rosejn/cortex

support for selu activation function

Open

#181 opened on Jun 11, 2017

View on GitHub
 (8 comments) (0 reactions) (0 assignees)Clojure (1,269 stars) (109 forks)batch import
enhancementharderhelp wanted

Description

Ran into an interesting discussion on hackers' news on an activation function that seems to work well, even with deeper fully connected networks. See comparison and paper here, and discussion at reddit and HN.

https://github.com/shaohua0116/Activation-Visualization-Histogram

https://www.reddit.com/r/MachineLearning/comments/6g5tg1/r_selfnormalizing_neural_networks_improved_elu/ https://news.ycombinator.com/item?id=14527686

Contributor guide

Tech stack
None
Domain
machine learningai
Issue type
feature
DifficultyEstimated implementation difficulty for a new contributor, from 1 for very small changes to 5 for expert-level work.
3
Estimated timeA rough time range for an experienced contributor to investigate, implement, test, and prepare a pull request.
half day
Activity statusHow available the issue appears right now: fresh, active, stale, blocked, or waiting on maintainer input.
stale
ClarityHow clearly the issue explains the expected change, acceptance criteria, and next step.
mostly clear
Prerequisites
Clojure familiaritySELU activation understandingCortex codebase structure
Newbie friendlinessA 1-100 score estimating how approachable this issue is for first-time contributors.
60
Research direction
Look into the cortex repository's activation function implementations, likely in the `cortex/nn/activation.clj` file. Study how existing activations like ReLU or tanh are defined. Implement the SELU function following the same pattern, referencing the paper and code from other libraries. Add unit tests to verify correctness.