Matches in SemOpenAlex for { <https://semopenalex.org/work/W2593156109> ?p ?o ?g. }
- W2593156109 abstract "Fully connected network has been widely used in deep learning, and its computation efficiency is highly benefited from the matrix multiplication algorithm with cuBLAS on GPU. However, We found that, there exist some drawbacks of cuBLAS in calculating matrix $textbf{A}$ multiplies the transpose of matrix $textbf{B}$ (i.e., NT operation). To reduce the impact of NT operation by cuBLAS, we exploit the out-of-place transpose of matrix $textbf{B}$ to avoid using NT operation, and then we apply our method to Caffe, which is a popular deep learning tool. Our contribution is two-fold. First, we propose a naive method (TNN) and model-based method (MTNN) to increase the performance in calculating $textbf{A}times textbf{B}^T$, and it achieves about 4.7 times performance enhancement in our tested cases on GTX1080 card. Second, we integrate MTNN method into Caffe to enhance the efficiency in training fully connected networks, which achieves about 70% speedup compared to the original Caffe in our configured fully connected networks on GTX1080 card." @default.
- W2593156109 created "2017-03-16" @default.
- W2593156109 creator A5016836702 @default.
- W2593156109 creator A5023321537 @default.
- W2593156109 creator A5053098275 @default.
- W2593156109 date "2017-02-10" @default.
- W2593156109 modified "2023-09-24" @default.
- W2593156109 title "Improving the Performance of Fully Connected Neural Networks by Out-of-Place Matrix Transpose" @default.
- W2593156109 cites W1495775210 @default.
- W2593156109 cites W1591180289 @default.
- W2593156109 cites W1596717185 @default.
- W2593156109 cites W1863336885 @default.
- W2593156109 cites W1978642402 @default.
- W2593156109 cites W1987882202 @default.
- W2593156109 cites W2039708501 @default.
- W2593156109 cites W2063186542 @default.
- W2593156109 cites W2090593986 @default.
- W2593156109 cites W2091883426 @default.
- W2593156109 cites W2093843662 @default.
- W2593156109 cites W2099021415 @default.
- W2593156109 cites W2119821739 @default.
- W2593156109 cites W2125055259 @default.
- W2593156109 cites W2129018774 @default.
- W2593156109 cites W2149706766 @default.
- W2593156109 cites W2153635508 @default.
- W2593156109 cites W2499931820 @default.
- W2593156109 cites W2514858228 @default.
- W2593156109 cites W2521497392 @default.
- W2593156109 cites W2950094539 @default.
- W2593156109 cites W2621593675 @default.
- W2593156109 hasPublicationYear "2017" @default.
- W2593156109 type Work @default.
- W2593156109 sameAs 2593156109 @default.
- W2593156109 citedByCount "1" @default.
- W2593156109 countsByYear W25931561092017 @default.
- W2593156109 crossrefType "posted-content" @default.
- W2593156109 hasAuthorship W2593156109A5016836702 @default.
- W2593156109 hasAuthorship W2593156109A5023321537 @default.
- W2593156109 hasAuthorship W2593156109A5053098275 @default.
- W2593156109 hasConcept C106487976 @default.
- W2593156109 hasConcept C108583219 @default.
- W2593156109 hasConcept C11413529 @default.
- W2593156109 hasConcept C114614502 @default.
- W2593156109 hasConcept C121332964 @default.
- W2593156109 hasConcept C154945302 @default.
- W2593156109 hasConcept C158693339 @default.
- W2593156109 hasConcept C159985019 @default.
- W2593156109 hasConcept C173608175 @default.
- W2593156109 hasConcept C192562407 @default.
- W2593156109 hasConcept C200106649 @default.
- W2593156109 hasConcept C2780595030 @default.
- W2593156109 hasConcept C2984842247 @default.
- W2593156109 hasConcept C33923547 @default.
- W2593156109 hasConcept C41008148 @default.
- W2593156109 hasConcept C45374587 @default.
- W2593156109 hasConcept C50644808 @default.
- W2593156109 hasConcept C62520636 @default.
- W2593156109 hasConcept C68339613 @default.
- W2593156109 hasConceptScore W2593156109C106487976 @default.
- W2593156109 hasConceptScore W2593156109C108583219 @default.
- W2593156109 hasConceptScore W2593156109C11413529 @default.
- W2593156109 hasConceptScore W2593156109C114614502 @default.
- W2593156109 hasConceptScore W2593156109C121332964 @default.
- W2593156109 hasConceptScore W2593156109C154945302 @default.
- W2593156109 hasConceptScore W2593156109C158693339 @default.
- W2593156109 hasConceptScore W2593156109C159985019 @default.
- W2593156109 hasConceptScore W2593156109C173608175 @default.
- W2593156109 hasConceptScore W2593156109C192562407 @default.
- W2593156109 hasConceptScore W2593156109C200106649 @default.
- W2593156109 hasConceptScore W2593156109C2780595030 @default.
- W2593156109 hasConceptScore W2593156109C2984842247 @default.
- W2593156109 hasConceptScore W2593156109C33923547 @default.
- W2593156109 hasConceptScore W2593156109C41008148 @default.
- W2593156109 hasConceptScore W2593156109C45374587 @default.
- W2593156109 hasConceptScore W2593156109C50644808 @default.
- W2593156109 hasConceptScore W2593156109C62520636 @default.
- W2593156109 hasConceptScore W2593156109C68339613 @default.
- W2593156109 hasLocation W25931561091 @default.
- W2593156109 hasOpenAccess W2593156109 @default.
- W2593156109 hasPrimaryLocation W25931561091 @default.
- W2593156109 hasRelatedWork W1857970128 @default.
- W2593156109 hasRelatedWork W2075920907 @default.
- W2593156109 hasRelatedWork W2079443367 @default.
- W2593156109 hasRelatedWork W2095363965 @default.
- W2593156109 hasRelatedWork W2145800083 @default.
- W2593156109 hasRelatedWork W2177387462 @default.
- W2593156109 hasRelatedWork W2260663238 @default.
- W2593156109 hasRelatedWork W2271936473 @default.
- W2593156109 hasRelatedWork W2272379656 @default.
- W2593156109 hasRelatedWork W2365142522 @default.
- W2593156109 hasRelatedWork W2474388053 @default.
- W2593156109 hasRelatedWork W2549412286 @default.
- W2593156109 hasRelatedWork W2903669223 @default.
- W2593156109 hasRelatedWork W2962988160 @default.
- W2593156109 hasRelatedWork W3036692157 @default.
- W2593156109 hasRelatedWork W3081486497 @default.
- W2593156109 hasRelatedWork W3129397352 @default.
- W2593156109 hasRelatedWork W3132695675 @default.
- W2593156109 hasRelatedWork W3142663483 @default.
- W2593156109 hasRelatedWork W2742563775 @default.