Learning Layer

[mab85]

@laurence-lin I want to use the learning layer to solve other problem in image processing field so, can the output of this layer be figure out as 2d image?.

[laurence-lin]

Hello, it has been some time since I've practiced this tempotron model. I don't have my notebook behind so I'll answer based on my memory, excuse me if there are any mistakes. Tempotron output is binary at each output neurons, '1' if output larger than threshold and '0' if output smaller than threshold. As for image case, I'm just learning to work on image task, and I've seen that others transform 2D image into 1D vector before input to network, and output as 1D vector, finally transform back to 2D matrix. In this case, if each output neuron represent a class to determine, then* I think it is available to apply tempotron for 2D image task* (if output neuron is 1, means the input sample belong to that class). After all, when I was working on tempotron I've found that due to its output, I could only use it as classification tools but not regression. mab85 <[email protected]> 於 2018年12月7日 週五 下午5:02寫道：

…

@laurence-lin <https://github.com/laurence-lin> I want to use the learning layer to solve other problem in image processing field so, can the output of this layer be figure out as 2d image?. — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#1>, or mute the thread <https://github.com/notifications/unsubscribe-auth/ASbaScKdHly1SKhgYWec4GOHOEvyGOHRks5u2i65gaJpZM4ZICkk> .