Information Technology Reference
In-Depth Information
W(t) is the weight vector. It represents the amount of attention allocated to each
attribute at each moment.
3 Case Study—Fashion Design Decision Making
3.1 Background
Intelligent fashion design system has been widely studied in the past decade. The
main idea of the system is to take a piece of fashion or garment as a combination of
several design items. So by searching from the database to find different items then
combine, the system can generate many new style of garments. Thus it can help those
who are not good at drawing to design their own fashion by making choices with aid
by computer. The major problem of this kind of system is that fashion design is a
creative and personal issue, there is not a general accepted rule set for the style. In
order to tackle this problem, many researchers tend to adopt interactive genetic algo-
rithm [12, 13, 14]. Many progresses have been made in capturing the user's personal
style orientation. They can handle well the general rational fashion choice, but for the
irrational phenomena that happen quite often in fashion choice or the stochastic cha-
racteristics of the user's choices, due to the limitation of GA, they failed.
Multi-alternative decision field theory models the dynamic, stochastic decision
process in an uncertain environment. It can successfully accounts for many irrational
behaviors in human multi-alternative decision making processes, such as similarity
effect, attraction effect and comprise effect. But it suffers from that it usually allows
less alternatives and attributes, and besides, the deliberation process is really time-
consuming.
Thus it is reasonable to use the proposed framework, integrating GA searching al-
gorithm with MDFT to have a better personal fashion decision making model.
3.2 Digitization
In order to ease the process of computer system, we digitize the original garment
design picture into 10 design attributes, namely outline, waist line, length, collar,
collar depth, segmentation, sleeve, sleeve length, ornaments, symmetry. They are
used in GA searching. The 7 independent design items of them are used to find the
semantic mapping relation of the objective feature space to the subjective feeling
space. So a garment in the objective feature space will be expressed as a 7-tuple for
semantic mapping:
Garment
oi
( outline
i
, waist
i
, length
i
, collar
i
,sleeve
i
, ornaments
i
, symmetry
i
)
Where,
Outline
i
, Waist
i
, Collar
i
, Sleeve
i
, Ornaments
i
each stands for the respective type of
the i
th
garment;
Length
i
is the length of the garment;
Symmetry
i
stands for whether the garment is symmetry or not.
