DynK-hydra: improved dynamic architecture ensembling for efficient inference

Abstract Accessibility on edge devices and the trade-off between latency and accuracy is an area of interest in deploying deep learning models. This paper explores a Mixture of Experts system, namely, DynK-Hydra, which allows training of an ensemble formed of multiple similar branches on data sets w...
Ausführliche Beschreibung

Gespeichert in:

Autor*in:	Ileni, Tudor Alexandru [verfasserIn] Darabant, Adrian Sergiu Borza, Diana Laura Marinescu, Alexandru Ion

Format:	E-Artikel
Sprache:	Englisch

Erschienen:	2022

Schlagwörter:	Dynamic network Ensemble Optimal inference Conditional computation

Anmerkung:	© The Author(s) 2022

Übergeordnetes Werk:	Enthalten in: Complex & intelligent systems - Berlin : SpringerOpen, 2015, 9(2022), 2 vom: 16. Nov., Seite 2177-2188
Übergeordnetes Werk:	volume:9 ; year:2022 ; number:2 ; day:16 ; month:11 ; pages:2177-2188

Links:	Volltext

DOI / URN:	10.1007/s40747-022-00897-1

Katalog-ID:	SPR050101056

Internformat


LEADER	01000naa a22002652 4500
001	SPR050101056
003	DE-627
005	20230419064805.0
007	cr uuu---uuuuu
008	230419s2022 xx \|\|\|\|\|o 00\| \|\|eng c
024	7		\|a 10.1007/s40747-022-00897-1 \|2 doi
035			\|a (DE-627)SPR050101056
035			\|a (SPR)s40747-022-00897-1-e
040			\|a DE-627 \|b ger \|c DE-627 \|e rakwb
041			\|a eng
100	1		\|a Ileni, Tudor Alexandru \|e verfasserin \|4 aut
245	1	0	\|a DynK-hydra: improved dynamic architecture ensembling for efficient inference
264		1	\|c 2022
336			\|a Text \|b txt \|2 rdacontent
337			\|a Computermedien \|b c \|2 rdamedia
338			\|a Online-Ressource \|b cr \|2 rdacarrier
500			\|a © The Author(s) 2022
520			\|a Abstract Accessibility on edge devices and the trade-off between latency and accuracy is an area of interest in deploying deep learning models. This paper explores a Mixture of Experts system, namely, DynK-Hydra, which allows training of an ensemble formed of multiple similar branches on data sets with a high number of classes, but uses, during the inference, only a subset of necessary branches. We achieve this by training a cohort of specialized branches (deep network of reduced size) and a gater/supervisor, that decides dynamically what branch to use for any specific input. An original contribution is that the number of chosen models is dynamically set, based on how confident the gater is (similar works use a static parameter for this). Another contribution is the way we ensure the branches’ specialization. We divide the data set classes into multiple clusters, and we assign a cluster to each branch while enforcing its specialization on this cluster by a separate loss function. We evaluate DynK-Hydra on CIFAR-100, Food-101, CUB-200, and ImageNet32 data sets and we obtain improvements of up to 4.3% accuracy compared with state-of-the-art ResNet. All this while reducing the number of inference flops by a factor of 2–5.5 times. Compared to a similar work (HydraRes), we obtain marginal accuracy improvements of up to 1.2% on the pairwise inference time architectures. However, we improve the inference times by up to 2.8 times compared to HydraRes.
650		4	\|a Dynamic network \|7 (dpeaa)DE-He213
650		4	\|a Ensemble \|7 (dpeaa)DE-He213
650		4	\|a Optimal inference \|7 (dpeaa)DE-He213
650		4	\|a Conditional computation \|7 (dpeaa)DE-He213
700	1		\|a Darabant, Adrian Sergiu \|0 (orcid)0000-0002-7580-5722 \|4 aut
700	1		\|a Borza, Diana Laura \|4 aut
700	1		\|a Marinescu, Alexandru Ion \|4 aut
773	0	8	\|i Enthalten in \|t Complex & intelligent systems \|d Berlin : SpringerOpen, 2015 \|g 9(2022), 2 vom: 16. Nov., Seite 2177-2188 \|w (DE-627)835589269 \|w (DE-600)2834740-7 \|x 2198-6053 \|7 nnns
773	1	8	\|g volume:9 \|g year:2022 \|g number:2 \|g day:16 \|g month:11 \|g pages:2177-2188
856	4	0	\|u https://dx.doi.org/10.1007/s40747-022-00897-1 \|z kostenfrei \|3 Volltext
912			\|a GBV_USEFLAG_A
912			\|a SYSFLAG_A
912			\|a GBV_SPRINGER
912			\|a GBV_ILN_11
912			\|a GBV_ILN_20
912			\|a GBV_ILN_22
912			\|a GBV_ILN_23
912			\|a GBV_ILN_24
912			\|a GBV_ILN_31
912			\|a GBV_ILN_39
912			\|a GBV_ILN_40
912			\|a GBV_ILN_60
912			\|a GBV_ILN_62
912			\|a GBV_ILN_63
912			\|a GBV_ILN_65
912			\|a GBV_ILN_69
912			\|a GBV_ILN_70
912			\|a GBV_ILN_73
912			\|a GBV_ILN_95
912			\|a GBV_ILN_105
912			\|a GBV_ILN_110
912			\|a GBV_ILN_151
912			\|a GBV_ILN_161
912			\|a GBV_ILN_170
912			\|a GBV_ILN_213
912			\|a GBV_ILN_230
912			\|a GBV_ILN_285
912			\|a GBV_ILN_293
912			\|a GBV_ILN_370
912			\|a GBV_ILN_602
912			\|a GBV_ILN_2014
912			\|a GBV_ILN_4012
912			\|a GBV_ILN_4037
912			\|a GBV_ILN_4112
912			\|a GBV_ILN_4125
912			\|a GBV_ILN_4126
912			\|a GBV_ILN_4249
912			\|a GBV_ILN_4305
912			\|a GBV_ILN_4306
912			\|a GBV_ILN_4307
912			\|a GBV_ILN_4313
912			\|a GBV_ILN_4322
912			\|a GBV_ILN_4323
912			\|a GBV_ILN_4324
912			\|a GBV_ILN_4325
912			\|a GBV_ILN_4326
912			\|a GBV_ILN_4335
912			\|a GBV_ILN_4338
912			\|a GBV_ILN_4367
912			\|a GBV_ILN_4700
951			\|a AR
952			\|d 9 \|j 2022 \|e 2 \|b 16 \|c 11 \|h 2177-2188

Indexfelder

author_variant	t a i ta tai a s d as asd d l b dl dlb a i m ai aim
matchkey_str	article:21986053:2022----::ykyripoednmcrhtcuenebigo
hierarchy_sort_str	2022
publishDate	2022
allfields	10.1007/s40747-022-00897-1 doi (DE-627)SPR050101056 (SPR)s40747-022-00897-1-e DE-627 ger DE-627 rakwb eng Ileni, Tudor Alexandru verfasserin aut DynK-hydra: improved dynamic architecture ensembling for efficient inference 2022 Text txt rdacontent Computermedien c rdamedia Online-Ressource cr rdacarrier © The Author(s) 2022 Abstract Accessibility on edge devices and the trade-off between latency and accuracy is an area of interest in deploying deep learning models. This paper explores a Mixture of Experts system, namely, DynK-Hydra, which allows training of an ensemble formed of multiple similar branches on data sets with a high number of classes, but uses, during the inference, only a subset of necessary branches. We achieve this by training a cohort of specialized branches (deep network of reduced size) and a gater/supervisor, that decides dynamically what branch to use for any specific input. An original contribution is that the number of chosen models is dynamically set, based on how confident the gater is (similar works use a static parameter for this). Another contribution is the way we ensure the branches’ specialization. We divide the data set classes into multiple clusters, and we assign a cluster to each branch while enforcing its specialization on this cluster by a separate loss function. We evaluate DynK-Hydra on CIFAR-100, Food-101, CUB-200, and ImageNet32 data sets and we obtain improvements of up to 4.3% accuracy compared with state-of-the-art ResNet. All this while reducing the number of inference flops by a factor of 2–5.5 times. Compared to a similar work (HydraRes), we obtain marginal accuracy improvements of up to 1.2% on the pairwise inference time architectures. However, we improve the inference times by up to 2.8 times compared to HydraRes. Dynamic network (dpeaa)DE-He213 Ensemble (dpeaa)DE-He213 Optimal inference (dpeaa)DE-He213 Conditional computation (dpeaa)DE-He213 Darabant, Adrian Sergiu (orcid)0000-0002-7580-5722 aut Borza, Diana Laura aut Marinescu, Alexandru Ion aut Enthalten in Complex & intelligent systems Berlin : SpringerOpen, 2015 9(2022), 2 vom: 16. Nov., Seite 2177-2188 (DE-627)835589269 (DE-600)2834740-7 2198-6053 nnns volume:9 year:2022 number:2 day:16 month:11 pages:2177-2188 https://dx.doi.org/10.1007/s40747-022-00897-1 kostenfrei Volltext GBV_USEFLAG_A SYSFLAG_A GBV_SPRINGER GBV_ILN_11 GBV_ILN_20 GBV_ILN_22 GBV_ILN_23 GBV_ILN_24 GBV_ILN_31 GBV_ILN_39 GBV_ILN_40 GBV_ILN_60 GBV_ILN_62 GBV_ILN_63 GBV_ILN_65 GBV_ILN_69 GBV_ILN_70 GBV_ILN_73 GBV_ILN_95 GBV_ILN_105 GBV_ILN_110 GBV_ILN_151 GBV_ILN_161 GBV_ILN_170 GBV_ILN_213 GBV_ILN_230 GBV_ILN_285 GBV_ILN_293 GBV_ILN_370 GBV_ILN_602 GBV_ILN_2014 GBV_ILN_4012 GBV_ILN_4037 GBV_ILN_4112 GBV_ILN_4125 GBV_ILN_4126 GBV_ILN_4249 GBV_ILN_4305 GBV_ILN_4306 GBV_ILN_4307 GBV_ILN_4313 GBV_ILN_4322 GBV_ILN_4323 GBV_ILN_4324 GBV_ILN_4325 GBV_ILN_4326 GBV_ILN_4335 GBV_ILN_4338 GBV_ILN_4367 GBV_ILN_4700 AR 9 2022 2 16 11 2177-2188
spelling	10.1007/s40747-022-00897-1 doi (DE-627)SPR050101056 (SPR)s40747-022-00897-1-e DE-627 ger DE-627 rakwb eng Ileni, Tudor Alexandru verfasserin aut DynK-hydra: improved dynamic architecture ensembling for efficient inference 2022 Text txt rdacontent Computermedien c rdamedia Online-Ressource cr rdacarrier © The Author(s) 2022 Abstract Accessibility on edge devices and the trade-off between latency and accuracy is an area of interest in deploying deep learning models. This paper explores a Mixture of Experts system, namely, DynK-Hydra, which allows training of an ensemble formed of multiple similar branches on data sets with a high number of classes, but uses, during the inference, only a subset of necessary branches. We achieve this by training a cohort of specialized branches (deep network of reduced size) and a gater/supervisor, that decides dynamically what branch to use for any specific input. An original contribution is that the number of chosen models is dynamically set, based on how confident the gater is (similar works use a static parameter for this). Another contribution is the way we ensure the branches’ specialization. We divide the data set classes into multiple clusters, and we assign a cluster to each branch while enforcing its specialization on this cluster by a separate loss function. We evaluate DynK-Hydra on CIFAR-100, Food-101, CUB-200, and ImageNet32 data sets and we obtain improvements of up to 4.3% accuracy compared with state-of-the-art ResNet. All this while reducing the number of inference flops by a factor of 2–5.5 times. Compared to a similar work (HydraRes), we obtain marginal accuracy improvements of up to 1.2% on the pairwise inference time architectures. However, we improve the inference times by up to 2.8 times compared to HydraRes. Dynamic network (dpeaa)DE-He213 Ensemble (dpeaa)DE-He213 Optimal inference (dpeaa)DE-He213 Conditional computation (dpeaa)DE-He213 Darabant, Adrian Sergiu (orcid)0000-0002-7580-5722 aut Borza, Diana Laura aut Marinescu, Alexandru Ion aut Enthalten in Complex & intelligent systems Berlin : SpringerOpen, 2015 9(2022), 2 vom: 16. Nov., Seite 2177-2188 (DE-627)835589269 (DE-600)2834740-7 2198-6053 nnns volume:9 year:2022 number:2 day:16 month:11 pages:2177-2188 https://dx.doi.org/10.1007/s40747-022-00897-1 kostenfrei Volltext GBV_USEFLAG_A SYSFLAG_A GBV_SPRINGER GBV_ILN_11 GBV_ILN_20 GBV_ILN_22 GBV_ILN_23 GBV_ILN_24 GBV_ILN_31 GBV_ILN_39 GBV_ILN_40 GBV_ILN_60 GBV_ILN_62 GBV_ILN_63 GBV_ILN_65 GBV_ILN_69 GBV_ILN_70 GBV_ILN_73 GBV_ILN_95 GBV_ILN_105 GBV_ILN_110 GBV_ILN_151 GBV_ILN_161 GBV_ILN_170 GBV_ILN_213 GBV_ILN_230 GBV_ILN_285 GBV_ILN_293 GBV_ILN_370 GBV_ILN_602 GBV_ILN_2014 GBV_ILN_4012 GBV_ILN_4037 GBV_ILN_4112 GBV_ILN_4125 GBV_ILN_4126 GBV_ILN_4249 GBV_ILN_4305 GBV_ILN_4306 GBV_ILN_4307 GBV_ILN_4313 GBV_ILN_4322 GBV_ILN_4323 GBV_ILN_4324 GBV_ILN_4325 GBV_ILN_4326 GBV_ILN_4335 GBV_ILN_4338 GBV_ILN_4367 GBV_ILN_4700 AR 9 2022 2 16 11 2177-2188
allfields_unstemmed	10.1007/s40747-022-00897-1 doi (DE-627)SPR050101056 (SPR)s40747-022-00897-1-e DE-627 ger DE-627 rakwb eng Ileni, Tudor Alexandru verfasserin aut DynK-hydra: improved dynamic architecture ensembling for efficient inference 2022 Text txt rdacontent Computermedien c rdamedia Online-Ressource cr rdacarrier © The Author(s) 2022 Abstract Accessibility on edge devices and the trade-off between latency and accuracy is an area of interest in deploying deep learning models. This paper explores a Mixture of Experts system, namely, DynK-Hydra, which allows training of an ensemble formed of multiple similar branches on data sets with a high number of classes, but uses, during the inference, only a subset of necessary branches. We achieve this by training a cohort of specialized branches (deep network of reduced size) and a gater/supervisor, that decides dynamically what branch to use for any specific input. An original contribution is that the number of chosen models is dynamically set, based on how confident the gater is (similar works use a static parameter for this). Another contribution is the way we ensure the branches’ specialization. We divide the data set classes into multiple clusters, and we assign a cluster to each branch while enforcing its specialization on this cluster by a separate loss function. We evaluate DynK-Hydra on CIFAR-100, Food-101, CUB-200, and ImageNet32 data sets and we obtain improvements of up to 4.3% accuracy compared with state-of-the-art ResNet. All this while reducing the number of inference flops by a factor of 2–5.5 times. Compared to a similar work (HydraRes), we obtain marginal accuracy improvements of up to 1.2% on the pairwise inference time architectures. However, we improve the inference times by up to 2.8 times compared to HydraRes. Dynamic network (dpeaa)DE-He213 Ensemble (dpeaa)DE-He213 Optimal inference (dpeaa)DE-He213 Conditional computation (dpeaa)DE-He213 Darabant, Adrian Sergiu (orcid)0000-0002-7580-5722 aut Borza, Diana Laura aut Marinescu, Alexandru Ion aut Enthalten in Complex & intelligent systems Berlin : SpringerOpen, 2015 9(2022), 2 vom: 16. Nov., Seite 2177-2188 (DE-627)835589269 (DE-600)2834740-7 2198-6053 nnns volume:9 year:2022 number:2 day:16 month:11 pages:2177-2188 https://dx.doi.org/10.1007/s40747-022-00897-1 kostenfrei Volltext GBV_USEFLAG_A SYSFLAG_A GBV_SPRINGER GBV_ILN_11 GBV_ILN_20 GBV_ILN_22 GBV_ILN_23 GBV_ILN_24 GBV_ILN_31 GBV_ILN_39 GBV_ILN_40 GBV_ILN_60 GBV_ILN_62 GBV_ILN_63 GBV_ILN_65 GBV_ILN_69 GBV_ILN_70 GBV_ILN_73 GBV_ILN_95 GBV_ILN_105 GBV_ILN_110 GBV_ILN_151 GBV_ILN_161 GBV_ILN_170 GBV_ILN_213 GBV_ILN_230 GBV_ILN_285 GBV_ILN_293 GBV_ILN_370 GBV_ILN_602 GBV_ILN_2014 GBV_ILN_4012 GBV_ILN_4037 GBV_ILN_4112 GBV_ILN_4125 GBV_ILN_4126 GBV_ILN_4249 GBV_ILN_4305 GBV_ILN_4306 GBV_ILN_4307 GBV_ILN_4313 GBV_ILN_4322 GBV_ILN_4323 GBV_ILN_4324 GBV_ILN_4325 GBV_ILN_4326 GBV_ILN_4335 GBV_ILN_4338 GBV_ILN_4367 GBV_ILN_4700 AR 9 2022 2 16 11 2177-2188
allfieldsGer	10.1007/s40747-022-00897-1 doi (DE-627)SPR050101056 (SPR)s40747-022-00897-1-e DE-627 ger DE-627 rakwb eng Ileni, Tudor Alexandru verfasserin aut DynK-hydra: improved dynamic architecture ensembling for efficient inference 2022 Text txt rdacontent Computermedien c rdamedia Online-Ressource cr rdacarrier © The Author(s) 2022 Abstract Accessibility on edge devices and the trade-off between latency and accuracy is an area of interest in deploying deep learning models. This paper explores a Mixture of Experts system, namely, DynK-Hydra, which allows training of an ensemble formed of multiple similar branches on data sets with a high number of classes, but uses, during the inference, only a subset of necessary branches. We achieve this by training a cohort of specialized branches (deep network of reduced size) and a gater/supervisor, that decides dynamically what branch to use for any specific input. An original contribution is that the number of chosen models is dynamically set, based on how confident the gater is (similar works use a static parameter for this). Another contribution is the way we ensure the branches’ specialization. We divide the data set classes into multiple clusters, and we assign a cluster to each branch while enforcing its specialization on this cluster by a separate loss function. We evaluate DynK-Hydra on CIFAR-100, Food-101, CUB-200, and ImageNet32 data sets and we obtain improvements of up to 4.3% accuracy compared with state-of-the-art ResNet. All this while reducing the number of inference flops by a factor of 2–5.5 times. Compared to a similar work (HydraRes), we obtain marginal accuracy improvements of up to 1.2% on the pairwise inference time architectures. However, we improve the inference times by up to 2.8 times compared to HydraRes. Dynamic network (dpeaa)DE-He213 Ensemble (dpeaa)DE-He213 Optimal inference (dpeaa)DE-He213 Conditional computation (dpeaa)DE-He213 Darabant, Adrian Sergiu (orcid)0000-0002-7580-5722 aut Borza, Diana Laura aut Marinescu, Alexandru Ion aut Enthalten in Complex & intelligent systems Berlin : SpringerOpen, 2015 9(2022), 2 vom: 16. Nov., Seite 2177-2188 (DE-627)835589269 (DE-600)2834740-7 2198-6053 nnns volume:9 year:2022 number:2 day:16 month:11 pages:2177-2188 https://dx.doi.org/10.1007/s40747-022-00897-1 kostenfrei Volltext GBV_USEFLAG_A SYSFLAG_A GBV_SPRINGER GBV_ILN_11 GBV_ILN_20 GBV_ILN_22 GBV_ILN_23 GBV_ILN_24 GBV_ILN_31 GBV_ILN_39 GBV_ILN_40 GBV_ILN_60 GBV_ILN_62 GBV_ILN_63 GBV_ILN_65 GBV_ILN_69 GBV_ILN_70 GBV_ILN_73 GBV_ILN_95 GBV_ILN_105 GBV_ILN_110 GBV_ILN_151 GBV_ILN_161 GBV_ILN_170 GBV_ILN_213 GBV_ILN_230 GBV_ILN_285 GBV_ILN_293 GBV_ILN_370 GBV_ILN_602 GBV_ILN_2014 GBV_ILN_4012 GBV_ILN_4037 GBV_ILN_4112 GBV_ILN_4125 GBV_ILN_4126 GBV_ILN_4249 GBV_ILN_4305 GBV_ILN_4306 GBV_ILN_4307 GBV_ILN_4313 GBV_ILN_4322 GBV_ILN_4323 GBV_ILN_4324 GBV_ILN_4325 GBV_ILN_4326 GBV_ILN_4335 GBV_ILN_4338 GBV_ILN_4367 GBV_ILN_4700 AR 9 2022 2 16 11 2177-2188
allfieldsSound	10.1007/s40747-022-00897-1 doi (DE-627)SPR050101056 (SPR)s40747-022-00897-1-e DE-627 ger DE-627 rakwb eng Ileni, Tudor Alexandru verfasserin aut DynK-hydra: improved dynamic architecture ensembling for efficient inference 2022 Text txt rdacontent Computermedien c rdamedia Online-Ressource cr rdacarrier © The Author(s) 2022 Abstract Accessibility on edge devices and the trade-off between latency and accuracy is an area of interest in deploying deep learning models. This paper explores a Mixture of Experts system, namely, DynK-Hydra, which allows training of an ensemble formed of multiple similar branches on data sets with a high number of classes, but uses, during the inference, only a subset of necessary branches. We achieve this by training a cohort of specialized branches (deep network of reduced size) and a gater/supervisor, that decides dynamically what branch to use for any specific input. An original contribution is that the number of chosen models is dynamically set, based on how confident the gater is (similar works use a static parameter for this). Another contribution is the way we ensure the branches’ specialization. We divide the data set classes into multiple clusters, and we assign a cluster to each branch while enforcing its specialization on this cluster by a separate loss function. We evaluate DynK-Hydra on CIFAR-100, Food-101, CUB-200, and ImageNet32 data sets and we obtain improvements of up to 4.3% accuracy compared with state-of-the-art ResNet. All this while reducing the number of inference flops by a factor of 2–5.5 times. Compared to a similar work (HydraRes), we obtain marginal accuracy improvements of up to 1.2% on the pairwise inference time architectures. However, we improve the inference times by up to 2.8 times compared to HydraRes. Dynamic network (dpeaa)DE-He213 Ensemble (dpeaa)DE-He213 Optimal inference (dpeaa)DE-He213 Conditional computation (dpeaa)DE-He213 Darabant, Adrian Sergiu (orcid)0000-0002-7580-5722 aut Borza, Diana Laura aut Marinescu, Alexandru Ion aut Enthalten in Complex & intelligent systems Berlin : SpringerOpen, 2015 9(2022), 2 vom: 16. Nov., Seite 2177-2188 (DE-627)835589269 (DE-600)2834740-7 2198-6053 nnns volume:9 year:2022 number:2 day:16 month:11 pages:2177-2188 https://dx.doi.org/10.1007/s40747-022-00897-1 kostenfrei Volltext GBV_USEFLAG_A SYSFLAG_A GBV_SPRINGER GBV_ILN_11 GBV_ILN_20 GBV_ILN_22 GBV_ILN_23 GBV_ILN_24 GBV_ILN_31 GBV_ILN_39 GBV_ILN_40 GBV_ILN_60 GBV_ILN_62 GBV_ILN_63 GBV_ILN_65 GBV_ILN_69 GBV_ILN_70 GBV_ILN_73 GBV_ILN_95 GBV_ILN_105 GBV_ILN_110 GBV_ILN_151 GBV_ILN_161 GBV_ILN_170 GBV_ILN_213 GBV_ILN_230 GBV_ILN_285 GBV_ILN_293 GBV_ILN_370 GBV_ILN_602 GBV_ILN_2014 GBV_ILN_4012 GBV_ILN_4037 GBV_ILN_4112 GBV_ILN_4125 GBV_ILN_4126 GBV_ILN_4249 GBV_ILN_4305 GBV_ILN_4306 GBV_ILN_4307 GBV_ILN_4313 GBV_ILN_4322 GBV_ILN_4323 GBV_ILN_4324 GBV_ILN_4325 GBV_ILN_4326 GBV_ILN_4335 GBV_ILN_4338 GBV_ILN_4367 GBV_ILN_4700 AR 9 2022 2 16 11 2177-2188
language	English
source	Enthalten in Complex & intelligent systems 9(2022), 2 vom: 16. Nov., Seite 2177-2188 volume:9 year:2022 number:2 day:16 month:11 pages:2177-2188
sourceStr	Enthalten in Complex & intelligent systems 9(2022), 2 vom: 16. Nov., Seite 2177-2188 volume:9 year:2022 number:2 day:16 month:11 pages:2177-2188
format_phy_str_mv	Article
institution	findex.gbv.de
topic_facet	Dynamic network Ensemble Optimal inference Conditional computation
isfreeaccess_bool	true
container_title	Complex & intelligent systems
authorswithroles_txt_mv	Ileni, Tudor Alexandru @@aut@@ Darabant, Adrian Sergiu @@aut@@ Borza, Diana Laura @@aut@@ Marinescu, Alexandru Ion @@aut@@
publishDateDaySort_date	2022-11-16T00:00:00Z
hierarchy_top_id	835589269
id	SPR050101056
language_de	englisch
fullrecord	<?xml version="1.0" encoding="UTF-8"?><collection xmlns="http://www.loc.gov/MARC21/slim"><record><leader>01000naa a22002652 4500</leader><controlfield tag="001">SPR050101056</controlfield><controlfield tag="003">DE-627</controlfield><controlfield tag="005">20230419064805.0</controlfield><controlfield tag="007">cr uuu---uuuuu</controlfield><controlfield tag="008">230419s2022 xx \|\|\|\|\|o 00\| \|\|eng c</controlfield><datafield tag="024" ind1="7" ind2=" "><subfield code="a">10.1007/s40747-022-00897-1</subfield><subfield code="2">doi</subfield></datafield><datafield tag="035" ind1=" " ind2=" "><subfield code="a">(DE-627)SPR050101056</subfield></datafield><datafield tag="035" ind1=" " ind2=" "><subfield code="a">(SPR)s40747-022-00897-1-e</subfield></datafield><datafield tag="040" ind1=" " ind2=" "><subfield code="a">DE-627</subfield><subfield code="b">ger</subfield><subfield code="c">DE-627</subfield><subfield code="e">rakwb</subfield></datafield><datafield tag="041" ind1=" " ind2=" "><subfield code="a">eng</subfield></datafield><datafield tag="100" ind1="1" ind2=" "><subfield code="a">Ileni, Tudor Alexandru</subfield><subfield code="e">verfasserin</subfield><subfield code="4">aut</subfield></datafield><datafield tag="245" ind1="1" ind2="0"><subfield code="a">DynK-hydra: improved dynamic architecture ensembling for efficient inference</subfield></datafield><datafield tag="264" ind1=" " ind2="1"><subfield code="c">2022</subfield></datafield><datafield tag="336" ind1=" " ind2=" "><subfield code="a">Text</subfield><subfield code="b">txt</subfield><subfield code="2">rdacontent</subfield></datafield><datafield tag="337" ind1=" " ind2=" "><subfield code="a">Computermedien</subfield><subfield code="b">c</subfield><subfield code="2">rdamedia</subfield></datafield><datafield tag="338" ind1=" " ind2=" "><subfield code="a">Online-Ressource</subfield><subfield code="b">cr</subfield><subfield code="2">rdacarrier</subfield></datafield><datafield tag="500" ind1=" " ind2=" "><subfield code="a">© The Author(s) 2022</subfield></datafield><datafield tag="520" ind1=" " ind2=" "><subfield code="a">Abstract Accessibility on edge devices and the trade-off between latency and accuracy is an area of interest in deploying deep learning models. This paper explores a Mixture of Experts system, namely, DynK-Hydra, which allows training of an ensemble formed of multiple similar branches on data sets with a high number of classes, but uses, during the inference, only a subset of necessary branches. We achieve this by training a cohort of specialized branches (deep network of reduced size) and a gater/supervisor, that decides dynamically what branch to use for any specific input. An original contribution is that the number of chosen models is dynamically set, based on how confident the gater is (similar works use a static parameter for this). Another contribution is the way we ensure the branches’ specialization. We divide the data set classes into multiple clusters, and we assign a cluster to each branch while enforcing its specialization on this cluster by a separate loss function. We evaluate DynK-Hydra on CIFAR-100, Food-101, CUB-200, and ImageNet32 data sets and we obtain improvements of up to 4.3% accuracy compared with state-of-the-art ResNet. All this while reducing the number of inference flops by a factor of 2–5.5 times. Compared to a similar work (HydraRes), we obtain marginal accuracy improvements of up to 1.2% on the pairwise inference time architectures. However, we improve the inference times by up to 2.8 times compared to HydraRes.</subfield></datafield><datafield tag="650" ind1=" " ind2="4"><subfield code="a">Dynamic network</subfield><subfield code="7">(dpeaa)DE-He213</subfield></datafield><datafield tag="650" ind1=" " ind2="4"><subfield code="a">Ensemble</subfield><subfield code="7">(dpeaa)DE-He213</subfield></datafield><datafield tag="650" ind1=" " ind2="4"><subfield code="a">Optimal inference</subfield><subfield code="7">(dpeaa)DE-He213</subfield></datafield><datafield tag="650" ind1=" " ind2="4"><subfield code="a">Conditional computation</subfield><subfield code="7">(dpeaa)DE-He213</subfield></datafield><datafield tag="700" ind1="1" ind2=" "><subfield code="a">Darabant, Adrian Sergiu</subfield><subfield code="0">(orcid)0000-0002-7580-5722</subfield><subfield code="4">aut</subfield></datafield><datafield tag="700" ind1="1" ind2=" "><subfield code="a">Borza, Diana Laura</subfield><subfield code="4">aut</subfield></datafield><datafield tag="700" ind1="1" ind2=" "><subfield code="a">Marinescu, Alexandru Ion</subfield><subfield code="4">aut</subfield></datafield><datafield tag="773" ind1="0" ind2="8"><subfield code="i">Enthalten in</subfield><subfield code="t">Complex & intelligent systems</subfield><subfield code="d">Berlin : SpringerOpen, 2015</subfield><subfield code="g">9(2022), 2 vom: 16. Nov., Seite 2177-2188</subfield><subfield code="w">(DE-627)835589269</subfield><subfield code="w">(DE-600)2834740-7</subfield><subfield code="x">2198-6053</subfield><subfield code="7">nnns</subfield></datafield><datafield tag="773" ind1="1" ind2="8"><subfield code="g">volume:9</subfield><subfield code="g">year:2022</subfield><subfield code="g">number:2</subfield><subfield code="g">day:16</subfield><subfield code="g">month:11</subfield><subfield code="g">pages:2177-2188</subfield></datafield><datafield tag="856" ind1="4" ind2="0"><subfield code="u">https://dx.doi.org/10.1007/s40747-022-00897-1</subfield><subfield code="z">kostenfrei</subfield><subfield code="3">Volltext</subfield></datafield><datafield tag="912" ind1=" " ind2=" "><subfield code="a">GBV_USEFLAG_A</subfield></datafield><datafield tag="912" ind1=" " ind2=" "><subfield code="a">SYSFLAG_A</subfield></datafield><datafield tag="912" ind1=" " ind2=" "><subfield code="a">GBV_SPRINGER</subfield></datafield><datafield tag="912" ind1=" " ind2=" "><subfield code="a">GBV_ILN_11</subfield></datafield><datafield tag="912" ind1=" " ind2=" "><subfield code="a">GBV_ILN_20</subfield></datafield><datafield tag="912" ind1=" " ind2=" "><subfield code="a">GBV_ILN_22</subfield></datafield><datafield tag="912" ind1=" " ind2=" "><subfield code="a">GBV_ILN_23</subfield></datafield><datafield tag="912" ind1=" " ind2=" "><subfield code="a">GBV_ILN_24</subfield></datafield><datafield tag="912" ind1=" " ind2=" "><subfield code="a">GBV_ILN_31</subfield></datafield><datafield tag="912" ind1=" " ind2=" "><subfield code="a">GBV_ILN_39</subfield></datafield><datafield tag="912" ind1=" " ind2=" "><subfield code="a">GBV_ILN_40</subfield></datafield><datafield tag="912" ind1=" " ind2=" "><subfield code="a">GBV_ILN_60</subfield></datafield><datafield tag="912" ind1=" " ind2=" "><subfield code="a">GBV_ILN_62</subfield></datafield><datafield tag="912" ind1=" " ind2=" "><subfield code="a">GBV_ILN_63</subfield></datafield><datafield tag="912" ind1=" " ind2=" "><subfield code="a">GBV_ILN_65</subfield></datafield><datafield tag="912" ind1=" " ind2=" "><subfield code="a">GBV_ILN_69</subfield></datafield><datafield tag="912" ind1=" " ind2=" "><subfield code="a">GBV_ILN_70</subfield></datafield><datafield tag="912" ind1=" " ind2=" "><subfield code="a">GBV_ILN_73</subfield></datafield><datafield tag="912" ind1=" " ind2=" "><subfield code="a">GBV_ILN_95</subfield></datafield><datafield tag="912" ind1=" " ind2=" "><subfield code="a">GBV_ILN_105</subfield></datafield><datafield tag="912" ind1=" " ind2=" "><subfield code="a">GBV_ILN_110</subfield></datafield><datafield tag="912" ind1=" " ind2=" "><subfield code="a">GBV_ILN_151</subfield></datafield><datafield tag="912" ind1=" " ind2=" "><subfield code="a">GBV_ILN_161</subfield></datafield><datafield tag="912" ind1=" " ind2=" "><subfield code="a">GBV_ILN_170</subfield></datafield><datafield tag="912" ind1=" " ind2=" "><subfield code="a">GBV_ILN_213</subfield></datafield><datafield tag="912" ind1=" " ind2=" "><subfield code="a">GBV_ILN_230</subfield></datafield><datafield tag="912" ind1=" " ind2=" "><subfield code="a">GBV_ILN_285</subfield></datafield><datafield tag="912" ind1=" " ind2=" "><subfield code="a">GBV_ILN_293</subfield></datafield><datafield tag="912" ind1=" " ind2=" "><subfield code="a">GBV_ILN_370</subfield></datafield><datafield tag="912" ind1=" " ind2=" "><subfield code="a">GBV_ILN_602</subfield></datafield><datafield tag="912" ind1=" " ind2=" "><subfield code="a">GBV_ILN_2014</subfield></datafield><datafield tag="912" ind1=" " ind2=" "><subfield code="a">GBV_ILN_4012</subfield></datafield><datafield tag="912" ind1=" " ind2=" "><subfield code="a">GBV_ILN_4037</subfield></datafield><datafield tag="912" ind1=" " ind2=" "><subfield code="a">GBV_ILN_4112</subfield></datafield><datafield tag="912" ind1=" " ind2=" "><subfield code="a">GBV_ILN_4125</subfield></datafield><datafield tag="912" ind1=" " ind2=" "><subfield code="a">GBV_ILN_4126</subfield></datafield><datafield tag="912" ind1=" " ind2=" "><subfield code="a">GBV_ILN_4249</subfield></datafield><datafield tag="912" ind1=" " ind2=" "><subfield code="a">GBV_ILN_4305</subfield></datafield><datafield tag="912" ind1=" " ind2=" "><subfield code="a">GBV_ILN_4306</subfield></datafield><datafield tag="912" ind1=" " ind2=" "><subfield code="a">GBV_ILN_4307</subfield></datafield><datafield tag="912" ind1=" " ind2=" "><subfield code="a">GBV_ILN_4313</subfield></datafield><datafield tag="912" ind1=" " ind2=" "><subfield code="a">GBV_ILN_4322</subfield></datafield><datafield tag="912" ind1=" " ind2=" "><subfield code="a">GBV_ILN_4323</subfield></datafield><datafield tag="912" ind1=" " ind2=" "><subfield code="a">GBV_ILN_4324</subfield></datafield><datafield tag="912" ind1=" " ind2=" "><subfield code="a">GBV_ILN_4325</subfield></datafield><datafield tag="912" ind1=" " ind2=" "><subfield code="a">GBV_ILN_4326</subfield></datafield><datafield tag="912" ind1=" " ind2=" "><subfield code="a">GBV_ILN_4335</subfield></datafield><datafield tag="912" ind1=" " ind2=" "><subfield code="a">GBV_ILN_4338</subfield></datafield><datafield tag="912" ind1=" " ind2=" "><subfield code="a">GBV_ILN_4367</subfield></datafield><datafield tag="912" ind1=" " ind2=" "><subfield code="a">GBV_ILN_4700</subfield></datafield><datafield tag="951" ind1=" " ind2=" "><subfield code="a">AR</subfield></datafield><datafield tag="952" ind1=" " ind2=" "><subfield code="d">9</subfield><subfield code="j">2022</subfield><subfield code="e">2</subfield><subfield code="b">16</subfield><subfield code="c">11</subfield><subfield code="h">2177-2188</subfield></datafield></record></collection>
author	Ileni, Tudor Alexandru
spellingShingle	Ileni, Tudor Alexandru misc Dynamic network misc Ensemble misc Optimal inference misc Conditional computation DynK-hydra: improved dynamic architecture ensembling for efficient inference
authorStr	Ileni, Tudor Alexandru
ppnlink_with_tag_str_mv	@@773@@(DE-627)835589269
format	electronic Article
delete_txt_mv	keep
author_role	aut aut aut aut
collection	springer
remote_str	true
illustrated	Not Illustrated
issn	2198-6053
topic_title	DynK-hydra: improved dynamic architecture ensembling for efficient inference Dynamic network (dpeaa)DE-He213 Ensemble (dpeaa)DE-He213 Optimal inference (dpeaa)DE-He213 Conditional computation (dpeaa)DE-He213
topic	misc Dynamic network misc Ensemble misc Optimal inference misc Conditional computation
topic_unstemmed	misc Dynamic network misc Ensemble misc Optimal inference misc Conditional computation
topic_browse	misc Dynamic network misc Ensemble misc Optimal inference misc Conditional computation
format_facet	Elektronische Aufsätze Aufsätze Elektronische Ressource
format_main_str_mv	Text Zeitschrift/Artikel
carriertype_str_mv	cr
hierarchy_parent_title	Complex & intelligent systems
hierarchy_parent_id	835589269
hierarchy_top_title	Complex & intelligent systems
isfreeaccess_txt	true
familylinks_str_mv	(DE-627)835589269 (DE-600)2834740-7
title	DynK-hydra: improved dynamic architecture ensembling for efficient inference
ctrlnum	(DE-627)SPR050101056 (SPR)s40747-022-00897-1-e
title_full	DynK-hydra: improved dynamic architecture ensembling for efficient inference
author_sort	Ileni, Tudor Alexandru
journal	Complex & intelligent systems
journalStr	Complex & intelligent systems
lang_code	eng
isOA_bool	true
recordtype	marc
publishDateSort	2022
contenttype_str_mv	txt
container_start_page	2177
author_browse	Ileni, Tudor Alexandru Darabant, Adrian Sergiu Borza, Diana Laura Marinescu, Alexandru Ion
container_volume	9
format_se	Elektronische Aufsätze
author-letter	Ileni, Tudor Alexandru
doi_str_mv	10.1007/s40747-022-00897-1
normlink	(ORCID)0000-0002-7580-5722
normlink_prefix_str_mv	(orcid)0000-0002-7580-5722
title_sort	dynk-hydra: improved dynamic architecture ensembling for efficient inference
title_auth	DynK-hydra: improved dynamic architecture ensembling for efficient inference
abstract	Abstract Accessibility on edge devices and the trade-off between latency and accuracy is an area of interest in deploying deep learning models. This paper explores a Mixture of Experts system, namely, DynK-Hydra, which allows training of an ensemble formed of multiple similar branches on data sets with a high number of classes, but uses, during the inference, only a subset of necessary branches. We achieve this by training a cohort of specialized branches (deep network of reduced size) and a gater/supervisor, that decides dynamically what branch to use for any specific input. An original contribution is that the number of chosen models is dynamically set, based on how confident the gater is (similar works use a static parameter for this). Another contribution is the way we ensure the branches’ specialization. We divide the data set classes into multiple clusters, and we assign a cluster to each branch while enforcing its specialization on this cluster by a separate loss function. We evaluate DynK-Hydra on CIFAR-100, Food-101, CUB-200, and ImageNet32 data sets and we obtain improvements of up to 4.3% accuracy compared with state-of-the-art ResNet. All this while reducing the number of inference flops by a factor of 2–5.5 times. Compared to a similar work (HydraRes), we obtain marginal accuracy improvements of up to 1.2% on the pairwise inference time architectures. However, we improve the inference times by up to 2.8 times compared to HydraRes. © The Author(s) 2022
abstractGer	Abstract Accessibility on edge devices and the trade-off between latency and accuracy is an area of interest in deploying deep learning models. This paper explores a Mixture of Experts system, namely, DynK-Hydra, which allows training of an ensemble formed of multiple similar branches on data sets with a high number of classes, but uses, during the inference, only a subset of necessary branches. We achieve this by training a cohort of specialized branches (deep network of reduced size) and a gater/supervisor, that decides dynamically what branch to use for any specific input. An original contribution is that the number of chosen models is dynamically set, based on how confident the gater is (similar works use a static parameter for this). Another contribution is the way we ensure the branches’ specialization. We divide the data set classes into multiple clusters, and we assign a cluster to each branch while enforcing its specialization on this cluster by a separate loss function. We evaluate DynK-Hydra on CIFAR-100, Food-101, CUB-200, and ImageNet32 data sets and we obtain improvements of up to 4.3% accuracy compared with state-of-the-art ResNet. All this while reducing the number of inference flops by a factor of 2–5.5 times. Compared to a similar work (HydraRes), we obtain marginal accuracy improvements of up to 1.2% on the pairwise inference time architectures. However, we improve the inference times by up to 2.8 times compared to HydraRes. © The Author(s) 2022
abstract_unstemmed	Abstract Accessibility on edge devices and the trade-off between latency and accuracy is an area of interest in deploying deep learning models. This paper explores a Mixture of Experts system, namely, DynK-Hydra, which allows training of an ensemble formed of multiple similar branches on data sets with a high number of classes, but uses, during the inference, only a subset of necessary branches. We achieve this by training a cohort of specialized branches (deep network of reduced size) and a gater/supervisor, that decides dynamically what branch to use for any specific input. An original contribution is that the number of chosen models is dynamically set, based on how confident the gater is (similar works use a static parameter for this). Another contribution is the way we ensure the branches’ specialization. We divide the data set classes into multiple clusters, and we assign a cluster to each branch while enforcing its specialization on this cluster by a separate loss function. We evaluate DynK-Hydra on CIFAR-100, Food-101, CUB-200, and ImageNet32 data sets and we obtain improvements of up to 4.3% accuracy compared with state-of-the-art ResNet. All this while reducing the number of inference flops by a factor of 2–5.5 times. Compared to a similar work (HydraRes), we obtain marginal accuracy improvements of up to 1.2% on the pairwise inference time architectures. However, we improve the inference times by up to 2.8 times compared to HydraRes. © The Author(s) 2022
collection_details	GBV_USEFLAG_A SYSFLAG_A GBV_SPRINGER GBV_ILN_11 GBV_ILN_20 GBV_ILN_22 GBV_ILN_23 GBV_ILN_24 GBV_ILN_31 GBV_ILN_39 GBV_ILN_40 GBV_ILN_60 GBV_ILN_62 GBV_ILN_63 GBV_ILN_65 GBV_ILN_69 GBV_ILN_70 GBV_ILN_73 GBV_ILN_95 GBV_ILN_105 GBV_ILN_110 GBV_ILN_151 GBV_ILN_161 GBV_ILN_170 GBV_ILN_213 GBV_ILN_230 GBV_ILN_285 GBV_ILN_293 GBV_ILN_370 GBV_ILN_602 GBV_ILN_2014 GBV_ILN_4012 GBV_ILN_4037 GBV_ILN_4112 GBV_ILN_4125 GBV_ILN_4126 GBV_ILN_4249 GBV_ILN_4305 GBV_ILN_4306 GBV_ILN_4307 GBV_ILN_4313 GBV_ILN_4322 GBV_ILN_4323 GBV_ILN_4324 GBV_ILN_4325 GBV_ILN_4326 GBV_ILN_4335 GBV_ILN_4338 GBV_ILN_4367 GBV_ILN_4700
container_issue	2
title_short	DynK-hydra: improved dynamic architecture ensembling for efficient inference
url	https://dx.doi.org/10.1007/s40747-022-00897-1
remote_bool	true
author2	Darabant, Adrian Sergiu Borza, Diana Laura Marinescu, Alexandru Ion
author2Str	Darabant, Adrian Sergiu Borza, Diana Laura Marinescu, Alexandru Ion
ppnlink	835589269
mediatype_str_mv	c
isOA_txt	true
hochschulschrift_bool	false
doi_str	10.1007/s40747-022-00897-1
up_date	2024-07-03T13:22:59.205Z
_version_	1803564329689153536
fullrecord_marcxml	<?xml version="1.0" encoding="UTF-8"?><collection xmlns="http://www.loc.gov/MARC21/slim"><record><leader>01000naa a22002652 4500</leader><controlfield tag="001">SPR050101056</controlfield><controlfield tag="003">DE-627</controlfield><controlfield tag="005">20230419064805.0</controlfield><controlfield tag="007">cr uuu---uuuuu</controlfield><controlfield tag="008">230419s2022 xx \|\|\|\|\|o 00\| \|\|eng c</controlfield><datafield tag="024" ind1="7" ind2=" "><subfield code="a">10.1007/s40747-022-00897-1</subfield><subfield code="2">doi</subfield></datafield><datafield tag="035" ind1=" " ind2=" "><subfield code="a">(DE-627)SPR050101056</subfield></datafield><datafield tag="035" ind1=" " ind2=" "><subfield code="a">(SPR)s40747-022-00897-1-e</subfield></datafield><datafield tag="040" ind1=" " ind2=" "><subfield code="a">DE-627</subfield><subfield code="b">ger</subfield><subfield code="c">DE-627</subfield><subfield code="e">rakwb</subfield></datafield><datafield tag="041" ind1=" " ind2=" "><subfield code="a">eng</subfield></datafield><datafield tag="100" ind1="1" ind2=" "><subfield code="a">Ileni, Tudor Alexandru</subfield><subfield code="e">verfasserin</subfield><subfield code="4">aut</subfield></datafield><datafield tag="245" ind1="1" ind2="0"><subfield code="a">DynK-hydra: improved dynamic architecture ensembling for efficient inference</subfield></datafield><datafield tag="264" ind1=" " ind2="1"><subfield code="c">2022</subfield></datafield><datafield tag="336" ind1=" " ind2=" "><subfield code="a">Text</subfield><subfield code="b">txt</subfield><subfield code="2">rdacontent</subfield></datafield><datafield tag="337" ind1=" " ind2=" "><subfield code="a">Computermedien</subfield><subfield code="b">c</subfield><subfield code="2">rdamedia</subfield></datafield><datafield tag="338" ind1=" " ind2=" "><subfield code="a">Online-Ressource</subfield><subfield code="b">cr</subfield><subfield code="2">rdacarrier</subfield></datafield><datafield tag="500" ind1=" " ind2=" "><subfield code="a">© The Author(s) 2022</subfield></datafield><datafield tag="520" ind1=" " ind2=" "><subfield code="a">Abstract Accessibility on edge devices and the trade-off between latency and accuracy is an area of interest in deploying deep learning models. This paper explores a Mixture of Experts system, namely, DynK-Hydra, which allows training of an ensemble formed of multiple similar branches on data sets with a high number of classes, but uses, during the inference, only a subset of necessary branches. We achieve this by training a cohort of specialized branches (deep network of reduced size) and a gater/supervisor, that decides dynamically what branch to use for any specific input. An original contribution is that the number of chosen models is dynamically set, based on how confident the gater is (similar works use a static parameter for this). Another contribution is the way we ensure the branches’ specialization. We divide the data set classes into multiple clusters, and we assign a cluster to each branch while enforcing its specialization on this cluster by a separate loss function. We evaluate DynK-Hydra on CIFAR-100, Food-101, CUB-200, and ImageNet32 data sets and we obtain improvements of up to 4.3% accuracy compared with state-of-the-art ResNet. All this while reducing the number of inference flops by a factor of 2–5.5 times. Compared to a similar work (HydraRes), we obtain marginal accuracy improvements of up to 1.2% on the pairwise inference time architectures. However, we improve the inference times by up to 2.8 times compared to HydraRes.</subfield></datafield><datafield tag="650" ind1=" " ind2="4"><subfield code="a">Dynamic network</subfield><subfield code="7">(dpeaa)DE-He213</subfield></datafield><datafield tag="650" ind1=" " ind2="4"><subfield code="a">Ensemble</subfield><subfield code="7">(dpeaa)DE-He213</subfield></datafield><datafield tag="650" ind1=" " ind2="4"><subfield code="a">Optimal inference</subfield><subfield code="7">(dpeaa)DE-He213</subfield></datafield><datafield tag="650" ind1=" " ind2="4"><subfield code="a">Conditional computation</subfield><subfield code="7">(dpeaa)DE-He213</subfield></datafield><datafield tag="700" ind1="1" ind2=" "><subfield code="a">Darabant, Adrian Sergiu</subfield><subfield code="0">(orcid)0000-0002-7580-5722</subfield><subfield code="4">aut</subfield></datafield><datafield tag="700" ind1="1" ind2=" "><subfield code="a">Borza, Diana Laura</subfield><subfield code="4">aut</subfield></datafield><datafield tag="700" ind1="1" ind2=" "><subfield code="a">Marinescu, Alexandru Ion</subfield><subfield code="4">aut</subfield></datafield><datafield tag="773" ind1="0" ind2="8"><subfield code="i">Enthalten in</subfield><subfield code="t">Complex & intelligent systems</subfield><subfield code="d">Berlin : SpringerOpen, 2015</subfield><subfield code="g">9(2022), 2 vom: 16. Nov., Seite 2177-2188</subfield><subfield code="w">(DE-627)835589269</subfield><subfield code="w">(DE-600)2834740-7</subfield><subfield code="x">2198-6053</subfield><subfield code="7">nnns</subfield></datafield><datafield tag="773" ind1="1" ind2="8"><subfield code="g">volume:9</subfield><subfield code="g">year:2022</subfield><subfield code="g">number:2</subfield><subfield code="g">day:16</subfield><subfield code="g">month:11</subfield><subfield code="g">pages:2177-2188</subfield></datafield><datafield tag="856" ind1="4" ind2="0"><subfield code="u">https://dx.doi.org/10.1007/s40747-022-00897-1</subfield><subfield code="z">kostenfrei</subfield><subfield code="3">Volltext</subfield></datafield><datafield tag="912" ind1=" " ind2=" "><subfield code="a">GBV_USEFLAG_A</subfield></datafield><datafield tag="912" ind1=" " ind2=" "><subfield code="a">SYSFLAG_A</subfield></datafield><datafield tag="912" ind1=" " ind2=" "><subfield code="a">GBV_SPRINGER</subfield></datafield><datafield tag="912" ind1=" " ind2=" "><subfield code="a">GBV_ILN_11</subfield></datafield><datafield tag="912" ind1=" " ind2=" "><subfield code="a">GBV_ILN_20</subfield></datafield><datafield tag="912" ind1=" " ind2=" "><subfield code="a">GBV_ILN_22</subfield></datafield><datafield tag="912" ind1=" " ind2=" "><subfield code="a">GBV_ILN_23</subfield></datafield><datafield tag="912" ind1=" " ind2=" "><subfield code="a">GBV_ILN_24</subfield></datafield><datafield tag="912" ind1=" " ind2=" "><subfield code="a">GBV_ILN_31</subfield></datafield><datafield tag="912" ind1=" " ind2=" "><subfield code="a">GBV_ILN_39</subfield></datafield><datafield tag="912" ind1=" " ind2=" "><subfield code="a">GBV_ILN_40</subfield></datafield><datafield tag="912" ind1=" " ind2=" "><subfield code="a">GBV_ILN_60</subfield></datafield><datafield tag="912" ind1=" " ind2=" "><subfield code="a">GBV_ILN_62</subfield></datafield><datafield tag="912" ind1=" " ind2=" "><subfield code="a">GBV_ILN_63</subfield></datafield><datafield tag="912" ind1=" " ind2=" "><subfield code="a">GBV_ILN_65</subfield></datafield><datafield tag="912" ind1=" " ind2=" "><subfield code="a">GBV_ILN_69</subfield></datafield><datafield tag="912" ind1=" " ind2=" "><subfield code="a">GBV_ILN_70</subfield></datafield><datafield tag="912" ind1=" " ind2=" "><subfield code="a">GBV_ILN_73</subfield></datafield><datafield tag="912" ind1=" " ind2=" "><subfield code="a">GBV_ILN_95</subfield></datafield><datafield tag="912" ind1=" " ind2=" "><subfield code="a">GBV_ILN_105</subfield></datafield><datafield tag="912" ind1=" " ind2=" "><subfield code="a">GBV_ILN_110</subfield></datafield><datafield tag="912" ind1=" " ind2=" "><subfield code="a">GBV_ILN_151</subfield></datafield><datafield tag="912" ind1=" " ind2=" "><subfield code="a">GBV_ILN_161</subfield></datafield><datafield tag="912" ind1=" " ind2=" "><subfield code="a">GBV_ILN_170</subfield></datafield><datafield tag="912" ind1=" " ind2=" "><subfield code="a">GBV_ILN_213</subfield></datafield><datafield tag="912" ind1=" " ind2=" "><subfield code="a">GBV_ILN_230</subfield></datafield><datafield tag="912" ind1=" " ind2=" "><subfield code="a">GBV_ILN_285</subfield></datafield><datafield tag="912" ind1=" " ind2=" "><subfield code="a">GBV_ILN_293</subfield></datafield><datafield tag="912" ind1=" " ind2=" "><subfield code="a">GBV_ILN_370</subfield></datafield><datafield tag="912" ind1=" " ind2=" "><subfield code="a">GBV_ILN_602</subfield></datafield><datafield tag="912" ind1=" " ind2=" "><subfield code="a">GBV_ILN_2014</subfield></datafield><datafield tag="912" ind1=" " ind2=" "><subfield code="a">GBV_ILN_4012</subfield></datafield><datafield tag="912" ind1=" " ind2=" "><subfield code="a">GBV_ILN_4037</subfield></datafield><datafield tag="912" ind1=" " ind2=" "><subfield code="a">GBV_ILN_4112</subfield></datafield><datafield tag="912" ind1=" " ind2=" "><subfield code="a">GBV_ILN_4125</subfield></datafield><datafield tag="912" ind1=" " ind2=" "><subfield code="a">GBV_ILN_4126</subfield></datafield><datafield tag="912" ind1=" " ind2=" "><subfield code="a">GBV_ILN_4249</subfield></datafield><datafield tag="912" ind1=" " ind2=" "><subfield code="a">GBV_ILN_4305</subfield></datafield><datafield tag="912" ind1=" " ind2=" "><subfield code="a">GBV_ILN_4306</subfield></datafield><datafield tag="912" ind1=" " ind2=" "><subfield code="a">GBV_ILN_4307</subfield></datafield><datafield tag="912" ind1=" " ind2=" "><subfield code="a">GBV_ILN_4313</subfield></datafield><datafield tag="912" ind1=" " ind2=" "><subfield code="a">GBV_ILN_4322</subfield></datafield><datafield tag="912" ind1=" " ind2=" "><subfield code="a">GBV_ILN_4323</subfield></datafield><datafield tag="912" ind1=" " ind2=" "><subfield code="a">GBV_ILN_4324</subfield></datafield><datafield tag="912" ind1=" " ind2=" "><subfield code="a">GBV_ILN_4325</subfield></datafield><datafield tag="912" ind1=" " ind2=" "><subfield code="a">GBV_ILN_4326</subfield></datafield><datafield tag="912" ind1=" " ind2=" "><subfield code="a">GBV_ILN_4335</subfield></datafield><datafield tag="912" ind1=" " ind2=" "><subfield code="a">GBV_ILN_4338</subfield></datafield><datafield tag="912" ind1=" " ind2=" "><subfield code="a">GBV_ILN_4367</subfield></datafield><datafield tag="912" ind1=" " ind2=" "><subfield code="a">GBV_ILN_4700</subfield></datafield><datafield tag="951" ind1=" " ind2=" "><subfield code="a">AR</subfield></datafield><datafield tag="952" ind1=" " ind2=" "><subfield code="d">9</subfield><subfield code="j">2022</subfield><subfield code="e">2</subfield><subfield code="b">16</subfield><subfield code="c">11</subfield><subfield code="h">2177-2188</subfield></datafield></record></collection>
score	7.397464

Nicht das Richtige dabei?

Schreiben Sie uns!

DynK-hydra: improved dynamic architecture ensembling for efficient inference

Nicht das Richtige dabei?

Zugang & Verfügbarkeit

Vorhandene Bände

Nicht das Richtige dabei?