1.1 --- a/thesis/cortex.org	Sun Mar 23 17:59:16 2014 -0400
     1.2 +++ b/thesis/cortex.org	Sun Mar 23 19:09:14 2014 -0400
     1.3 @@ -4,86 +4,64 @@
     1.4  #+description: Using embodied AI to facilitate Artificial Imagination.
     1.5  #+keywords: AI, clojure, embodiment
     1.6  
     1.7 --- show hand
     1.8 +* Embodiment is a critical component of Intelligence
     1.9  
    1.10 -* Embodiment is a critical component to Intelligence
    1.11 +** Recognizing actions in video is extremely difficult
    1.12 +   cat drinking, mimes, leaning, common sense
    1.13  
    1.14 -* To explore embodiment, we need a world, body, and senses
    1.15 +** Embodiment is the the right language for the job
    1.16  
    1.17 -* Because of Time, simulation is perferable to reality
    1.18 +   a new possibility for the question ``what is a chair?'' -- it's the
    1.19 +   feeling of your butt on something and your knees bent, with your
    1.20 +   back muscles and legs relaxed.
    1.21  
    1.22 -* Video game engines are a great starting point
    1.23 +** =CORTEX= is a system for exploring embodiment
    1.24  
    1.25 -* Bodies are composed of segments connected by joints
    1.26 +   Hand integration demo
    1.27  
    1.28 -* Eyes reuse standard video game components
    1.29 +** =CORTEX= solves recognition problems using empathy
    1.30 +   
    1.31 +   worm empathy demo
    1.32  
    1.33 -* Hearing is hard; =CORTEX= does it right
    1.34 +** Overview
    1.35  
    1.36 -* Touch uses hundreds of hair-like elements
    1.37 +* Building =CORTEX=
    1.38  
    1.39 -* Proprioception is the force that makes everything ``real''
    1.40 +** To explore embodiment, we need a world, body, and senses
    1.41  
    1.42 -* Muscles are both effectors and sensors
    1.43 +** Because of Time, simulation is perferable to reality
    1.44  
    1.45 -* =CORTEX= brings complex creatures to life!
    1.46 +** Video game engines are a great starting point
    1.47  
    1.48 -* =CORTEX= enables many possiblities for further research
    1.49 +** Bodies are composed of segments connected by joints
    1.50  
    1.51 -* =CORTEX= User Guide
    1.52 +** Eyes reuse standard video game components
    1.53 +
    1.54 +** Hearing is hard; =CORTEX= does it right
    1.55 +
    1.56 +** Touch uses hundreds of hair-like elements
    1.57 +
    1.58 +** Proprioception is the force that makes everything ``real''
    1.59 +
    1.60 +** Muscles are both effectors and sensors
    1.61 +
    1.62 +** =CORTEX= brings complex creatures to life!
    1.63 +
    1.64 +** =CORTEX= enables many possiblities for further research
    1.65  
    1.66  * Empathy in a simulated worm
    1.67  
    1.68 -* Embodiment factors action recognition into managable parts
    1.69 +** Embodiment factors action recognition into managable parts
    1.70  
    1.71 -* Action recognition is easy with a full gamut of senses
    1.72 +** Action recognition is easy with a full gamut of senses
    1.73  
    1.74 -* Digression: bootstrapping with multiple senses
    1.75 +** Digression: bootstrapping with multiple senses
    1.76  
    1.77 -* \Phi-space describes the worm's experiences
    1.78 +** \Phi-space describes the worm's experiences
    1.79  
    1.80 -* Empathy is the process of tracing though \Phi-space 
    1.81 +** Empathy is the process of tracing though \Phi-space 
    1.82    
    1.83 -* Efficient action recognition via empathy
    1.84 -
    1.85 -* Contributions
    1.86 -
    1.87 -
    1.88 -* Vision 
    1.89 -
    1.90 -  System for understanding what the actors in a video are doing --
    1.91 -  Action Recognition.
    1.92 -  
    1.93 -  Separate action recognition into three components:
    1.94 -
    1.95 -  - free play
    1.96 -  - embodied action predicates
    1.97 -  - model alignment 
    1.98 -  - sensory imagination
    1.99 -
   1.100 -* Steps 
   1.101 -  
   1.102 - - Build cortex, a simulated environment for sensate AI
   1.103 -   - solid bodies w/ joints
   1.104 -   - vision
   1.105 -   - touch
   1.106 -   - vision
   1.107 -   - hearing
   1.108 -   - proprioception
   1.109 -   - muscle contraction
   1.110 -
   1.111 - - Build experimental framework for worm-actions
   1.112 -  - embodied stream predicates
   1.113 -  - \phi-space 
   1.114 -  - \phi-scan
   1.115 -
   1.116 -* News
   1.117 -  
   1.118 -  Experimental results:
   1.119 -
   1.120 -  - \phi-space actually works very well for the worm!
   1.121 -  - self organizing touch map
   1.122 -
   1.123 +** Efficient action recognition via empathy
   1.124  
   1.125  * Contributions
   1.126    - Built =CORTEX=, a comprehensive platform for embodied AI
   1.127 @@ -92,9 +70,30 @@
   1.128    - created a novel concept for action recognition by using artificial
   1.129      imagination. 
   1.130  
   1.131 +* =CORTEX= User Guide
   1.132  
   1.133  
   1.134  
   1.135 +In the second half of the thesis I develop a computational model of
   1.136 +empathy, using =CORTEX= as a base. Empathy in this context is the
   1.137 +ability to observe another creature and infer what sorts of sensations
   1.138 +that creature is feeling. My empathy algorithm involves multiple
   1.139 +phases. First is free-play, where the creature moves around and gains
   1.140 +sensory experience. From this experience I construct a representation
   1.141 +of the creature's sensory state space, which I call \phi-space. Using
   1.142 +\phi-space, I construct an efficient function for enriching the
   1.143 +limited data that comes from observing another creature with a full
   1.144 +compliment of imagined sensory data based on previous experience. I
   1.145 +can then use the imagined sensory data to recognize what the observed
   1.146 +creature is doing and feeling, using straightforward embodied action
   1.147 +predicates. This is all demonstrated with using a simple worm-like
   1.148 +creature, and recognizing worm-actions based on limited data.
   1.149  
   1.150 +Embodied representation using multiple senses such as touch,
   1.151 +proprioception, and muscle tension turns out be be exceedingly
   1.152 +efficient at describing body-centered actions. It is the ``right
   1.153 +language for the job''. For example, it takes only around 5 lines of
   1.154 +LISP code to describe the action of ``curling'' using embodied
   1.155 +primitives. It takes about 8 lines to describe the seemingly
   1.156 +complicated action of wiggling.
   1.157  
   1.158 -