EyeMoSt
EyeMoSt copied to clipboard
Cocofeat
【MICCAI 2023 Early Accept & MedIA submission】EyeMost "Reliable Multimodality Eye Disease Screening via Mixture of Student's t Distributions"
Dear author, Thank you for your work. I know you have processed the image data into a pkl file. But I didn't find the code or the processed data in...
Dear author, Thank you for your work, it's an outstanding contribution, but I have a problem when running your code.  The meaning of this error suggests that when reading...
First of all, thanks for your excellent paper "Confidence-aware multi-modality learning for eye disease screening". It added to the area uncertainty estimation good improvement. I have downloaded your code and...
Dear Author, when training on the GAMMA dataset, the model does not predict early glaucoma in the test set; it only predicts normal and late-stage glaucoma. Could you please let...
Thanks for your impressive work EyeMoSt and EyeMoSt+ in MICCAI2023 and MIA 2024. When I reproduced the work based on this code, I found that the calculated loss would be...
这篇论文的方法有种杀敌1000，自损1500的感觉。 1.首先TMC就是为了解决OOD才引入了证据学习到多视图融合。OOD的单视图检测能力并不弱。Intro中的OOD检测能力不行说法不成立。 2.TMC中为了处理视图融合，自带基于两视图不确定性融合的规则。intro中的没有考虑模态冲突说法不成立。 3.将TMC的多分类问题（多项式分布）强行视为回归问题，直接导致方法只能做个2分类。简单来说就是回归一个二分类的正确率，以此把分类建模为回归。TMC里本身就有把融合2个迪利克雷分布的方法。在这篇论文强行把单视图后验分布置换为NIG后，引入了混合学生分布。这种做法在分类问题上没什么意义。 这篇论文把要方法里本就不存在的缺陷说成了是缺陷，然后毫无依据的换了其他方法。
