Dataset_quality.md

This is quick evaluation of trainig set quality impact on performance on ImageNet-2012.

The architecture is similar to CaffeNet, but has differences:

1. Images are resized to small side = 128 for speed reasons.
2. fc6 and fc7 layers have 2048 neurons instead of 4096.
3. Networks are initialized with LSUV-init
4. No LRN layers.

Default augmentation: random crop 128x128 from 144xN image, 50% random horizontal flip.

Dataset quality

Name	Accuracy	LogLoss	Comments
Default, clean labels	0.471	2.36
5% incorrect labels	0.458	2.45
10% incorrect labels	0.447	2.58
15% incorrect labels	0.437	2.69
50% incorrect labels	0.347	3.44

logs