DTU DeepLearning: exercise 6

时间：2019-10-18 16:57:03 阅读：68 评论：0 收藏：0 [点我收藏+]

Wenjing Dai 3:41 PM · 3:41 PM

Wenjing Dai 3:41 PM · 3:41 PM

Hi everyone, I‘m a little confused about 6.1_exercise_4, the relationship between network selection (FFNN, CNN, RNN) and input types (image, margin, shape, texture). As my understanding, CNN is a type of FFNN, and image classification can be conducted by CNN. What about the other three features? Does each feature has a special relation with a network structure? I just don‘t get the point. Thanks in advance for your help.

Today

new messages

Christian Kolding 8:19 AM · 8:19 AM

Christian Kolding 8:19 AM · 8:19 AM

Christian Kolding 8:19 AM

The way I see it is that you use the CNN to preprocess the image and save the output from that as features. Same for the RNN but with other features. Finally, you collect all the features and feed them through one or more linear layers.

Does batch_size have any affects in results quality? how to set the optimal batch size and number of iterations?

1. the batch size, a : the number of training data feed into neural network once.

2. the number of iterations, b : how many time we should feed into NN.

a*b = the number of training examples.

https://stats.stackexchange.com/questions/164876/tradeoff-batch-size-vs-number-of-iterations-to-train-a-neural-network

https://machinelearningmastery.com/how-to-control-the-speed-and-stability-of-training-neural-networks-with-gradient-descent-batch-size/

https://datascience.stackexchange.com/questions/12532/does-batch-size-in-keras-have-any-effects-in-results-quality

https://medium.com/@BorisAKnyazev/tutorial-on-graph-neural-networks-for-computer-vision-and-beyond-part-1-3d9fada3b80d

DTU DeepLearning: exercise 6

原文：https://www.cnblogs.com/dulun/p/11698471.html

踩

(0)

评论一句话评论（0）