4.4. EXPERIMENT 41
regarding the rule confidence learned by the attention mechanism. As can be seen, different lev-
els of rule confidence can be assigned for the same rule (“no silk C knit”) with different triplets.
In addition, we found that the rules pertaining to the category are usually assigned higher con-
fidence levels. is may be attributed to that people tend to put the category attribute at the
first place when they make outfits compared to other attributes. Furthermore, we also noted
that although the contextual metadata indicates that the third triplet activates the rule “stripe
C stripe,” the learned rule confidence is not much high. is may be due to the fact that the
given rule is a bit fuzzy and general and the visual signals imply the incompatibility between the
stripes in the given top i and bottom k. Accordingly, to certain extent, the attention mechanism
can be helpful to overcome the limitation of the human-defined fuzzy rules.
Silk Top Knit Shorts Chiffon Skirt
Silk Blouse Skinny Jeans Knit Dress
Stripe
Tank Top
i
Denim
Short
j
Striped Silk
Maxi Skirt
k
(1)
(2)
(3)
Activated Rules λ
No silk + knit 0.22
No silk + chiffon 0.78
Activated Rules λ
No blouse + dress 0.66
No silk + knit 0.34
Activated Rules λ
Tank Top + short 0.63
Stripe + Stripe 0.37
Figure 4.6: Illustration of attentive rule confidences.
4.4.4 ON FASHION ITEM RETRIEVAL (RQ3)
To assess the practical value of our work, we evaluated the proposed AKD-DBPR toward the
complementary fashion item retrieval. As it is time-consuming to rank all the bottoms for each
top, we adopted the common strategy [37] that feeds each top t
i
appeared in S
test
as a query, and
randomly selected T bottoms as the ranking candidates, where there is only one positive bottom.
ereafter, by passing them to the trained neural networks, getting their latent representations
and calculating the compatibility score m
ij
according to Eq. (4.2), we generated a ranking list
of these bottoms for the given top. In our setting, we focused on the average position of the
positive bottom in the ranking list and thus adopted the MRR metric [52, 129, 135].
..................Content has been hidden....................

You can't read the all page of ebook, please click here login for view all page.
Reset