Offloading the computational complexity of transfer learning with generic features
Deep learning approaches are generally complex, requiring extensive computational resources and having high time complexity. Transfer learning is a state-of-the-art approach to reducing the requirements of high computational resources by using pre-trained models without compromising accuracy and per...
| Main Authors: | , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
PeerJ Inc.
2024-03-01
|
| Series: | PeerJ Computer Science |
| Subjects: | |
| Online Access: | https://peerj.com/articles/cs-1938.pdf |
| _version_ | 1797238992694935552 |
|---|---|
| author | Muhammad Safdar Ali Khan Arif Husen Shafaq Nisar Hasnain Ahmed Syed Shah Muhammad Shabib Aftab |
| author_facet | Muhammad Safdar Ali Khan Arif Husen Shafaq Nisar Hasnain Ahmed Syed Shah Muhammad Shabib Aftab |
| author_sort | Muhammad Safdar Ali Khan |
| collection | DOAJ |
| description | Deep learning approaches are generally complex, requiring extensive computational resources and having high time complexity. Transfer learning is a state-of-the-art approach to reducing the requirements of high computational resources by using pre-trained models without compromising accuracy and performance. In conventional studies, pre-trained models are trained on datasets from different but similar domains with many domain-specific features. The computational requirements of transfer learning are directly dependent on the number of features that include the domain-specific and the generic features. This article investigates the prospects of reducing the computational requirements of the transfer learning models by discarding domain-specific features from a pre-trained model. The approach is applied to breast cancer detection using the dataset curated breast imaging subset of the digital database for screening mammography and various performance metrics such as precision, accuracy, recall, F1-score, and computational requirements. It is seen that discarding the domain-specific features to a specific limit provides significant performance improvements as well as minimizes the computational requirements in terms of training time (reduced by approx. 12%), processor utilization (reduced approx. 25%), and memory usage (reduced approx. 22%). The proposed transfer learning strategy increases accuracy (approx. 7%) and offloads computational complexity expeditiously. |
| first_indexed | 2024-04-24T17:44:27Z |
| format | Article |
| id | doaj.art-e52eedd3ae8a409dade49873af42170a |
| institution | Directory Open Access Journal |
| issn | 2376-5992 |
| language | English |
| last_indexed | 2024-04-24T17:44:27Z |
| publishDate | 2024-03-01 |
| publisher | PeerJ Inc. |
| record_format | Article |
| series | PeerJ Computer Science |
| spelling | doaj.art-e52eedd3ae8a409dade49873af42170a2024-03-27T15:05:12ZengPeerJ Inc.PeerJ Computer Science2376-59922024-03-0110e193810.7717/peerj-cs.1938Offloading the computational complexity of transfer learning with generic featuresMuhammad Safdar Ali Khan0Arif Husen1Shafaq Nisar2Hasnain Ahmed3Syed Shah Muhammad4Shabib Aftab5Department of Computer Science and Information Technology, Virtual University of Pakistan, Lahore, Punjab, PakistanDepartment of Computer Science and Information Technology, Virtual University of Pakistan, Lahore, Punjab, PakistanDepartment of Computer Science and Information Technology, Virtual University of Pakistan, Lahore, Punjab, PakistanDepartment of Computer Science and Information Technology, Virtual University of Pakistan, Lahore, Punjab, PakistanDepartment of Computer Science and Information Technology, Virtual University of Pakistan, Lahore, Punjab, PakistanDepartment of Computer Science and Information Technology, Virtual University of Pakistan, Lahore, Punjab, PakistanDeep learning approaches are generally complex, requiring extensive computational resources and having high time complexity. Transfer learning is a state-of-the-art approach to reducing the requirements of high computational resources by using pre-trained models without compromising accuracy and performance. In conventional studies, pre-trained models are trained on datasets from different but similar domains with many domain-specific features. The computational requirements of transfer learning are directly dependent on the number of features that include the domain-specific and the generic features. This article investigates the prospects of reducing the computational requirements of the transfer learning models by discarding domain-specific features from a pre-trained model. The approach is applied to breast cancer detection using the dataset curated breast imaging subset of the digital database for screening mammography and various performance metrics such as precision, accuracy, recall, F1-score, and computational requirements. It is seen that discarding the domain-specific features to a specific limit provides significant performance improvements as well as minimizes the computational requirements in terms of training time (reduced by approx. 12%), processor utilization (reduced approx. 25%), and memory usage (reduced approx. 22%). The proposed transfer learning strategy increases accuracy (approx. 7%) and offloads computational complexity expeditiously.https://peerj.com/articles/cs-1938.pdfGeneric featuresDomain specific featuresDeep learningCancer detectionCancer classificationTransfer learning |
| spellingShingle | Muhammad Safdar Ali Khan Arif Husen Shafaq Nisar Hasnain Ahmed Syed Shah Muhammad Shabib Aftab Offloading the computational complexity of transfer learning with generic features PeerJ Computer Science Generic features Domain specific features Deep learning Cancer detection Cancer classification Transfer learning |
| title | Offloading the computational complexity of transfer learning with generic features |
| title_full | Offloading the computational complexity of transfer learning with generic features |
| title_fullStr | Offloading the computational complexity of transfer learning with generic features |
| title_full_unstemmed | Offloading the computational complexity of transfer learning with generic features |
| title_short | Offloading the computational complexity of transfer learning with generic features |
| title_sort | offloading the computational complexity of transfer learning with generic features |
| topic | Generic features Domain specific features Deep learning Cancer detection Cancer classification Transfer learning |
| url | https://peerj.com/articles/cs-1938.pdf |
| work_keys_str_mv | AT muhammadsafdaralikhan offloadingthecomputationalcomplexityoftransferlearningwithgenericfeatures AT arifhusen offloadingthecomputationalcomplexityoftransferlearningwithgenericfeatures AT shafaqnisar offloadingthecomputationalcomplexityoftransferlearningwithgenericfeatures AT hasnainahmed offloadingthecomputationalcomplexityoftransferlearningwithgenericfeatures AT syedshahmuhammad offloadingthecomputationalcomplexityoftransferlearningwithgenericfeatures AT shabibaftab offloadingthecomputationalcomplexityoftransferlearningwithgenericfeatures |