afagarap / cnn-svm Goto Github PK

Hello, I was wondering how do I train my own data set？

Thanks for this meaningful open source project！
Now that we use the MNIST dataset, I want to know how can I manipulate some pictures for model training?

For example, I have many pictures of "dogs", how do I use cnn-svm, thank you

This is the predicted output part of the svm network：
output = tf.identity(tf.sign(output), name='prediction')
correct_prediction = tf.equal(tf.argmax(output, 1), tf.argmax(y_input, 1))

After the output is processed by the function tf.sign()，a two-category tag that is processed into (1,-1).This is not predictable for multi-class handwritten data sets.
I want to ask you this question.

Hello i have tried to run your code but it is not able to create log directory. i have created manually and pass path of that but then also it is showing the same error.
"tensorflow.python.framework.errors_impl.InvalidArgumentError: Failed to create a directory: C:/Users/Anamika/Downloads/cnn-svm-master/cnn-svm-master/logsThu Aug 9 18:32:45 2018-training; Invalid argument"

In cnn-svm/model/cnn_svm.py , output of one versus rest svm is written as :
output = tf.identity(tf.sign(output), name='prediction')
correct_prediction = tf.equal(tf.argmax(output, 1), tf.argmax(y_input, 1))

tf.sign(output) seems incorrect. It should be removed. SVM here use max margin decision. When output margins of multiple binary linear SVM is converted to 1 and -1, tf.argmax(output, 1) may return a false value, because more than one binary linear SVM's output may be converted to 1 as they may all output a positive margin.

how to execute the code can you please me steps on how to solve it

Write a training function for the convolutional network model.

You are supposed to provide codes loading custo-made data

Hi,

I cant run the code with the following command:
python3 main.py --model 2 --dataset ./MNIST_data --penalty_parameter 1 --checkpoint_path ./checkpoint --log_path ./logs
I would get the error below:
"ModuleNotFoundError: No module named 'tensorflow.examples'"

How can I solve the Issue?

Hi, I am trying to run the code however I am getting this error.

InvalidArgumentError: You must feed a value for placeholder tensor 'input_4/x_input' with dtype float and shape [?,784]
[[{{node input_4/x_input}}]]

Implement the CNN-SVM in model/cnn_svm.py using TF2.

Implement the CNN model in model/cnn_softmax.py using TF2.

Add a PyTorch implementation of the baseline convolutional network.

Add a PyTorch implementation of the CNN-SVM model.

Why didn't you provide the hinge loss version?

• Implement a function for loading datasets
• Implement a function for creating a data loader.