如何将多分类转换为二分类

[LiuShiWang]

hello！Thank you for the code you shared.This is very helpful to me. The blog you recommend is in the case of multi-category application， and your code is in the case of two classifications. The question is how to convert multiple classifications into two classifications？ Do you mind sharing your own data set? You are excellent, I want to learn from you, thank you again.

[jiangyiqiao]

您好，

1. 关于是如何将多分类转换为二分类,主要修改点为重定义网络的最后一层全连接输出层神经元个数为类别数 主要代码包括： train.py line 180： n_classes = len(image_lists.keys())

   获取分类类别长度len(["classification-1","classification-2"])

   line 203-207 # 定义一层全连接层解决新的图片分类问题 with tf.name_scope('final_training_ops'): weights = tf.Variable(tf.truncated_normal([BOTTLENECK_TENSOR_SIZE, n_classes], stddev=0.1)) 2.关于训练数据集，因为涉及实验室的数据，不方便给出请见谅~ 3.我也是新人，之前做过一段时间的图像，都是那开源的学习，最近在做NLP类的东西，可惜都是业务上的东西，很久没接触神经网络的东西了，希望你能找的合适的值得学习的人。 祝好~

On Mon, 13 Aug 2018 at 21:26, LiuShiWang notifications@github.com wrote:

hello！Thank you for the code you shared.This is very helpful to me. The blog you recommend is in the case of multi-category application， and your code is in the case of two classifications. The question is how to convert multiple classifications into two classifications？ Do you mind sharing your own data set? You are excellent, I want to learn from you, thank you again.

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/jiangyiqiao/fundus_transfer_v3_classification/issues/2, or mute the thread https://github.com/notifications/unsubscribe-auth/ASuJ3_WX0teCQ99-YrqJ54viqOaMgOgOks5uQX6igaJpZM4V6gKY .

[LiuShiWang]

谢谢你，YiQiao。我已经成功的运行代码，并且精度有较好的表现。 但是那个精度曲线图（accuracy，loss）我无法通过tensorboard --logdir=./runs/1521208186/summaries/train进行下载，是域外访问的限制吗？还是什么原因？ 我之前是做激光测量的，偏硬件，编程基础比较薄弱，最近这个月才开始接触编程语言，转学图像，经常是一脸懵bi。谢谢你给予我的帮助。

[jiangyiqiao]

hello，我也不知道为什么你的tensorbord无法查看

1. 可以查看路径是否正确； 2.域名都是本地的测试，没啥限制吧... 3.我学习也是蒙，加油啦~

On Tue, 14 Aug 2018 at 13:49, Liushiwang notifications@github.com wrote:

谢谢你，YiQiao。我已经成功的运行代码，并且精度有较好的表现。 但是那个精度曲线图（accuracy，loss）我无法通过tensorboard --logdir=./runs/1521208186/summaries/train进行下载，是域外访问的限制吗？还是什么原因？ 我之前是做激光测量的，偏硬件，编程基础比较薄弱，最近这个月才开始接触编程语言，转学图像，经常是一脸懵bi。谢谢你给予我的帮助。

— You are receiving this because you commented. Reply to this email directly, view it on GitHub https://github.com/jiangyiqiao/fundus_transfer_v3_classification/issues/2#issuecomment-412761648, or mute the thread https://github.com/notifications/unsubscribe-auth/ASuJ32aMA433ShveRDh17kZfJZHdSo5Mks5uQmTZgaJpZM4V6gKY .

[LiuShiWang]

谢谢你，昨晚收到邮件，我都快睡着了，xidada说年轻人要少熬夜！要注意休息a！