Emotion model, apparatus, and method for adaptively modifying personality features of emotion model
View Patent ↗Disclosed are an emotion model, and an apparatus and method for adaptively learning personality of the emotion model. The emotion model, which maintains personality information, creates emotion information according to the personality information and takes a predetermined behavior according to the emotion information. The personality information may change adaptively in correspondence to a user's response to the behavior performed by the emotion model. Accordingly, the emotion model may react adaptively to the user through interactions with the user.
1. An emotion modeling system, comprising:
a processor configured to:
store personality information; and
create emotion information based on the personality information;
a motion selector configured to select motion information for deciding a motion, based on the emotion information;
a user's response receiver configured to receive a user's response to the motion corresponding to the motion information; and
a personality updating unit configured to:
create a learning parameter for the personality information using the user's response; and
update the personality information using the learning parameter, wherein
the learning parameter is inversely proportional to (1+e −sx ), where “x” is a value representing the intensity of the user's response and “s” is a value for adjusting the speed of learning.
2. The emotion modeling system of claim 1 , wherein:
the emotion creating unit is further configured to provide dependency information, the dependency information comprising a dependency relationship between the personality information and the emotion information, to the personality updating unit; and
at least one of the emotion creating unit and the motion selector is further configured to provide the emotion information to the personality updating unit.
3. The emotion modeling system of claim 2 , wherein the personality updating unit is further configured to:
select personality information to be updated, using the emotion information and the dependency information;
create a learning parameter for the selected personality information using the received user's response; and
update the selected personality information using the learning parameter.
4. The modeling system of claim 1 , wherein:
the user's response is classified as a positive response or a negative response of a particular intensity; and
the personality updating unit is further configured to classify the user's response as the positive response or the negative response with reference to a response classification tree.
5. The emotion modeling system of claim 1 , wherein:
the personality information comprises a personality characteristic and an intensity for the personality characteristic; and
the gain amount adjusts the intensity of the personality characteristic.
6. The emotion modeling system of claim 1 , wherein:
the learning parameter is derived from the relation:
G
(
x
)
=
a
1
+
ⅇ
-
s
·
x
·
(
-
1
)
r
,
(
x
≠
0
,
if
x
=
0
,
G
(
x
)
=
0
)
,
G(x) is the learning parameter, “x” is a negative or positive value representing the intensity of the user's response, “a” is a gain amount, and “r” is an even number if the user's response is positive and an odd number if the user's response is negative.
7. A personality learning apparatus for an emotion model, comprising:
a non-transitory first memory configured to store:
emotion information created based on personality information; and
a user's response to a motion corresponding to motion information selected based on the emotion information;
a second memory configured to store dependency information relating to a dependency relationship between the emotion information and the personality information; and
a learning controller configured to:
select personality information to be updated, using the emotion information and the dependency information,
generate a learning parameter for the selected personality information using the user's response; and
update the selected personality information using the learning parameter, wherein
the learning parameter is inversely proportional to (1+e −sx ), where “x” is a value representing the intensity of the user's response and “s” is a value for adjusting the speed of learning.
8. The personality learning apparatus of claim 7 , wherein:
the user's response is classified as a positive response or a negative response of a particular intensity; and
the second memory unit is further configured to store a response classification tree to classify the user's response.
9. The personality learning apparatus of claim 8 , wherein the learning controller is further configured to classify the user's response as the positive response or the negative response with reference to the response classification tree.
10. The personality learning apparatus of claim 7 , wherein:
the personality information comprises a personality characteristic and an intensity for the personality characteristic; and
the gain amount adjusts the intensity of the personality characteristic.
11. The personality learning apparatus of claim 7 , wherein:
the learning parameter is derived from the relation:
G
(
x
)
=
a
1
+
ⅇ
-
s
·
x
·
(
-
1
)
r
,
(
x
≠
0
,
if
x
=
0
,
G
(
x
)
=
0
)
,
G(x) is the learning parameter, “x” is a negative or positive value representing the intensity of the user's response, “a” is a gain amount, and “r” is an even number if the user's response is positive and an odd number if the user's response is negative.
12. A personality learning method which is performed by an emotion model, the method comprising:
creating, with a processor, emotion information based on personality information;
selecting motion information for deciding a motion, based on the emotion information;
receiving a user's response to the motion corresponding to the motion information; and
selecting personality information to be updated using the emotion information and dependency information, the dependency information relating to a dependency relationship between the personality information and the emotion information;
generating a learning parameter for the selected personality information using the user's response; and
updating the selected personality information using the learning parameter, wherein
the learning parameter is inversely proportional to (1+e −sx ), where “x” is a value representing the intensity of the user's response and “s” is a value for adjusting the speed of learning.
13. The personality learning method of claim 12 , wherein the updating of the personality information comprises classifying the user's response as a positive response or a negative response of a particular intensity.
14. The personality learning method of claim 12 , wherein:
the personality information comprises a personality characteristic and an intensity for the personality characteristic; and
the gain amount adjusts the intensity of the personality characteristic.
15. The personality learning method of claim 12 , wherein:
the learning parameter is derived from the relation:
G
(
x
)
=
a
1
+
ⅇ
-
s
·
x
·
(
-
1
)
r
,
(
x
≠
0
,
if
x
=
0
,
G
(
x
)
=
0
)
,
G(x) is the learning parameter, “x” is a negative or positive value representing the intensity of the user's response, “a” is a gain amount, and “r” is an even number if the user's response is positive and an odd number if the user's response is negative.
16. A personality learning method which is performed by an emotion model, the method comprising:
selecting, with a processor, a personality characteristic of personality information to be updated, using emotion information and dependency information relating to a dependency relationship between the emotion information and the personality information;
generating a learning parameter for the selected personality characteristic using a user's response; and
updating an intensity of the selected personality characteristic using the learning parameter, wherein
the learning parameter is inversely proportional to (1+e −sx ), where “x” is a value representing the intensity of the user's response and “s” is a value for adjusting the speed of learning.
17. The personality learning method of claim 16 , wherein the generating of the learning parameter comprises classifying the user's response as a positive response or a negative response of a particular intensity.
18. The personality learning method of claim 16 , wherein:
the learning parameter is derived from the relation:
G
(
x
)
=
a
1
+
ⅇ
-
s
·
x
·
(
-
1
)
r
,
(
x
≠
0
,
if
x
=
0
,
G
(
x
)
=
0
)
,
G(x) is the learning parameter, “x” is a negative or positive value representing the intensity of the user's response, “a” is a gain amount, and “r” is an even number if the user's response is positive and an odd number if the user's response is negative.