稀疏编码自编码表达
From Ufldl
(→Batching examples into mini-batches) |
|||
| Line 299: | Line 299: | ||
[原文] | [原文] | ||
| - | === Batching examples into mini-batches === | + | === Batching examples into mini-batches [将样本处理为“迷你块”]=== |
If you try running the simple iterative algorithm on a large dataset of say 10 000 patches at one go, you will find that each iteration takes a long time, and the algorithm may hence take a long time to converge. To increase the rate of convergence, you can instead run the algorithm on mini-batches instead. To do this, instead of running the algorithm on all 10 000 patches, in each iteration, select a mini-batch - a (different) random subset of say 2000 patches from the 10 000 patches - and run the algorithm on that mini-batch for the iteration instead. This accomplishes two things - firstly, it speeds up each iteration, since now each iteration is operating on 2000 rather than 10 000 patches; secondly, and more importantly, it increases the rate of convergence [[(TODO]]: explain why). | If you try running the simple iterative algorithm on a large dataset of say 10 000 patches at one go, you will find that each iteration takes a long time, and the algorithm may hence take a long time to converge. To increase the rate of convergence, you can instead run the algorithm on mini-batches instead. To do this, instead of running the algorithm on all 10 000 patches, in each iteration, select a mini-batch - a (different) random subset of say 2000 patches from the 10 000 patches - and run the algorithm on that mini-batch for the iteration instead. This accomplishes two things - firstly, it speeds up each iteration, since now each iteration is operating on 2000 rather than 10 000 patches; secondly, and more importantly, it increases the rate of convergence [[(TODO]]: explain why). | ||
| Line 305: | Line 305: | ||
[初译] | [初译] | ||
| + | 如果在一个大规模数据集（例如，10 000 patches）上运行上面描述的简单迭代算法，每次迭代都会耗费大量时间，因此算法收敛速度极慢。为了提高收敛速度，可以选择在迷你块上运行该算法。每次迭代中，选择一个迷你块代替10 000 patches 的大数据集合，然后在迷你块上运行算法。此处的迷你块是从10 000 patches中随机选取2000 patches。这样做有两点好处，一是加速了每次迭代，因为此时算法是在2000 patches而非10 000 patches上运行；更重要的是，加速了收敛速度。 | ||
[一审] | [一审] | ||
| + | |||
| + | 如果在一个大规模数据集（例如，10 000 patches）上运行上面描述的简单迭代算法，每次迭代都会耗费大量时间，因此算法收敛速度极慢。为了提高收敛速度，可以选择在迷你块上运行该算法。每次迭代中，选择一个迷你块代替10 000 patches 的大数据集合，然后在迷你块上运行算法。此处的迷你块是从10 000 patches中随机选取2000 patches。这样做有两点好处，一是加速了每次迭代，因为此时算法是在2000 patches而非10 000 patches上运行；更重要的是，加速了收敛速度。 | ||
[原文] | [原文] | ||
