Comparison of the computational cost and the number of weights for the uncompressed, structurally-compressed, and deep-compressed models. The computational cost is quantified by FLOPS, i.e., the number of floating-point operations required in each layer to classify one image. The compression ratio is defined as the number of FLOPS (or weights) required for the uncompressed model divided by those required for the compressed model. Note that for the structurally-compression model, the compression ratio based on the number of FLOPS is equal to the compression ratio based on the number weights because all filters are removed during this form of compression. Structural compression is also not defined for the fully connected layers; therefore no number is reported for these layers.
<p>Comparison of the computational cost and the number of weights for the uncompressed, structurally-compressed, and deep-compressed models. The computational cost is quantified by FLOPS, i.e., the number of floating-point operations required in each layer to classify one image. The compressio...
Saved in:
| Main Author: | |
|---|---|
| Other Authors: | , , |
| Published: |
2025
|
| Subjects: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1852022593141342208 |
|---|---|
| author | Fatemeh Kamali (20752576) |
| author2 | Amir Abolfazl Suratgar (20752579) Mohammadbagher Menhaj (20752582) Reza Abbasi-Asl (5953967) |
| author2_role | author author author |
| author_facet | Fatemeh Kamali (20752576) Amir Abolfazl Suratgar (20752579) Mohammadbagher Menhaj (20752582) Reza Abbasi-Asl (5953967) |
| author_role | author |
| dc.creator.none.fl_str_mv | Fatemeh Kamali (20752576) Amir Abolfazl Suratgar (20752579) Mohammadbagher Menhaj (20752582) Reza Abbasi-Asl (5953967) |
| dc.date.none.fl_str_mv | 2025-02-19T18:33:06Z |
| dc.identifier.none.fl_str_mv | 10.1371/journal.pcbi.1012822.t001 |
| dc.relation.none.fl_str_mv | https://figshare.com/articles/dataset/Comparison_of_the_computational_cost_and_the_number_of_weights_for_the_uncompressed_structurally-compressed_and_deep-compressed_models_The_computational_cost_is_quantified_by_FLOPS_i_e_the_number_of_floating-point_operations_required_in_eac/28446763 |
| dc.rights.none.fl_str_mv | CC BY 4.0 info:eu-repo/semantics/openAccess |
| dc.subject.none.fl_str_mv | Genetics Biotechnology Evolutionary Biology Ecology Cancer Biological Sciences not elsewhere classified Mathematical Sciences not elsewhere classified Information Systems not elsewhere classified voxelwise pattern selectivity ventral visual pathway principal component analysis identifying visual stimuli convolutional neural networks brain activity evoked superior predictive performance select receptive fields compressed models reveal compressed models offer based voxelwise models model compression improves predictive models based models uncompressed models receptive field optimal model widely used test set stable interpretation reduce dimensionality optimal center natural movies huge number enabled interpretability centralized along |
| dc.title.none.fl_str_mv | Comparison of the computational cost and the number of weights for the uncompressed, structurally-compressed, and deep-compressed models. The computational cost is quantified by FLOPS, i.e., the number of floating-point operations required in each layer to classify one image. The compression ratio is defined as the number of FLOPS (or weights) required for the uncompressed model divided by those required for the compressed model. Note that for the structurally-compression model, the compression ratio based on the number of FLOPS is equal to the compression ratio based on the number weights because all filters are removed during this form of compression. Structural compression is also not defined for the fully connected layers; therefore no number is reported for these layers. |
| dc.type.none.fl_str_mv | Dataset info:eu-repo/semantics/publishedVersion dataset |
| description | <p>Comparison of the computational cost and the number of weights for the uncompressed, structurally-compressed, and deep-compressed models. The computational cost is quantified by FLOPS, i.e., the number of floating-point operations required in each layer to classify one image. The compression ratio is defined as the number of FLOPS (or weights) required for the uncompressed model divided by those required for the compressed model. Note that for the structurally-compression model, the compression ratio based on the number of FLOPS is equal to the compression ratio based on the number weights because all filters are removed during this form of compression. Structural compression is also not defined for the fully connected layers; therefore no number is reported for these layers.</p> |
| eu_rights_str_mv | openAccess |
| id | Manara_4c7a705b2e06bb3e6b3bc6edcf83bfef |
| identifier_str_mv | 10.1371/journal.pcbi.1012822.t001 |
| network_acronym_str | Manara |
| network_name_str | ManaraRepo |
| oai_identifier_str | oai:figshare.com:article/28446763 |
| publishDate | 2025 |
| repository.mail.fl_str_mv | |
| repository.name.fl_str_mv | |
| repository_id_str | |
| rights_invalid_str_mv | CC BY 4.0 |
| spelling | Comparison of the computational cost and the number of weights for the uncompressed, structurally-compressed, and deep-compressed models. The computational cost is quantified by FLOPS, i.e., the number of floating-point operations required in each layer to classify one image. The compression ratio is defined as the number of FLOPS (or weights) required for the uncompressed model divided by those required for the compressed model. Note that for the structurally-compression model, the compression ratio based on the number of FLOPS is equal to the compression ratio based on the number weights because all filters are removed during this form of compression. Structural compression is also not defined for the fully connected layers; therefore no number is reported for these layers.Fatemeh Kamali (20752576)Amir Abolfazl Suratgar (20752579)Mohammadbagher Menhaj (20752582)Reza Abbasi-Asl (5953967)GeneticsBiotechnologyEvolutionary BiologyEcologyCancerBiological Sciences not elsewhere classifiedMathematical Sciences not elsewhere classifiedInformation Systems not elsewhere classifiedvoxelwise pattern selectivityventral visual pathwayprincipal component analysisidentifying visual stimuliconvolutional neural networksbrain activity evokedsuperior predictive performanceselect receptive fieldscompressed models revealcompressed models offerbased voxelwise modelsmodel compression improvespredictive modelsbased modelsuncompressed modelsreceptive fieldoptimal modelwidely usedtest setstable interpretationreduce dimensionalityoptimal centernatural movieshuge numberenabled interpretabilitycentralized along<p>Comparison of the computational cost and the number of weights for the uncompressed, structurally-compressed, and deep-compressed models. The computational cost is quantified by FLOPS, i.e., the number of floating-point operations required in each layer to classify one image. The compression ratio is defined as the number of FLOPS (or weights) required for the uncompressed model divided by those required for the compressed model. Note that for the structurally-compression model, the compression ratio based on the number of FLOPS is equal to the compression ratio based on the number weights because all filters are removed during this form of compression. Structural compression is also not defined for the fully connected layers; therefore no number is reported for these layers.</p>2025-02-19T18:33:06ZDatasetinfo:eu-repo/semantics/publishedVersiondataset10.1371/journal.pcbi.1012822.t001https://figshare.com/articles/dataset/Comparison_of_the_computational_cost_and_the_number_of_weights_for_the_uncompressed_structurally-compressed_and_deep-compressed_models_The_computational_cost_is_quantified_by_FLOPS_i_e_the_number_of_floating-point_operations_required_in_eac/28446763CC BY 4.0info:eu-repo/semantics/openAccessoai:figshare.com:article/284467632025-02-19T18:33:06Z |
| spellingShingle | Comparison of the computational cost and the number of weights for the uncompressed, structurally-compressed, and deep-compressed models. The computational cost is quantified by FLOPS, i.e., the number of floating-point operations required in each layer to classify one image. The compression ratio is defined as the number of FLOPS (or weights) required for the uncompressed model divided by those required for the compressed model. Note that for the structurally-compression model, the compression ratio based on the number of FLOPS is equal to the compression ratio based on the number weights because all filters are removed during this form of compression. Structural compression is also not defined for the fully connected layers; therefore no number is reported for these layers. Fatemeh Kamali (20752576) Genetics Biotechnology Evolutionary Biology Ecology Cancer Biological Sciences not elsewhere classified Mathematical Sciences not elsewhere classified Information Systems not elsewhere classified voxelwise pattern selectivity ventral visual pathway principal component analysis identifying visual stimuli convolutional neural networks brain activity evoked superior predictive performance select receptive fields compressed models reveal compressed models offer based voxelwise models model compression improves predictive models based models uncompressed models receptive field optimal model widely used test set stable interpretation reduce dimensionality optimal center natural movies huge number enabled interpretability centralized along |
| status_str | publishedVersion |
| title | Comparison of the computational cost and the number of weights for the uncompressed, structurally-compressed, and deep-compressed models. The computational cost is quantified by FLOPS, i.e., the number of floating-point operations required in each layer to classify one image. The compression ratio is defined as the number of FLOPS (or weights) required for the uncompressed model divided by those required for the compressed model. Note that for the structurally-compression model, the compression ratio based on the number of FLOPS is equal to the compression ratio based on the number weights because all filters are removed during this form of compression. Structural compression is also not defined for the fully connected layers; therefore no number is reported for these layers. |
| title_full | Comparison of the computational cost and the number of weights for the uncompressed, structurally-compressed, and deep-compressed models. The computational cost is quantified by FLOPS, i.e., the number of floating-point operations required in each layer to classify one image. The compression ratio is defined as the number of FLOPS (or weights) required for the uncompressed model divided by those required for the compressed model. Note that for the structurally-compression model, the compression ratio based on the number of FLOPS is equal to the compression ratio based on the number weights because all filters are removed during this form of compression. Structural compression is also not defined for the fully connected layers; therefore no number is reported for these layers. |
| title_fullStr | Comparison of the computational cost and the number of weights for the uncompressed, structurally-compressed, and deep-compressed models. The computational cost is quantified by FLOPS, i.e., the number of floating-point operations required in each layer to classify one image. The compression ratio is defined as the number of FLOPS (or weights) required for the uncompressed model divided by those required for the compressed model. Note that for the structurally-compression model, the compression ratio based on the number of FLOPS is equal to the compression ratio based on the number weights because all filters are removed during this form of compression. Structural compression is also not defined for the fully connected layers; therefore no number is reported for these layers. |
| title_full_unstemmed | Comparison of the computational cost and the number of weights for the uncompressed, structurally-compressed, and deep-compressed models. The computational cost is quantified by FLOPS, i.e., the number of floating-point operations required in each layer to classify one image. The compression ratio is defined as the number of FLOPS (or weights) required for the uncompressed model divided by those required for the compressed model. Note that for the structurally-compression model, the compression ratio based on the number of FLOPS is equal to the compression ratio based on the number weights because all filters are removed during this form of compression. Structural compression is also not defined for the fully connected layers; therefore no number is reported for these layers. |
| title_short | Comparison of the computational cost and the number of weights for the uncompressed, structurally-compressed, and deep-compressed models. The computational cost is quantified by FLOPS, i.e., the number of floating-point operations required in each layer to classify one image. The compression ratio is defined as the number of FLOPS (or weights) required for the uncompressed model divided by those required for the compressed model. Note that for the structurally-compression model, the compression ratio based on the number of FLOPS is equal to the compression ratio based on the number weights because all filters are removed during this form of compression. Structural compression is also not defined for the fully connected layers; therefore no number is reported for these layers. |
| title_sort | Comparison of the computational cost and the number of weights for the uncompressed, structurally-compressed, and deep-compressed models. The computational cost is quantified by FLOPS, i.e., the number of floating-point operations required in each layer to classify one image. The compression ratio is defined as the number of FLOPS (or weights) required for the uncompressed model divided by those required for the compressed model. Note that for the structurally-compression model, the compression ratio based on the number of FLOPS is equal to the compression ratio based on the number weights because all filters are removed during this form of compression. Structural compression is also not defined for the fully connected layers; therefore no number is reported for these layers. |
| topic | Genetics Biotechnology Evolutionary Biology Ecology Cancer Biological Sciences not elsewhere classified Mathematical Sciences not elsewhere classified Information Systems not elsewhere classified voxelwise pattern selectivity ventral visual pathway principal component analysis identifying visual stimuli convolutional neural networks brain activity evoked superior predictive performance select receptive fields compressed models reveal compressed models offer based voxelwise models model compression improves predictive models based models uncompressed models receptive field optimal model widely used test set stable interpretation reduce dimensionality optimal center natural movies huge number enabled interpretability centralized along |