GithubHelp home page GithubHelp logo

airtai / monotonic-nn Goto Github PK

View Code? Open in Web Editor NEW
21.0 4.0 2.0 20.85 MB

Keras implementation of the constrained monotonic neural networks

Home Page: https://monotonic.airt.ai

License: Other

Jupyter Notebook 97.75% Python 2.10% CSS 0.06% JavaScript 0.02% HTML 0.06%
keras monotonic neural-networks tensorflow

monotonic-nn's People

Contributors

davorrunje avatar harishmohanraj avatar sharath-sms avatar

Stargazers

avatar avatar avatar avatar avatar avatar avatar avatar avatar avatar avatar avatar avatar avatar avatar avatar avatar avatar avatar

Watchers

avatar avatar avatar avatar

Forkers

freekang ericxue92

monotonic-nn's Issues

Keras saving

Can you let me know how to save a model with monodense implementation?

How can this work with time-series NN ?

Thanks for the great work! It is very useful for me. But I wander how it can be used in the RNN structure or Transformer model? I am doing a time-series analysis. Thanks!

Question about saturated activation

Hi! I was skimming through the source code and I am unsure about the computation for the saturated activation function.

In the function saturated_activation we can see that the following line is executed:
(

monotonic-nn/airt/_components/mono_dense_layer.py

Line 37 in ffbcf63

ccc = concave_activation(-tf.ones_like(x) * c)
)

However, ccc is never used in the remaining of the function, as the return statement uses the cc variable but never ccc:

monotonic-nn/airt/_components/mono_dense_layer.py

Lines 38 to 42 in ffbcf63

return a * tf.where(
x <= 0,
convex_activation(x + c) - cc,
concave_activation(x - c) + cc,
)

Could this be a bug? Otherwise, the computation for ccc is not required and can be ommited.

Thank you for the paper and your effort in the library!

Recommend Projects

  • React photo React

    A declarative, efficient, and flexible JavaScript library for building user interfaces.

  • Vue.js photo Vue.js

    🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.

  • Typescript photo Typescript

    TypeScript is a superset of JavaScript that compiles to clean JavaScript output.

  • TensorFlow photo TensorFlow

    An Open Source Machine Learning Framework for Everyone

  • Django photo Django

    The Web framework for perfectionists with deadlines.

  • D3 photo D3

    Bring data to life with SVG, Canvas and HTML. 📊📈🎉

Recommend Topics

  • javascript

    JavaScript (JS) is a lightweight interpreted programming language with first-class functions.

  • web

    Some thing interesting about web. New door for the world.

  • server

    A server is a program made to process requests and deliver data to clients.

  • Machine learning

    Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.

  • Game

    Some thing interesting about game, make everyone happy.

Recommend Org

  • Facebook photo Facebook

    We are working to build community through open source technology. NB: members must have two-factor auth.

  • Microsoft photo Microsoft

    Open source projects and samples from Microsoft.

  • Google photo Google

    Google ❤️ Open Source for everyone.

  • D3 photo D3

    Data-Driven Documents codes.