Abstract | ||
---|---|---|
We propose ways to improve the performance of fully connected networks. We found that two approaches in particular have a strong effect on performance: linear bottleneck layers and unsupervised pre-training using autoencoders without hidden unit biases. We show how both approaches can be related to improving gradient flow and reducing sparsity in the network. We show that a fully connected network can yield approximately 70% classification accuracy on the permutation-invariant CIFAR-10 task, which is much higher than the current state-of-the-art. By adding deformations to the training data, the fully connected network achieves 78% accuracy, which is just 10% short of a decent convolutional network. |
Year | Venue | Field |
---|---|---|
2015 | arXiv: Learning | Training set,Bottleneck,Computer science,Convolution,Artificial intelligence,Balanced flow,Machine learning |
DocType | Volume | Citations |
Journal | abs/1511.02580 | 3 |
PageRank | References | Authors |
0.46 | 9 | 3 |
Name | Order | Citations | PageRank |
---|---|---|---|
Zhouhan Lin | 1 | 419 | 17.51 |
Roland Memisevic | 2 | 1116 | 65.87 |
Kishore Reddy Konda | 3 | 428 | 18.22 |