Dream to Learn is shutting down...

We are very sorry to say that Dream to Learn will be shutting down as of December 28th, 2019. If you have content that you wish to keep, you should make a copy of it before that date.


0COMMENTS0RECOMMENDS

Harry Potter Sorting Hat 2.0 – Brain Gain

06
POSTED IN: Cognitive Wingman

Harry Potter Sorting Hat 2.0 – Brain Gain

 

This article describes an idea for expanding the Harry Potter Sorting Hat hack, from verbal signal input, to also include EEG brain signals, to assign Harry Potter house.

 

Some background - a couple of years ago I built a Harry Potter sorting hat powered by IBM Watson Speech to Text and Natural Language Classifier

http://www.businessinsider.com/real-life-harry-potter-sorting-hat-uses-watson-2016-6

 

instructions for a simple version here:

https://dreamtolearn.com/ryan/r_journey_to_watson/49

 

the ground truth for the NLC is here:

https://github.com/rustyoldrake/Harry_Potter_Sorting_Hat_Simple

 

And lately, I’ve been experimenting lately with EEG waves to move things and do things. Here, I’m driving around a hacked electric wheelchair

https://dreamtolearn.com/ryan/cognitivewingman/5

I was inspired by my colleague Josh Carr who did a similar thing with a BB8 droid

https://developer.ibm.com/recipes/tutorials/control-spherobb8-by-voice-through-ibm-watson-iot-platform/

 

HYPOTHESIS: 

I believe that the Emotive EEG 14 channel headset can inject additional signal into the Harry Potter Sorting Hat prototype (voice + EEG) but can operate on EEG only to produce reliable and reproducible sorts.

 

TEST:

The test for this is repeatable, blind-test able hat interpretation of thought signals.

  • Aggression, rage, anger (Slytherin)
  • Pride, Joy, Nobility (Gryffindor)  
  • Cerebral, thoughtfulness, cognitive load like math problems (Ravenclaw)
  • (for now I’m going to leave Hufflepuff out as a low intensity signal, or perhaps as default)
  • To begin, just on 1 subject.

 

 

DEVELOPING:

Anyway, time permitting, I’m going to start building this in April 2017.. will see how it goes.

 

 

Key Learnings:

 

  • Blunt force (non-EEG) signals, which I suspect are largely muscular, are available out of the box and need almost no training by user.   For example, for ‘four coordinate’ control  for forward, backward, left and right – you might use ‘clench’; ‘smile’; look/blink left and look/blink right
  • EEG “thought driven” commands take a little bit of work, but they do work.  One of my colleagues imagined hitting a three point shot to win a basketball game.  That was a ‘go to’ memory that invoked both memory; positive emotion; and reliably lit up parts of his brain to move the needle
  • Training and tuning, according to online posts, are essential for good performance, so it’s unclear how well ‘just putting the hat on’ will work across varying people and brains
  • I’ve also been doing a little work in the “Autism” use case area – leveraging emotion and tone analysis for kids on the spectrum.  https://dreamtolearn.com/ryan/cognitivewingman/3  - so there’s a component from this work that may dovetail here
  • COST / HACK? Emotiv gear is cool, and really opened the door to consumer grade EEG access, but it has a few drawbacks.  It’s expensive, nearly $1000 for the 14 channel hardware and $300 for the 5 channel ‘dry’;  to get to the useful stuff it’s a $50 monthly subscription;  I’ve seen reviews and heard from peers, and from my own experience, it’s delicate and breaks;.  I’m pleased I bought it, but for technical folks considering a purchase,  I’d consider doing some research into hacking / building your own low fidelity version.  I’ve seen a few ‘guitar pick’ youtube videos.   A DIY approach will also allow greater understanding of the signals, amplification, etc.
  • FMRI (portable0 I think within 10 years, possibly 5, we’re going to see some very interesting developments in FMRI.  At CES this year, I saw a  booth that had a demo that was not much bigger than a motorcycle helmet

 

 

Interested in more content by this author?

About the Author

Ryan Anderson

Ryan Anderson

Hi! I like to play with data, analytics and hack around with robots and gadgets in my garage. Lately I've been learning about machine learning.

About this blog

Cognitive Wingman - Never Walk Alone

Created: December 20, 2016

English

This Blog Appears in

Up Next