Matches in SemOpenAlex for { <https://semopenalex.org/work/W3191506988> ?p ?o ?g. }
Showing items 1 to 86 of
86
with 100 items per page.
- W3191506988 abstract "CNN is the most widely used method for image classification. In many occasions, its accuracy is improved by combining it with several boosting techniques. Determining such an efficient boosting model can be very time conserving and benevolent for many image classification applications. This study aims at analyzing and evaluating the performance of CNN added with various boosting techniques, namely XGBoost, Gradient Boost and AdaBoost. At first, CNN mode is trained, and image features are extracted by using convolutional layers, then for training the boosting algorithms intermediate data is extracted from three different layers of the CNN model. After comparing the performance of each individual techniques, it is revealed that boosted CNNs overperforms the others. While comparing the performances of the three CNN-Boosting algorithms, using data extracted from the flatten layer and the first dense layer, CNN-Gradient Boost performs better with an average of approximately 0.1921% more test accuracy and 0.2455% more F1 score for the flatten layer and 0.0505% more test accuracy and 0.0479% more F1 score for the first dense layer. For the third dense layer, CNN-XGBoost achieves better result with 0.0505% more test accuracy and 0.0479% more F1 score than the others. This finding is beneficial to improve performance of any application requiring image classification." @default.
- W3191506988 created "2021-08-16" @default.
- W3191506988 creator A5006349493 @default.
- W3191506988 creator A5061984654 @default.
- W3191506988 creator A5065064280 @default.
- W3191506988 creator A5086217459 @default.
- W3191506988 date "2021-06-25" @default.
- W3191506988 modified "2023-10-18" @default.
- W3191506988 title "Analyzing and Evaluating Boosting-Based CNN Algorithms for Image Classification" @default.
- W3191506988 cites W1559535184 @default.
- W3191506988 cites W1968969471 @default.
- W3191506988 cites W1972290682 @default.
- W3191506988 cites W2007339694 @default.
- W3191506988 cites W2028571532 @default.
- W3191506988 cites W2044465660 @default.
- W3191506988 cites W2070493638 @default.
- W3191506988 cites W2123175289 @default.
- W3191506988 cites W2168809519 @default.
- W3191506988 cites W2520826941 @default.
- W3191506988 cites W2554892747 @default.
- W3191506988 cites W2734358244 @default.
- W3191506988 cites W2747876883 @default.
- W3191506988 cites W2766908198 @default.
- W3191506988 cites W2791460829 @default.
- W3191506988 cites W2964015640 @default.
- W3191506988 cites W3014666486 @default.
- W3191506988 cites W3046220160 @default.
- W3191506988 cites W3102476541 @default.
- W3191506988 cites W3127686408 @default.
- W3191506988 cites W4244895750 @default.
- W3191506988 doi "https://doi.org/10.1109/conit51480.2021.9498328" @default.
- W3191506988 hasPublicationYear "2021" @default.
- W3191506988 type Work @default.
- W3191506988 sameAs 3191506988 @default.
- W3191506988 citedByCount "1" @default.
- W3191506988 countsByYear W31915069882022 @default.
- W3191506988 crossrefType "proceedings-article" @default.
- W3191506988 hasAuthorship W3191506988A5006349493 @default.
- W3191506988 hasAuthorship W3191506988A5061984654 @default.
- W3191506988 hasAuthorship W3191506988A5065064280 @default.
- W3191506988 hasAuthorship W3191506988A5086217459 @default.
- W3191506988 hasConcept C110083411 @default.
- W3191506988 hasConcept C11413529 @default.
- W3191506988 hasConcept C115961682 @default.
- W3191506988 hasConcept C119857082 @default.
- W3191506988 hasConcept C12267149 @default.
- W3191506988 hasConcept C141404830 @default.
- W3191506988 hasConcept C153180895 @default.
- W3191506988 hasConcept C154945302 @default.
- W3191506988 hasConcept C169258074 @default.
- W3191506988 hasConcept C41008148 @default.
- W3191506988 hasConcept C46686674 @default.
- W3191506988 hasConcept C70153297 @default.
- W3191506988 hasConcept C75294576 @default.
- W3191506988 hasConcept C81363708 @default.
- W3191506988 hasConceptScore W3191506988C110083411 @default.
- W3191506988 hasConceptScore W3191506988C11413529 @default.
- W3191506988 hasConceptScore W3191506988C115961682 @default.
- W3191506988 hasConceptScore W3191506988C119857082 @default.
- W3191506988 hasConceptScore W3191506988C12267149 @default.
- W3191506988 hasConceptScore W3191506988C141404830 @default.
- W3191506988 hasConceptScore W3191506988C153180895 @default.
- W3191506988 hasConceptScore W3191506988C154945302 @default.
- W3191506988 hasConceptScore W3191506988C169258074 @default.
- W3191506988 hasConceptScore W3191506988C41008148 @default.
- W3191506988 hasConceptScore W3191506988C46686674 @default.
- W3191506988 hasConceptScore W3191506988C70153297 @default.
- W3191506988 hasConceptScore W3191506988C75294576 @default.
- W3191506988 hasConceptScore W3191506988C81363708 @default.
- W3191506988 hasLocation W31915069881 @default.
- W3191506988 hasOpenAccess W3191506988 @default.
- W3191506988 hasPrimaryLocation W31915069881 @default.
- W3191506988 hasRelatedWork W2108043611 @default.
- W3191506988 hasRelatedWork W2944309180 @default.
- W3191506988 hasRelatedWork W3210229324 @default.
- W3191506988 hasRelatedWork W4200057378 @default.
- W3191506988 hasRelatedWork W4200061235 @default.
- W3191506988 hasRelatedWork W4293069612 @default.
- W3191506988 hasRelatedWork W4296081764 @default.
- W3191506988 hasRelatedWork W4304142064 @default.
- W3191506988 hasRelatedWork W4318716840 @default.
- W3191506988 hasRelatedWork W4323060081 @default.
- W3191506988 isParatext "false" @default.
- W3191506988 isRetracted "false" @default.
- W3191506988 magId "3191506988" @default.
- W3191506988 workType "article" @default.