s4vslO6_9uqfZ35yyt7vla_HAy6f7x-K25L2wZuBBGQ3KBvDGBL3SylSEuJIKc68kyLxCELleeRHAeiRrKEAlgyXlpdZznDL5e7ffdOktda5q28_eCXAHTS51duhSa3dU4g6a8AcDLlhW |
ContentType | Book Chapter |
Copyright | The Author(s), under exclusive license to Springer Nature Switzerland AG 2022 |
Copyright_xml | – notice: The Author(s), under exclusive license to Springer Nature Switzerland AG 2022 |
DOI | 10.1007/978-3-031-21753-1_44 |
DatabaseTitleList | |
DeliveryMethod | fulltext_linktorsrc |
Discipline | Computer Science |
EISBN | 9783031217531 3031217535 |
EISSN | 1611-3349 |
Editor | Camacho, David Yin, Hujun Tino, Peter |
Editor_xml | – sequence: 1 givenname: Hujun orcidid: 0000-0002-9198-5401 surname: Yin fullname: Yin, Hujun email: hujun.yin@manchester.ac.uk – sequence: 2 givenname: David orcidid: 0000-0002-0535-7450 surname: Camacho fullname: Camacho, David email: david.camacho@upm.es – sequence: 3 givenname: Peter orcidid: 0000-0003-2330-128X surname: Tino fullname: Tino, Peter email: p.tino@cs.bham.ac.uk |
EndPage | 464 |
GroupedDBID | 38. AABBV AAZWU ABSVR ABTHU ABVND ACHZO ACPMC ADNVS AEDXK AEJLV AEKFX AHVRR AIYYB ALMA_UNASSIGNED_HOLDINGS BBABE CZZ IEZ SBO TPJZQ TSXQS Z5O Z7R Z7U Z7W Z7X Z7Y Z7Z Z81 Z83 Z84 Z85 Z87 Z88 |
ID | FETCH-LOGICAL-s198t-c487049c0a88c7d748be33d2435a42025b30157da1c09a3e98ec0b2c9e4a27763 |
ISBN | 9783031217524 3031217527 |
ISSN | 0302-9743 |
IngestDate | Tue Oct 01 19:40:36 EDT 2024 |
IsPeerReviewed | true |
IsScholarly | true |
Language | English |
LinkModel | OpenURL |
MergedId | FETCHMERGED-LOGICAL-s198t-c487049c0a88c7d748be33d2435a42025b30157da1c09a3e98ec0b2c9e4a27763 |
Notes | K. Saadi—Independent Researcher.This work was supported by Fatima Al-Fihri predoctoral fellowship program (https://fatimafellowship.com/). |
ORCID | 0000-0002-6542-9217 |
PageCount | 9 |
ParticipantIDs | springer_books_10_1007_978_3_031_21753_1_44 |
PublicationPlace | Cham |
PublicationPlace_xml | – name: Cham |
PublicationSeriesTitle | Lecture Notes in Computer Science |
PublicationSeriesTitleAlternate | Lect.Notes Computer |
PublicationSubtitle | 23rd International Conference, IDEAL 2022, Manchester, UK, November 24–26, 2022, Proceedings |
PublicationTitle | Intelligent Data Engineering and Automated Learning – IDEAL 2022 |
Publisher | Springer International Publishing |
Publisher_xml | – name: Springer International Publishing |
RelatedPersons | Hartmanis, Juris Gao, Wen Steffen, Bernhard Bertino, Elisa Goos, Gerhard Yung, Moti |
RelatedPersons_xml | – sequence: 1 givenname: Gerhard surname: Goos fullname: Goos, Gerhard – sequence: 2 givenname: Juris surname: Hartmanis fullname: Hartmanis, Juris – sequence: 3 givenname: Elisa surname: Bertino fullname: Bertino, Elisa – sequence: 4 givenname: Wen surname: Gao fullname: Gao, Wen – sequence: 5 givenname: Bernhard orcidid: 0000-0001-9619-1558 surname: Steffen fullname: Steffen, Bernhard – sequence: 6 givenname: Moti orcidid: 0000-0003-0848-0873 surname: Yung fullname: Yung, Moti |
SSID | ssj0002792 ssj0002829692 |
Score | 2.2390158 |
Snippet | Lifelong machine learning or continual learning models attempt to learn incrementally by accumulating knowledge across a sequence of tasks. Therefore, these... |
SourceID | springer |
SourceType | Publisher |
StartPage | 456 |
SubjectTerms | ang-DEN Catastrophic forgetting Continual deep learning Semantic drift |
Title | Effective Prevention of Semantic Drift in Continual Deep Learning |
URI | http://link.springer.com/10.1007/978-3-031-21753-1_44 |
hasFullText | 1 |
inHoldings | 1 |
isFullTextHit | |
isPrint | |
link | http://utb.summon.serialssolutions.com/2.0.0/link/0/eLvHCXMwnV27btswFCWcZOvQNmnRFk3AoZkEFZJIieLQwbBdpIGTJQ9kE_gyYqC2CltaOvdD8i35sl6KoizHWZJFMAjDpO-x7_seIvSNxCLSsbJF94SHNDMs5MKQUEujcy5SHrsG2cvs7Iae36V3g8G_XtdSXcnv6u-zcyWvQRXWAFc7JfsCZLsPhQV4DfjCExCG5xPndzvN2rULOjLNCrCrRJ9asCkJDOuqBH8UPMqpz3_41gYS_BpPhtMgiZKuu_hKCO1K-_dl_bus9SaoBzzLFay7ZOlFfS8WC7HVCe9IkG0XkueEavzQ01FyOoyuzALwm6tgvJrP7JUEbtkSY80bdv-xMX-6Qzo9Z_mX1z-mbYXjsqyaxrHAX0LhddJOgvJJinOTZduKaMGixhAlpW6w2k92gdaGuMcpQuMUdWbpF4mjO22VL02znh2njh19x0T0u0Jgs9DuRsK4oHQP7TEOmvFgODmf3nb23FIsdmk7W3jOeFubcqeyE0P-1MxxOm2-RW9a87ktd-rvjVtz_Q69saMu2M6ggNDeo4FZHqK3Xsa4lfERGnXw4g28uJw9PnhocQMtni8fHzpYsYUVe1g_oJufk-vRWdjevhGuY55XoYJQFsJHFYk8V0wzmktDiE7AvxYUfp6pBNuQMi1iFXFBDM-NimSiuKEiYWC2PqL9Zbk0nxA2qWQEFD3RGaegCeQsISRl9qoFwXMhP6PAS6Gw_6d14cm0QWYFKUBmRSOzwsrsy4ve_RXtV6vaHIPrWMmTFtv_x3Zmkg |
link.rule.ids | 785,786,790,799,27958 |
linkProvider | Library Specific Holdings |
openUrl | ctx_ver=Z39.88-2004&ctx_enc=info%3Aofi%2Fenc%3AUTF-8&rfr_id=info%3Asid%2Fsummon.serialssolutions.com&rft_val_fmt=info%3Aofi%2Ffmt%3Akev%3Amtx%3Abook&rft.genre=bookitem&rft.title=Intelligent+Data+Engineering+and+Automated+Learning+%E2%80%93+IDEAL+2022&rft.au=Saadi%2C+Khouloud&rft.au=Taimoor+Khan%2C+Muhammad&rft.atitle=Effective+Prevention+of%C2%A0Semantic+Drift+in%C2%A0Continual+Deep+Learning&rft.series=Lecture+Notes+in+Computer+Science&rft.pub=Springer+International+Publishing&rft.isbn=9783031217524&rft.issn=0302-9743&rft.eissn=1611-3349&rft.spage=456&rft.epage=464&rft_id=info:doi/10.1007%2F978-3-031-21753-1_44 |
thumbnail_l | http://covers-cdn.summon.serialssolutions.com/index.aspx?isbn=/lc.gif&issn=0302-9743&client=summon |
thumbnail_m | http://covers-cdn.summon.serialssolutions.com/index.aspx?isbn=/mc.gif&issn=0302-9743&client=summon |
thumbnail_s | http://covers-cdn.summon.serialssolutions.com/index.aspx?isbn=/sc.gif&issn=0302-9743&client=summon |