DARTS algorithm process.
<div><p>Image classification, as the core task of computer vision, has broad application value in fields such as medical diagnosis and intelligent transportation.However, the ability of differentiable neural architecture to search (NAS) for local information is weak, which limits the acc...
Saved in:
| Main Author: | |
|---|---|
| Other Authors: | , |
| Published: |
2025
|
| Subjects: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1852017627070726144 |
|---|---|
| author | Yuxuan Ji (13991895) |
| author2 | Wenshu Li (762627) Nan Yu (757123) |
| author2_role | author author |
| author_facet | Yuxuan Ji (13991895) Wenshu Li (762627) Nan Yu (757123) |
| author_role | author |
| dc.creator.none.fl_str_mv | Yuxuan Ji (13991895) Wenshu Li (762627) Nan Yu (757123) |
| dc.date.none.fl_str_mv | 2025-08-13T20:22:00Z |
| dc.identifier.none.fl_str_mv | 10.1371/journal.pone.0329480.g001 |
| dc.relation.none.fl_str_mv | https://figshare.com/articles/figure/DARTS_algorithm_process_/29904843 |
| dc.rights.none.fl_str_mv | CC BY 4.0 info:eu-repo/semantics/openAccess |
| dc.subject.none.fl_str_mv | Biotechnology Biological Sciences not elsewhere classified Information Systems not elsewhere classified runtime memory usage original convolution operator differentiable neural architecture broad application value adds residual structure model &# 8217 search parameter required research results show information acquisition ability research model results indicate macro structure research method improved model baseline model significantly better search time point operations minimum number medical diagnosis mainstream algorithms local information intelligent transportation imagenet dataset effectively solve core task computer vision 600 rounds 52 %, 2 %. 100 dataset 10 dataset |
| dc.title.none.fl_str_mv | DARTS algorithm process. |
| dc.type.none.fl_str_mv | Image Figure info:eu-repo/semantics/publishedVersion image |
| description | <div><p>Image classification, as the core task of computer vision, has broad application value in fields such as medical diagnosis and intelligent transportation.However, the ability of differentiable neural architecture to search (NAS) for local information is weak, which limits the accuracy and long-distance information capture capability of the algorithm. Therefore, based on this, the study introduces visual attention mechanism and proposes an improved model that replaces the original convolution operator and adds residual structure in the macro structure to enhance the model’s information acquisition ability and classification accuracy. The research results show that after 600 rounds of training on the CIFAR-10 dataset, the final accuracy of the improved model reached 97.2%. The runtime memory usage on the CIFAR-100 dataset is only 44.52%, a decrease of 44.56% compared to the baseline model. In the testing on the ImageNet dataset, the classification accuracy of the research model is 94.01, the search parameter required is only 4.8MB, the search time is shortened to 0.5d, and the minimum number of floating-point operations is 3.7G, significantly better than other mainstream algorithms. The above results indicate that the research method can effectively solve the shortcomings of traditional differentiable neural architecture search in local and remote information acquisition capabilities, providing important technical support for improving the accuracy and efficiency of image classification technology.</p></div> |
| eu_rights_str_mv | openAccess |
| id | Manara_58bfae5dd3f155f3492ddc4fc26e4e81 |
| identifier_str_mv | 10.1371/journal.pone.0329480.g001 |
| network_acronym_str | Manara |
| network_name_str | ManaraRepo |
| oai_identifier_str | oai:figshare.com:article/29904843 |
| publishDate | 2025 |
| repository.mail.fl_str_mv | |
| repository.name.fl_str_mv | |
| repository_id_str | |
| rights_invalid_str_mv | CC BY 4.0 |
| spelling | DARTS algorithm process.Yuxuan Ji (13991895)Wenshu Li (762627)Nan Yu (757123)BiotechnologyBiological Sciences not elsewhere classifiedInformation Systems not elsewhere classifiedruntime memory usageoriginal convolution operatordifferentiable neural architecturebroad application valueadds residual structuremodel &# 8217search parameter requiredresearch results showinformation acquisition abilityresearch modelresults indicatemacro structureresearch methodimproved modelbaseline modelsignificantly bettersearch timepoint operationsminimum numbermedical diagnosismainstream algorithmslocal informationintelligent transportationimagenet dataseteffectively solvecore taskcomputer vision600 rounds52 %,2 %.100 dataset10 dataset<div><p>Image classification, as the core task of computer vision, has broad application value in fields such as medical diagnosis and intelligent transportation.However, the ability of differentiable neural architecture to search (NAS) for local information is weak, which limits the accuracy and long-distance information capture capability of the algorithm. Therefore, based on this, the study introduces visual attention mechanism and proposes an improved model that replaces the original convolution operator and adds residual structure in the macro structure to enhance the model’s information acquisition ability and classification accuracy. The research results show that after 600 rounds of training on the CIFAR-10 dataset, the final accuracy of the improved model reached 97.2%. The runtime memory usage on the CIFAR-100 dataset is only 44.52%, a decrease of 44.56% compared to the baseline model. In the testing on the ImageNet dataset, the classification accuracy of the research model is 94.01, the search parameter required is only 4.8MB, the search time is shortened to 0.5d, and the minimum number of floating-point operations is 3.7G, significantly better than other mainstream algorithms. The above results indicate that the research method can effectively solve the shortcomings of traditional differentiable neural architecture search in local and remote information acquisition capabilities, providing important technical support for improving the accuracy and efficiency of image classification technology.</p></div>2025-08-13T20:22:00ZImageFigureinfo:eu-repo/semantics/publishedVersionimage10.1371/journal.pone.0329480.g001https://figshare.com/articles/figure/DARTS_algorithm_process_/29904843CC BY 4.0info:eu-repo/semantics/openAccessoai:figshare.com:article/299048432025-08-13T20:22:00Z |
| spellingShingle | DARTS algorithm process. Yuxuan Ji (13991895) Biotechnology Biological Sciences not elsewhere classified Information Systems not elsewhere classified runtime memory usage original convolution operator differentiable neural architecture broad application value adds residual structure model &# 8217 search parameter required research results show information acquisition ability research model results indicate macro structure research method improved model baseline model significantly better search time point operations minimum number medical diagnosis mainstream algorithms local information intelligent transportation imagenet dataset effectively solve core task computer vision 600 rounds 52 %, 2 %. 100 dataset 10 dataset |
| status_str | publishedVersion |
| title | DARTS algorithm process. |
| title_full | DARTS algorithm process. |
| title_fullStr | DARTS algorithm process. |
| title_full_unstemmed | DARTS algorithm process. |
| title_short | DARTS algorithm process. |
| title_sort | DARTS algorithm process. |
| topic | Biotechnology Biological Sciences not elsewhere classified Information Systems not elsewhere classified runtime memory usage original convolution operator differentiable neural architecture broad application value adds residual structure model &# 8217 search parameter required research results show information acquisition ability research model results indicate macro structure research method improved model baseline model significantly better search time point operations minimum number medical diagnosis mainstream algorithms local information intelligent transportation imagenet dataset effectively solve core task computer vision 600 rounds 52 %, 2 %. 100 dataset 10 dataset |