maozezhong / focal_loss_multi_class Goto Github PK
View Code? Open in Web Editor NEWmutil-class focal loss implemented in keras
mutil-class focal loss implemented in keras
You have next line in the code, which rewrites default alpha=0.25 param
alpha = array_ops.where(tf.greater(target_tensor, zeros), classes_weight, zeros)
balanced_fl = alpha * FT
Was it on purpose?
Hi, I want to know if this function can be directly applied in the multi-label task.. Should I make some changes or not ? thanks
firstly, you should get a list which contains each class number, like classes_nu=[1,2,3] means index_0 class have 1 pic, index_1 class have 1 pics.
then, use the focal loss function like below:
model.compile(optimizer=SGD(lr=learning_rate, momentum=0.9), loss=[focal_loss(classes_num)], metrics=['accuracy'])
dear maozezhong:
classes_nu=[1,2,3] means index_0 class have 1 pic, index_1 class have 1 pics.
or classes_nu=[1,2,3] means index_0 class have 1 pic, index_1 class have 2 pics, index_2 class have 3 pics?
Sparse_categorical_corssentorpy is fitable? Not categorical_corssentorpy!
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.