FRINK Linked Data Fragments server

Matches in SemOpenAlex for { <https://semopenalex.org/work/W2768348081> ?p ?o ?g. }

• W2768348081 endingPage "3157" @default.
• W2768348081 startingPage "3149" @default.
• W2768348081 abstract "Gradient Boosting Decision Tree (GBDT) is a popular machine learning algorithm, and has quite a few effective implementations such as XGBoost and pGBRT. Although many engineering optimizations have been adopted in these implementations, the efficiency and scalability are still unsatisfactory when the feature dimension is high and data size is large. A major reason is that for each feature, they need to scan all the data instances to estimate the information gain of all possible split points, which is very time consuming. To tackle this problem, we propose two novel techniques: Gradient-based One-Side Sampling (GOSS) and Exclusive Feature Bundling (EFB). With GOSS, we exclude a significant proportion of data instances with small gradients, and only use the rest to estimate the information gain. We prove that, since the data instances with larger gradients play a more important role in the computation of information gain, GOSS can obtain quite accurate estimation of the information gain with a much smaller data size. With EFB, we bundle mutually exclusive features (i.e., they rarely take nonzero values simultaneously), to reduce the number of features. We prove that finding the optimal bundling of exclusive features is NP-hard, but a greedy algorithm can achieve quite good approximation ratio (and thus can effectively reduce the number of features without hurting the accuracy of split point determination by much). We call our new GBDT implementation with GOSS and EFB LightGBM. Our experiments on multiple public datasets show that, LightGBM speeds up the training process of conventional GBDT by up to over 20 times while achieving almost the same accuracy." @default.
• W2768348081 created "2017-12-04" @default.
• W2768348081 creator A5007444897 @default.
• W2768348081 creator A5017541508 @default.
• W2768348081 creator A5044802273 @default.
• W2768348081 creator A5061200287 @default.
• W2768348081 creator A5068656698 @default.
• W2768348081 creator A5070990160 @default.
• W2768348081 creator A5074676968 @default.
• W2768348081 creator A5080903368 @default.
• W2768348081 date "2017-12-04" @default.
• W2768348081 modified "2023-10-03" @default.
• W2768348081 title "LightGBM: a highly efficient gradient boosting decision tree" @default.
• W2768348081 cites W1483135265 @default.
• W2768348081 cites W1489069776 @default.
• W2768348081 cites W1530210183 @default.
• W2768348081 cites W1542798451 @default.
• W2768348081 cites W1558918611 @default.
• W2768348081 cites W1576962511 @default.
• W2768348081 cites W1678356000 @default.
• W2768348081 cites W1987356990 @default.
• W2768348081 cites W2024046085 @default.
• W2768348081 cites W2070493638 @default.
• W2768348081 cites W2090883204 @default.
• W2768348081 cites W2109395398 @default.
• W2768348081 cites W2113635748 @default.
• W2768348081 cites W2115328185 @default.
• W2768348081 cites W2120391124 @default.
• W2768348081 cites W2165653993 @default.
• W2768348081 cites W2171012768 @default.
• W2768348081 cites W2188460664 @default.
• W2768348081 cites W2553372643 @default.
• W2768348081 cites W2604808181 @default.
• W2768348081 cites W2652074165 @default.
• W2768348081 cites W2962979321 @default.
• W2768348081 cites W3099514962 @default.
• W2768348081 cites W3102476541 @default.
• W2768348081 hasPublicationYear "2017" @default.
• W2768348081 type Work @default.
• W2768348081 sameAs 2768348081 @default.
• W2768348081 citedByCount "591" @default.
• W2768348081 countsByYear W27683480812017 @default.
• W2768348081 countsByYear W27683480812018 @default.
• W2768348081 countsByYear W27683480812019 @default.
• W2768348081 countsByYear W27683480812020 @default.
• W2768348081 countsByYear W27683480812021 @default.
• W2768348081 countsByYear W27683480812022 @default.
• W2768348081 crossrefType "proceedings-article" @default.
• W2768348081 hasAuthorship W2768348081A5007444897 @default.
• W2768348081 hasAuthorship W2768348081A5017541508 @default.
• W2768348081 hasAuthorship W2768348081A5044802273 @default.
• W2768348081 hasAuthorship W2768348081A5061200287 @default.
• W2768348081 hasAuthorship W2768348081A5068656698 @default.
• W2768348081 hasAuthorship W2768348081A5070990160 @default.
• W2768348081 hasAuthorship W2768348081A5074676968 @default.
• W2768348081 hasAuthorship W2768348081A5080903368 @default.
• W2768348081 hasConcept C11413529 @default.
• W2768348081 hasConcept C119857082 @default.
• W2768348081 hasConcept C124101348 @default.
• W2768348081 hasConcept C138885662 @default.
• W2768348081 hasConcept C154945302 @default.
• W2768348081 hasConcept C169258074 @default.
• W2768348081 hasConcept C199360897 @default.
• W2768348081 hasConcept C26713055 @default.
• W2768348081 hasConcept C2776401178 @default.
• W2768348081 hasConcept C41008148 @default.
• W2768348081 hasConcept C41895202 @default.
• W2768348081 hasConcept C45374587 @default.
• W2768348081 hasConcept C46686674 @default.
• W2768348081 hasConcept C48044578 @default.
• W2768348081 hasConcept C70153297 @default.
• W2768348081 hasConcept C77088390 @default.
• W2768348081 hasConcept C84525736 @default.
• W2768348081 hasConceptScore W2768348081C11413529 @default.
• W2768348081 hasConceptScore W2768348081C119857082 @default.
• W2768348081 hasConceptScore W2768348081C124101348 @default.
• W2768348081 hasConceptScore W2768348081C138885662 @default.
• W2768348081 hasConceptScore W2768348081C154945302 @default.
• W2768348081 hasConceptScore W2768348081C169258074 @default.
• W2768348081 hasConceptScore W2768348081C199360897 @default.
• W2768348081 hasConceptScore W2768348081C26713055 @default.
• W2768348081 hasConceptScore W2768348081C2776401178 @default.
• W2768348081 hasConceptScore W2768348081C41008148 @default.
• W2768348081 hasConceptScore W2768348081C41895202 @default.
• W2768348081 hasConceptScore W2768348081C45374587 @default.
• W2768348081 hasConceptScore W2768348081C46686674 @default.
• W2768348081 hasConceptScore W2768348081C48044578 @default.
• W2768348081 hasConceptScore W2768348081C70153297 @default.
• W2768348081 hasConceptScore W2768348081C77088390 @default.
• W2768348081 hasConceptScore W2768348081C84525736 @default.
• W2768348081 hasOpenAccess W2768348081 @default.
• W2768348081 hasRelatedWork W1678356000 @default.
• W2768348081 hasRelatedWork W1988790447 @default.
• W2768348081 hasRelatedWork W2056132907 @default.
• W2768348081 hasRelatedWork W2064675550 @default.
• W2768348081 hasRelatedWork W2070493638 @default.
• W2768348081 hasRelatedWork W2101234009 @default.
• W2768348081 hasRelatedWork W2115584760 @default.

• next