FRINK Linked Data Fragments server

Matches in SemOpenAlex for { <https://semopenalex.org/work/W2914811257> ?p ?o ?g. }

• W2914811257 abstract "Empirical studies show that gradient-based methods can learn deep neural networks (DNNs) with very good generalization performance in the over-parameterization regime, where DNNs can easily fit a random labeling of the training data. Very recently, a line of work explains in theory that with over-parameterization and proper random initialization, gradient-based methods can find the global minima of the training loss for DNNs. However, existing generalization error bounds are unable to explain the good generalization performance of over-parameterized DNNs. The major limitation of most existing generalization bounds is that they are based on uniform convergence and are independent of the training algorithm. In this work, we derive an algorithm-dependent generalization error bound for deep ReLU networks, and show that under certain assumptions on the data distribution, gradient descent (GD) with proper random initialization is able to train a sufficiently over-parameterized DNN to achieve arbitrarily small generalization error. Our work sheds light on explaining the good generalization performance of over-parameterized deep neural networks." @default.
• W2914811257 created "2019-02-21" @default.
• W2914811257 creator A5051448391 @default.
• W2914811257 creator A5086329948 @default.
• W2914811257 date "2019-02-04" @default.
• W2914811257 modified "2023-09-23" @default.
• W2914811257 title "Generalization Error Bounds of Gradient Descent for Learning Over-parameterized Deep ReLU Networks" @default.
• W2914811257 cites W1677182931 @default.
• W2914811257 cites W1944672 @default.
• W2914811257 cites W2123395972 @default.
• W2914811257 cites W2160815625 @default.
• W2914811257 cites W2163605009 @default.
• W2914811257 cites W2194775991 @default.
• W2914811257 cites W2208555118 @default.
• W2914811257 cites W2257979135 @default.
• W2914811257 cites W2399994860 @default.
• W2914811257 cites W2528305538 @default.
• W2914811257 cites W2563275261 @default.
• W2914811257 cites W2565538933 @default.
• W2914811257 cites W2579923771 @default.
• W2914811257 cites W2591714514 @default.
• W2914811257 cites W2593709294 @default.
• W2914811257 cites W2593958421 @default.
• W2914811257 cites W2614119628 @default.
• W2914811257 cites W2618003483 @default.
• W2914811257 cites W2739309726 @default.
• W2914811257 cites W2765428107 @default.
• W2914811257 cites W2766099245 @default.
• W2914811257 cites W2787132992 @default.
• W2914811257 cites W2788997738 @default.
• W2914811257 cites W2790253170 @default.
• W2914811257 cites W2796744096 @default.
• W2914811257 cites W2798986185 @default.
• W2914811257 cites W2806252860 @default.
• W2914811257 cites W2807299122 @default.
• W2914811257 cites W2808157231 @default.
• W2914811257 cites W2808958252 @default.
• W2914811257 cites W2809090039 @default.
• W2914811257 cites W2886067286 @default.
• W2914811257 cites W2891942459 @default.
• W2914811257 cites W2899476926 @default.
• W2914811257 cites W2899748887 @default.
• W2914811257 cites W2900959181 @default.
• W2914811257 cites W2911742574 @default.
• W2914811257 cites W2913473169 @default.
• W2914811257 cites W2949521215 @default.
• W2914811257 cites W2952469083 @default.
• W2914811257 cites W2962698540 @default.
• W2914811257 cites W2962726660 @default.
• W2914811257 cites W2962767131 @default.
• W2914811257 cites W2962810483 @default.
• W2914811257 cites W2962857907 @default.
• W2914811257 cites W2962930448 @default.
• W2914811257 cites W2962990163 @default.
• W2914811257 cites W2963095610 @default.
• W2914811257 cites W2963097630 @default.
• W2914811257 cites W2963100491 @default.
• W2914811257 cites W2963220074 @default.
• W2914811257 cites W2963236897 @default.
• W2914811257 cites W2963239103 @default.
• W2914811257 cites W2963285844 @default.
• W2914811257 cites W2963326517 @default.
• W2914811257 cites W2963376662 @default.
• W2914811257 cites W2963383839 @default.
• W2914811257 cites W2963417959 @default.
• W2914811257 cites W2963427613 @default.
• W2914811257 cites W2963446085 @default.
• W2914811257 cites W2963485221 @default.
• W2914811257 cites W2963509076 @default.
• W2914811257 cites W2963519230 @default.
• W2914811257 cites W2963569411 @default.
• W2914811257 cites W2963664410 @default.
• W2914811257 cites W2963739978 @default.
• W2914811257 cites W2963791871 @default.
• W2914811257 cites W2963794891 @default.
• W2914811257 cites W2963798163 @default.
• W2914811257 cites W2963826371 @default.
• W2914811257 cites W2963827833 @default.
• W2914811257 cites W2964010725 @default.
• W2914811257 cites W2964031251 @default.
• W2914811257 cites W2964034630 @default.
• W2914811257 cites W2964161337 @default.
• W2914811257 cites W2965497096 @default.
• W2914811257 cites W2965772785 @default.
• W2914811257 cites W2966530573 @default.
• W2914811257 cites W2970176397 @default.
• W2914811257 cites W2970330753 @default.
• W2914811257 cites W2970332347 @default.
• W2914811257 cites W2970443625 @default.
• W2914811257 cites W2970862128 @default.
• W2914811257 cites W2981407587 @default.
• W2914811257 cites W3104969455 @default.
• W2914811257 cites W3119586787 @default.
• W2914811257 cites W3137695714 @default.
• W2914811257 cites W607505555 @default.
• W2914811257 cites W813605148 @default.
• W2914811257 doi "https://doi.org/10.48550/arxiv.1902.01384" @default.
• W2914811257 hasPublicationYear "2019" @default.
• W2914811257 type Work @default.
• W2914811257 sameAs 2914811257 @default.

• next