Tensorflow Ruby API

on YouTube ↗42m

" summary

Built a Ruby binding that calls the TensorFlow C API (wrapped with SWIG) and uses protocol buffers to load models trained elsewhere, so saved Python models run unchanged in Ruby. Demonstrates running Inception v3 on images (golden retriever 85% confidence, suspension bridge 89%), and visualizing embeddings with PCA and t-SNE via TensorBoard. Repository is available but not published to rubygems yet; works end-to-end though ongoing maintenance and TF 2.x compatibility remain open.

AI-written summary · about

" chapters (11)

1. 0:00Introduction and background
2. 2:17What is TensorFlow
3. 9:00TensorFlow timeline and features
4. 15:27Why build TensorFlow Ruby API
5. 18:01Building the Ruby API
6. 22:57Protocol buffers in TensorFlow
7. 25:02Ruby API demo and code walkthrough
8. 30:00Running ImageNet model in Ruby
9. 34:36Results and learning resources
10. 35:41Community and gratitude
11. 37:06Q&A session

" related talks

• 36mIntroducing Tensorflow Ruby APIArafat Dad Khan · EuRuKo 2017
• 29mDiscover Machine Learning in RubyJustin Bowen · RubyConf 2022
• 23mDeep Learning with RPaige Bailey · rstudio::conf 2019
• 26mDetecting and classifying object images using rubyFabio Leandro Janiszevski · RubyConf 2024
• 24mTensorFlow.js + NodeNick Kreeger, Yannick Assogba · EmpireJS 2018
• 52mPractical Uses of Scala Tensorflow for Image AnalysisXavier Tordoir · ScalaIO 2019
• 42mDown the Wabbit HoleChris McKinlay · Scale By The Bay 2018
• 04mTensorFlow and GoNatalie Pistunovich · dotGo 2019