submissionsin / mflvc Goto Github PK

Hi, could you please upload the visualization code by t-sne? I try to reproduce the figure2 in the paper, but I cannot get the correct output.

Hi author!
Could you share your codes?

Thanks

I want to know the calculation process of cost matrix M in semantic clustering with high-level features. I didn't understand the formula you wrote in your paper. There is also the specific meaning of bool matrix A. Thank you for answering my questions.

Hi, thanks for your great job and well-written paper!

I wanna know which view of the MNIST-USPS to be chosen for the t-SNE vis in Figure 2? Or the whole view in an average way?

could you please tell how does the mask works? I find it hard to understand it. If you could answer it, I would appreciate it very much. Thank U for your kindness!

Can you please tell me the effect of "math.log(p_i.size(0))" has on line 49 in "loss.py" and 'p_i /= p_i.sum()' has on line 48 in "loss.py"?

Could you tell me the reason why you choose just a single linear layer as the body of the Feature MLP?Whether a more complicated MLP may cause bad effect?Thank you for answering my question.

There might be an error in the 85-th line on metric.py, it should include in the 'for' loop.

Hi,

I'm unable to access the data and model from baidu link, is it possible for you to upload the repository on any other cloud, I can not even create an account on Baidu without a Chinese number to access Baidu Netdisk. Also could you please explain what data pre-processing /feature extraction techniques have you applied to create these npy or mat files.