and Paths in Continuous Spaces Jacob Andreas and Dan Klein UC Berkeley B erkeley N L P Formal grounding On June 26 th Facebook stock cost 65 per share quote date 20140626 ID: 788743
Download The PPT/PDF document "Grounding Language with Points" is the property of its rightful owner. Permission is granted to download and print the materials on this web site for personal, non-commercial use only, and to display it on your personal computer provided you do not modify the materials and that you retain all copyright notices contained in the materials. By downloading content from our website, you accept the terms of this agreement.
Slide1
Grounding Language with Points and Paths in Continuous Spaces
Jacob Andreas and Dan KleinUC Berkeley
B
erkeley
N L P
Slide2Formal groundingOn June 26
th, Facebook stock cost $65 per share
quote {
date: 2014-06-26, stock: FB,
price: $65}
Slide3Perceptual groundingOn June 26
th, Facebook stock reboundedafter a bruising swoon
?
Slide4Perceptual groundingOn June 26
th, Facebook stock reboundedafter a bruising swoon
Slide5Perceptual groundingOn June 26
th, Facebook stock reboundedafter a bruising swoon
A after
B A, B
A before B
B, A rebounded
{ sgn(slope) = +1 } bruising
{ sgn(slope) = -1,
abs(slope) = +2.3 }
Slide6Continuous spaces everywhere
On June 26th, Facebook stock reboundedafter a bruising swoonA deep red sunset
Keep a little to the left of the post
Beat the eggs gently, until they form stiff peaks
Slide7Three tasksColor
Time series
Navigation
Slide8Predicting colors
blue
p
astel blue
d
ark pastel blue
H
V
S
Slide9Regression model
blue
p
astel blue
d
ark pastel blue
H
V
S
Slide10Regression model
H 216
S 43
V 75
dark
pastel
blue
H
S
V
0
0
-40
0
-37
25
216
80
90
Slide11Regression model
darkpastel
blue
0
0
-40
0
-37
-25
216
80
90
H
S
V
216
43
75
+
+
=
Slide12Regression modeldark
pastel
blue
Slide13dark pastel blue
{dark, pastel, blue}
Regression model
H 216
S 43
V 75
Slide14Experiment setup
Slide15Sample predictions
e
lectric green
p
ale blue
d
ark brown
p
ale green
indigo
Slide16Prediction error
Slide17A guessing game
p
ale blue
Slide18A guessing game
Slide19Predicting time series
stocks rebounded
a
fter a bruising swoon
1
2
2
1
Slide20Predicting time series
stocks rebounded
a
fter a bruising swoon
Slide21Predicting time series
stocks rebounded
a
fter a bruising swoon
{
stocks, rebounded}
{after, a, bruising, swoon}
2
1
sgn
(slope):
-1
abs(slope):
3.1
curvature: 0.5
sgn
(slope):
1
abs(slope):
2.7
curvature: -0.1
Slide22Learning & inferenceNeed parameters for linear prediction model & log-linear alignment model: easy with EM
For small number of path segments, possible to sum exactly over latent alignmentsOtherwise, approximation of your choice
Slide23Experiment setup
Market rallies
to new highs
Slide24Sample predictions
Reference
Predicted
U.S. stocks end lower
as economic worries persist
[
U.S. stocks end lower
]
2
[
as economic worries persist
]
1
Slide25A guessing game
Slide26Peeking at parameters
sgn
(slope)
abs(slope)
rise
swoon
sharply
0.27
-0.57
-0.22
-0.78
0
0.28
Slide27Following instructions…
and then we're going to turn north againand immediat-- well a distance below that turning point there's a fenced meadowbut you should be avoiding that by quite a distance
okay so we've turned and we're going up north againcontinue straight up north
and then we're going to turn to the west on a curvature right sort of…
Slide28Navigation results
Slide29ConclusionsNew model for predicting grounded representations of meaning in arbitrary real-valued spaces
Beats strong baselines on a diverse range of tasksCode and data available online athttp://
cs.berkeley.edu
/~jda