Matches in SemOpenAlex for { <https://semopenalex.org/work/W3089324491> ?p ?o ?g. }
- W3089324491 abstract "Normalization techniques are essential for accelerating the training and improving the generalization of deep neural networks (DNNs), and have successfully been used in various applications. This paper reviews and comments on the past, present and future of normalization methods in the context of DNN training. We provide a unified picture of the main motivation behind different approaches from the perspective of optimization, and present a taxonomy for understanding the similarities and differences between them. Specifically, we decompose the pipeline of the most representative normalizing activation methods into three components: the normalization area partitioning, normalization operation and normalization representation recovery. In doing so, we provide insight for designing new normalization technique. Finally, we discuss the current progress in understanding normalization methods, and provide a comprehensive review of the applications of normalization for particular tasks, in which it can effectively solve the key issues." @default.
- W3089324491 created "2020-10-01" @default.
- W3089324491 creator A5040027309 @default.
- W3089324491 creator A5061215231 @default.
- W3089324491 creator A5063481044 @default.
- W3089324491 creator A5066270537 @default.
- W3089324491 creator A5069879969 @default.
- W3089324491 creator A5082634513 @default.
- W3089324491 date "2020-09-27" @default.
- W3089324491 modified "2023-10-12" @default.
- W3089324491 title "Normalization Techniques in Training DNNs: Methodology, Analysis and Application" @default.
- W3089324491 cites W122483837 @default.
- W3089324491 cites W1483804921 @default.
- W3089324491 cites W1533861849 @default.
- W3089324491 cites W1581755290 @default.
- W3089324491 cites W1665214252 @default.
- W3089324491 cites W1677182931 @default.
- W3089324491 cites W1815076433 @default.
- W3089324491 cites W1836465849 @default.
- W3089324491 cites W1861492603 @default.
- W3089324491 cites W1877062207 @default.
- W3089324491 cites W196761320 @default.
- W3089324491 cites W2021661079 @default.
- W3089324491 cites W2049633694 @default.
- W3089324491 cites W2073459066 @default.
- W3089324491 cites W2084894614 @default.
- W3089324491 cites W2086953401 @default.
- W3089324491 cites W2095705004 @default.
- W3089324491 cites W2097117768 @default.
- W3089324491 cites W2097428361 @default.
- W3089324491 cites W2099471712 @default.
- W3089324491 cites W2100495367 @default.
- W3089324491 cites W2117539524 @default.
- W3089324491 cites W2125389028 @default.
- W3089324491 cites W2127230474 @default.
- W3089324491 cites W2139398462 @default.
- W3089324491 cites W2144513243 @default.
- W3089324491 cites W2146502635 @default.
- W3089324491 cites W2152424459 @default.
- W3089324491 cites W2155894447 @default.
- W3089324491 cites W2161388792 @default.
- W3089324491 cites W2163605009 @default.
- W3089324491 cites W2173520492 @default.
- W3089324491 cites W2183341477 @default.
- W3089324491 cites W2188333210 @default.
- W3089324491 cites W2190691619 @default.
- W3089324491 cites W2194775991 @default.
- W3089324491 cites W2237132896 @default.
- W3089324491 cites W2299668505 @default.
- W3089324491 cites W2302255633 @default.
- W3089324491 cites W2405756170 @default.
- W3089324491 cites W2475287302 @default.
- W3089324491 cites W2502312327 @default.
- W3089324491 cites W2533142838 @default.
- W3089324491 cites W2545656684 @default.
- W3089324491 cites W2546257475 @default.
- W3089324491 cites W2546302380 @default.
- W3089324491 cites W2548275288 @default.
- W3089324491 cites W2549139847 @default.
- W3089324491 cites W2551557006 @default.
- W3089324491 cites W2552391848 @default.
- W3089324491 cites W2556833785 @default.
- W3089324491 cites W2560023338 @default.
- W3089324491 cites W2560609797 @default.
- W3089324491 cites W2562512778 @default.
- W3089324491 cites W2564591810 @default.
- W3089324491 cites W2568343048 @default.
- W3089324491 cites W2581955877 @default.
- W3089324491 cites W2583938035 @default.
- W3089324491 cites W2594653239 @default.
- W3089324491 cites W2594833348 @default.
- W3089324491 cites W2603777577 @default.
- W3089324491 cites W2607512710 @default.
- W3089324491 cites W2614311475 @default.
- W3089324491 cites W2619441027 @default.
- W3089324491 cites W2626017178 @default.
- W3089324491 cites W2724651715 @default.
- W3089324491 cites W2737818580 @default.
- W3089324491 cites W2739741052 @default.
- W3089324491 cites W2742021527 @default.
- W3089324491 cites W2750933313 @default.
- W3089324491 cites W2757196798 @default.
- W3089324491 cites W2757910899 @default.
- W3089324491 cites W2762473905 @default.
- W3089324491 cites W2780149736 @default.
- W3089324491 cites W2785783085 @default.
- W3089324491 cites W2787911558 @default.
- W3089324491 cites W2789027062 @default.
- W3089324491 cites W2789134901 @default.
- W3089324491 cites W2789621689 @default.
- W3089324491 cites W2792287754 @default.
- W3089324491 cites W2794363191 @default.
- W3089324491 cites W2795358072 @default.
- W3089324491 cites W2795900505 @default.
- W3089324491 cites W2798405286 @default.
- W3089324491 cites W2804078698 @default.
- W3089324491 cites W2807205843 @default.
- W3089324491 cites W2889900721 @default.
- W3089324491 cites W2891172060 @default.
- W3089324491 cites W2892218381 @default.