Methods and systems for budgeted and simplified training of deep neural networks
Methods and systems for budgeted and simplified training of deep neural networks (DNNs) are disclosed. In one example, a trainer is to train a DNN using a plurality of training sub-images derived from a down-sampled training image. A tester is to test the trained DNN using a plurality of testing sub-images derived from a down-sampled testing image. In another example, in a recurrent deep Q-network (RDQN) having a local attention mechanism located between a convolutional neural network (CNN) and a long-short time memory (LSTM), a plurality of feature maps are generated by the CNN from an input image. Hard-attention is applied by the local attention mechanism to the generated plurality of feature maps by selecting a subset of the generated feature maps. Soft attention is applied by the local attention mechanism to the selected subset of generated feature maps by providing weights to the selected subset of generated feature maps in obtaining weighted feature maps. The weighted feature maps are stored in the LSTM. A Q value is calculated for different actions based on the weighted feature maps stored in the LSTM.
1. A method for a deep neural network (DNN), comprising:
sub-sampling a training image into a plurality of training sub-images;
dividing the plurality of training sub-images into batches of training sub-images, each of the batches including one or more of the plurality of training sub-images; and
training the DNN with randomly selected batches of training sub-images, including:
randomly selecting a first batch of the training sub-images,
processing the randomly selected first batch of training sub-images with the DNN,
adjusting weights of the DNN based on classification results determined from processing the randomly selected first batch of training sub-images,
randomly selecting a second batch of the training sub-images,
processing the randomly selected second batch of training sub-images with the DNN, and
adjusting the weights of the DNN based on second classification results determined from processing the randomly selected second batch of training sub-images.
2. The method of claim 1 , further comprising:
sub-sampling a testing image into a plurality of testing sub-images;
randomly selecting a first testing sub-image; and
testing the trained DNN with the randomly selected first testing sub-image to obtain a first testing result.
3. The method of claim 2 , further comprising:
randomly selecting a second testing sub-image;
testing the trained DNN with the randomly selected second testing sub-image to obtain a second testing result; and
selecting one of the first testing result and second testing result as a final result.
4. The method of claim 2 , wherein each of the plurality of testing sub-images has a smaller resolution than the testing image.
5. The method of claim 1 , wherein each of the plurality of training sub-images has a smaller resolution than the training image.
6. A system for a server comprising:
a processing core having a deep neural network (DNN);
an I/O controller hub coupled to the processing core and to provide network, data storage, and DNN access for the processing core; and
a graphics processor coupled to the I/O controller hub, the graphics processor to:
train the DNN with randomly selected batches of training sub-images, including to:
randomly select a first batch of the training sub-images,
process the randomly selected first batch of training sub-images with the DNN,
adjust weights of the DNN based on classification results determined from processing the randomly selected first batch of training sub-images,
randomly select a second batch of the training sub-images,
process the randomly selected second batch of training sub-images with the DNN, and
adjust the weights of the DNN based on second classification results determined from processing the randomly selected second batch of training sub-images, and
test the trained DNN using a plurality of testing sub-images derived from a down-sampled testing image.
7. The system of claim 6 , wherein the graphics processor is to:
sub-sample a testing image into a plurality of testing sub-images,
randomly select a first testing sub-image, and
test the trained DNN with the randomly selected first testing sub-image to obtain a first testing result.
8. The system of claim 7 , wherein the graphics processor is to
randomly select a second testing sub-image,
test the trained DNN with the randomly selected second testing sub-image to obtain a second testing result, and
selecting one of the first testing result and second testing result as a final result.
9. The system of claim 7 , wherein each testing sub-image has a smaller resolution than the testing image.
10. The system of claim 6 , wherein each training sub-image has a smaller resolution than the training image.
11. A non-transitory computer-readable medium having instructions stored thereon, which when executed by one or more processors, cause the one or more processors to perform a method for a deep neural network (DNN), the method comprising:
sub-sampling a training image into a plurality of training sub-images;
dividing the plurality of training sub-images into batches of training sub-images, each of the batches including one or more of the plurality of training sub-images; and
training the DNN with randomly selected batches of training sub-images, including;
randomly selecting a first batch of the training sub-images,
processing the randomly selected first batch of training sub-images with the DNN,
adjusting weights of the DNN based on classification results determined from processing the randomly selected first batch of training sub-images,
randomly selecting a second batch of the training sub-images,
processing the randomly selected second batch of training sub-images with the DNN, and
adjusting the weights of the DNN based on second classification results determined from processing the randomly selected second batch of training sub-images.
12. The non-transitory computer-readable medium of claim 11 , wherein the method further comprises:
sub-sampling a testing image into a plurality of testing sub-images;
randomly selecting a first testing sub-image; and
testing the trained DNN with the randomly selected first testing sub-image to obtain a first testing result.
13. The non-transitory computer-readable medium of claim 12 , wherein the method further comprises:
randomly selecting a second testing sub-image;
testing the trained DNN with the randomly selected second testing sub-image to obtain a second testing result; and
selecting one of the first testing result and second testing result as a final result.
14. The non-transitory computer-readable medium of claim 12 , wherein:
each of the plurality of testing sub-images has a smaller resolution than the testing image.
15. The non-transitory computer-readable medium of claim 11 , wherein:
each of the plurality of training sub-images has a smaller resolution than the training image.