haoqing-wang / cdfsl-ata Goto Github PK
View Code? Open in Web Editor NEW[IJCAI 2021 & AIJ 2023] Cross-Domain Few-Shot Classification via Adversarial Task Augmentation
[IJCAI 2021 & AIJ 2023] Cross-Domain Few-Shot Classification via Adversarial Task Augmentation
Nice work! However, I cannot reimplement the result in table 1 using the GNN method. I notice that your finetune.py document also covers the augmentation method. I finetuned the GNN checkpoint which was trained by step 1 and get very low accuracy using the finetune.py from https://github.com/IBM/cdfsl-benchmark. So can you provide me some detailes on how to get the GNN results in table 1. Also, I noticed that your code can not cover more shots(5way-20s or 5w-50s), since this may cost more CUDA memerory, have you solved this issue?
Hello, your work has inspired me a lot! I hope to further study and re-implement your result, but I found that using different pre-training models (specifically the "399.tar" in your code) had a great impact on the result. However, I used a variety of pre-training weights but did not get the result as in your paper. Could you please public pretrained model weights you used for different methods? Thanks a lot!
Hi, thanks for you sharing codes. Here I have a question about which model exactly you used when test in cross-domain few-shot dataset in table 1. For example, you use the one tested best in mini-test or you have another evaluation to choose the model which can generalize best in all this datasets? Looking forward to your response, thanks.
Hello, the dataset download links of cars, miniImagenet and planate in filelists->process.py are wrong, can you provide a right link? Many thanks.
hello, I tried to get plantae dataset by link you provided in process.py but the website it points responded with "404 not found".
Could you please provide a new link? Thanks!
Dear authors:
I find no supplementary material for the paper "Cross-Domain Few-Shot Classification via Adversarial Task Augmentation".
The conclusions in Lemma1 and Lemma 2 seem a little hard to deduce by myself. And the given references are full of too much distractors that I can't grasp the key content.
Could you please give me more explanation or some proof material about Lemma1 and Lemma 2?
A declarative, efficient, and flexible JavaScript library for building user interfaces.
๐ Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
An Open Source Machine Learning Framework for Everyone
The Web framework for perfectionists with deadlines.
A PHP framework for web artisans
Bring data to life with SVG, Canvas and HTML. ๐๐๐
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
Some thing interesting about web. New door for the world.
A server is a program made to process requests and deliver data to clients.
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
Some thing interesting about visualization, use data art
Some thing interesting about game, make everyone happy.
We are working to build community through open source technology. NB: members must have two-factor auth.
Open source projects and samples from Microsoft.
Google โค๏ธ Open Source for everyone.
Alibaba Open Source for everyone
Data-Driven Documents codes.
China tencent open source team.