2 added 93 characters in body
source | link

In https://arxiv.org/pdf/1809.09529.pdf it is said

If the new dataset is similar to original dataset, we expect higher-level features in the CNN to be relevant to this dataset. Thus, it is recommended to train a linear classifier on the CNN codes. Otherwise, if the new dataset is large then we might try to fine-tune through the full network. However if the new dataset is different from the original dataset and it has a small size, then it might work better to train a linear classifier from activations somewhere earlier in the network.

Fine tuning is quite popular in image classification problems where you take a model pre-trained on ImageNet and fine tune it to your dataset. But ImageNet classifies images on a very broad spectrum (1000 classes), so what does similar and different even mean precisely? How do I know if my dataset is similar to such a big and broad dataset? What does that mean?

In https://arxiv.org/pdf/1809.09529.pdf it is said

If the new dataset is similar to original dataset, we expect higher-level features in the CNN to be relevant to this dataset. Thus, it is recommended to train a linear classifier on the CNN codes. Otherwise, if the new dataset is large then we might try to fine-tune through the full network. However if the new dataset is different from the original dataset and it has a small size, then it might work better to train a linear classifier from activations somewhere earlier in the network.

Fine tuning is quite popular in image classification problems where you take a model pre-trained on ImageNet and fine tune it to your dataset. But ImageNet classifies images on a very broad spectrum (1000 classes), so what does similar and different even mean precisely?

In https://arxiv.org/pdf/1809.09529.pdf it is said

If the new dataset is similar to original dataset, we expect higher-level features in the CNN to be relevant to this dataset. Thus, it is recommended to train a linear classifier on the CNN codes. Otherwise, if the new dataset is large then we might try to fine-tune through the full network. However if the new dataset is different from the original dataset and it has a small size, then it might work better to train a linear classifier from activations somewhere earlier in the network.

Fine tuning is quite popular in image classification problems where you take a model pre-trained on ImageNet and fine tune it to your dataset. But ImageNet classifies images on a very broad spectrum (1000 classes), so what does similar and different even mean precisely? How do I know if my dataset is similar to such a big and broad dataset? What does that mean?

1
source | link

What does a "similar" dataset mean in the context of fine tuning a CNN?

In https://arxiv.org/pdf/1809.09529.pdf it is said

If the new dataset is similar to original dataset, we expect higher-level features in the CNN to be relevant to this dataset. Thus, it is recommended to train a linear classifier on the CNN codes. Otherwise, if the new dataset is large then we might try to fine-tune through the full network. However if the new dataset is different from the original dataset and it has a small size, then it might work better to train a linear classifier from activations somewhere earlier in the network.

Fine tuning is quite popular in image classification problems where you take a model pre-trained on ImageNet and fine tune it to your dataset. But ImageNet classifies images on a very broad spectrum (1000 classes), so what does similar and different even mean precisely?