AI gives your project superpowers (talk)

Frank Carey
Bots and Brains
Published in
2 min readApr 25, 2017

--

Drupalcon Baltimore

I’ll be presenting at DrupalCon Baltimore this year on how “Artificial Intelligence gives your project superpowers”, diving into what ‘normal’ developers should know about Artificial Intelligence and Machine Learning.

How AI gives your project superpowers — Drupalcon Talk

Here are some of the resources that attendees and others might find useful.

Quest:

How do you debug Machine Learning / Deep Learning? Tensorflow now has a debugger called tfdbg you could experiment with, but debugging a neural network can be quite a different experience than traditional development.

What is a common/popular (local) setup? What libs do I need? I recommend Python to begin with, then here are some good libraries to look at.

  • Tensorflow, but start with keras which is a higher-level way of creating networks with tensorflow.
  • Pandas — A library for data science, making it easier to deal with datasets in a similar way as R
  • Scikit-learn — A general machine learning library for python. Lots of examples on the project site.
  • Jupyter (aka iPython) Notebooks — Great way to cleanup and visualize your datasets using plots of graphs and images. Many tutorials use this.

Public examples/use cases/demos, etc. (?)

  • cs231n.stanford.edu — GREAT class, with videos of lectures and assignments all available online for free. Lectures are by Andrej Karpathy who’s a great presenter and really helps you to understand the material.
  • Udacity ML Nanodegree — non-free, but very high quality assuming it’s like the driverless car program.
  • http://www.deeplearningbook.org/ — it seems like this book is considered to be one of the best resources for Deep Learning.

Services and Companies mentioned:

Feel free to experiment! Almost all of these services provide free trials or even free tiers, so take advantage to try a few and find out what works for you.

  • Google Machine Learning APIs, including Vision, Speech-To-Text, and Translation.
  • Clarifai Vision APIs for both Images and Video.
  • Mollom spam filtering service.
  • IBM Watson has various APIs including sentiment analysis.
  • Heliocore provides consulting and Machine Learning development (note, I’m a founder)

Code and Examples:

  • My Yelp Kaggle Competition code predicts if restaurants are good for kids, expensive, have booze, etc.. all by just looking at the pictures people take. [Tensorflow]
  • My German Traffic Sign code predicts which one of 43 traffic signs it was given like 100 km/hr speed limits, rights of ways, hazard signs. [Tensorflow]

Learn More:

--

--

I live in the future. #AI #Robotics #Tech #Neuroscience If you do to, check out @BotsAndBrains and our NYC Meetup http://www.meetup.com/Bots-and-Brains-NYC/