UserPreferences

SRNModuleExperiments

Pyro SRN Experiments

This section will explore a series of experiments originally designed and run by Jeffery Elman.

First, we will generate a corpus of data of the form: 

dog walks .
dog who dogs who hit chase feeds .
John chases .
girl who dogs who feed live walks .
Mary lives .
boy sees .
Mary hits .
John hits .
cat sees .
girls who hear walk .
boy who hits feeds .
cat lives .
girls who hear hit .

http://bubo.brynmawr.edu/~dblank/startSmall/grammar-1.gif

http://bubo.brynmawr.edu/~dblank/startSmall/table1.gif

http://bubo.brynmawr.edu/~dblank/startSmall/network.gif

Results

Experiment #1: The network was trained on a corpus of sentences, and the it failed miserably. It learned some, but overall very poorly. Maybe Gold was right?

Experiment #2: The network was trained according to a strict schedule, 5 epochs each:

Phase 1 10,000 simple sentences 0 complex sentences
Phase 2 7,500 simple sentences 2,500 complex sentences
Phase 3 5,000 simple sentences 5,000 complex sentences
Phase 4 2,500 simple sentences 7,500 complex sentences
Phase 5 0 simple sentences 10,000 complex sentences

This worked! However, it was very ad hoc and probably took some fiddling by a graduate student to get it to work correctly. In addition, the environment was manipulated in a manner very much unlike the way that children learn language; they are exposed to it in all its complexity from early on. Could it be learned without such a strict manipulative schedule?

Experiment #3: The context bank of the network was randomly wiped out every 2 or 3 words with random patterns. The length between wipe-outs was slowly increased, until no wipe-outs were made.

This worked!

http://bubo.brynmawr.edu/~dblank/startSmall/fig2.gif

http://bubo.brynmawr.edu/~dblank/startSmall/fig3.gif

http://bubo.brynmawr.edu/~dblank/startSmall/fig4.gif

http://bubo.brynmawr.edu/~dblank/startSmall/fig5.gif

http://bubo.brynmawr.edu/~dblank/startSmall/hiddenspace.gif 

  1 
  2 
  3 
  4 
  5 
  6 
  7 
  8 
  9 
 10 
 11 
 12 
 13 
 14 
 15 
 16 
 17 
 18 
 19 
 20 
 21 
 22 
 23 
 24 
 25 
 26 
 27 
 28 
 29 
 30 
 31 
 32 
 33 
 34 
 35 
 36 
 37 
 38 
 39 
 40 
 41 
 42 
 43 
 44 
 45 
 46 
 47 
 48 
 49 
 50 
 51 
 52 
 53 
 54 
 55 
 56 
 57 
 58 
 59 
 60 
 61 
 62 


from pyrobot.brain.conx import SRN
import pyrobot.system.debug

#class mySRN(SRN):
#    def preStep(self):
#        print "pre step..."

network = SRN()
network.setSequenceType("random-continuous")

network.addLayer("input", 26)
network.addContextLayer("context", 70, "hidden")
network.addLayer("hid1", 10)
network.addLayer("hidden", 70)
network.addLayer("hid2", 10)
network.addLayer("output", 26)

network.connect("input", "hid1")
network.connect("hid1", "hidden")
network.connect("context", "hidden")
network.connect("hidden", "hid2")
network.connect("hid2", "output")

def makePattern(n):
    retval = [0.0] * 26
    retval[-n] = 1.0
    return retval

network.predict("input", "output")

network.setPatterns( {".": makePattern(1),
                      "boy": makePattern(2),
                      "girl": makePattern(3),
                      "cat": makePattern(4),
                      "dog": makePattern(5),
                      "chase": makePattern(6),
                      "feed": makePattern(7),
                      "see": makePattern(8),
                      "hear": makePattern(9),
                      "walk": makePattern(10),
                      "live": makePattern(11),
                      "boys": makePattern(12),
                      "girls": makePattern(13),
                      "cats": makePattern(14),
                      "dogs": makePattern(15),
                      "chases": makePattern(16),
                      "feeds": makePattern(17),
                      "sees": makePattern(18),
                      "hears": makePattern(19),
                      "walks": makePattern(20),
                      "lives": makePattern(21),
                      "John": makePattern(22),
                      "Mary": makePattern(23),
                      "hit": makePattern(24),
                      "hits": makePattern(25),
                      "who": makePattern(26) } )

network.loadInputPatternsFromFile("elman100.dat", checkEven=0)
#network.setInteractive(1)
network.setReportRate(1)
network.train()
[StartSmallProgram]Edit

Pyro Modules Table of Contents

Modules

  1. PyroModuleIntroduction

  2. PyroModuleObjectOverview

  3. PyroModulePythonIntro

  4. PyroModuleDirectControl

  5. PyroModuleSequencingControl

  6. PyroModuleBehaviorBasedControl

  7. PyroModuleReinforcementLearning

  8. PyroModuleNeuralNetworks

  9. PyroModuleEvolutionaryAlgorithms

  10. PyroModuleComputerVision

  11. PyroModuleMapping

  12. PyroModuleMultirobot

  13. FurtherReading

Additional Resources

  1. PyroIndex

  2. PyroAdvancedTopics

  3. PyroUserManual

  4. [WWW]Pyro Tutorial Movies

Reference: PyroSiteNotes

[PyroModulesContents]Edit