alanchou / truncated-loss Goto Github PK
View Code? Open in Web Editor NEWPyTorch implementation of the paper "Generalized Cross Entropy Loss for Training Deep Neural Networks with Noisy Labels" in NIPS 2018
PyTorch implementation of the paper "Generalized Cross Entropy Loss for Training Deep Neural Networks with Noisy Labels" in NIPS 2018
the weight of Truncat loss correspond to the data in training dataset, rather than the test dataset. Should we reuse the critierion at the test time ? which may calculate the wrong loss.
Dear Alan,
I hope that you're doing well.
Thank you for sharing the open-source implementation of this impressive publication.
May I please ask what one should do in case of multi-label output - e.g. BCE is used?
Can you also maybe implement and share the state-of-the-art in this area, SuperLoss:
https://proceedings.neurips.cc//paper/2020/file/2cfa8f9e50e0f510ede9d12338a5f564-Paper.pdf
Thank you very much again. Sincerely,
what to do if the loss is binary cross entropy (torch.nn.BCEWithLogitsLoss) , it does not work
I tried testing the model both during the training time and testing, the model is not learning when I'm using the implementation in this repo. So the accuracy is not improving on the dataset.
loss = ((1-(Ygself.q))/self.q)*self.weight[indexes] - ((1-(self.kself.q))/self.q)*self.weight[indexes]
here the loss will quickly become negative.
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.