How do I know and tune the batch size while using patternnet for training a neural network classification model

Question

Siddarthan CS 2021-10-9

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/1470511-how-do-i-know-and-tune-the-batch-size-while-using-patternnet-for-training-a-neural-network-classific

回答： Sahil Jain 2021-10-19

So I tried using the patternet to classify the type of wines using the inbult wines dataset. I was wondering, if there was an option to tune the parameter such as batch size and etc? I know how to change the training fucntion and cost function but I'm unable to figure out how to tweak or tune the batch size. Any help would be appreciated.

0 个评论
显示 -2更早的评论隐藏 -2更早的评论

请先登录，再进行评论。

请先登录，再回答此问题。

Answer 1

Sahil Jain 2021-10-19

0
链接

此回答的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/1470511-how-do-i-know-and-tune-the-batch-size-while-using-patternnet-for-training-a-neural-network-classific#answer_811633

Hi. As of 2021b, shallow networks such as "patternnet" do not support minibatch training. The training function "traingd" uses the batch gradient descent algorithm and uses the entire dataset for each epoch. For minibatch training, consider using a deep neural network with the "trainNetwork" function. The Create Simple Image Classification Network documentation page will help you get started.