SS symmetry

Article
Probabilistic Unsupervised Machine Learning
Approach for a Similar Image Recommender System
for E-Commerce
Ssvr Kumar Addagarla and Anthoniraj Amalanathan *
School of Computer Science and Engineering, Vellore Institute of Technology, Vellore 632014, Tamil Nadu, India;
[email protected]
* Correspondence: [email protected]




Received: 28 September 2020; Accepted: 23 October 2020; Published: 27 October 2020


Abstract: The recommender system is the most profound research area for e-commerce product
recommendations. Currently, many e-commerce platforms use a text-based product search, which has
limitations to fetch the most similar products. An image-based similarity search for recommendations
had considerable gains in popularity for many areas, especially for the e-commerce platforms
giving a better visual search experience by the users. In our research work, we proposed a
machine-learning-based approach for a similar image-based recommender system. We applied a
dimensionality reduction technique using Principal Component Analysis (PCA) through Singular
Value Decomposition (SVD) for transforming the extracted features into lower-dimensional space.
Further, we applied the K-Means++ clustering approach for the possible cluster identification for a
similar group of products. Later, we computed the Manhattan distance measure for the input image to
the target clusters set for fetching the top-N similar products with low distance measure. We compared
our approach with five different unsupervised clustering algorithms, namely Minibatch, K-Mediod,
Agglomerative, Brich, and the Gaussian Mixture Model (GMM), and used the 40,000 fashion product
image dataset from the Kaggle web platform for the product recommendation process. We computed
various cluster performance metrics on K-means++ and achieved a Silhouette Coefficient (SC) of
0.1414, a Calinski-Harabasz (CH) index score of 669.4, and a Davies–Bouldin (DB) index score of 1.8538.
Finally, our proposed PCA-SVD transformed K-mean++ approach showed superior performance
compared to the other five clustering approaches for similar image product recommendations.

Keywords: PCA-SVD dimensionality reduction; K-means++ clustering; similar image

recommender system; Manhattan distance; cluster similarity

1. Introduction
In the digital era, the e-commerce industry is growing rapidly, especially since many of the users
are migrating towards online shopping from traditional offline shopping in many developing countries.
In countries such as India, China, Singapore, Malaysia, Japan, etc., the growth rate consistently
increases, and millions of users are interested in purchases through e-commerce platforms [1,2].
There is a wider scope of increase in the demand for e-commerce purchases in many developing nations
due to the COVID-19 pandemic situation, and millions of people prefer to purchase through online
shopping now-a-days. Especially several categories of products are in high demand like clothing,
electronics, furniture, sports, etc. [2]. Out of which apparels and some of the electronic products mostly
rely on visual appearance to attract the users to purchase the products. These e-commerce portals
consist of millions of images relevant to various products, and bringing the desired product of the
customer is a challenging issue. Many researchers had come up with several possibilities to address

Symmetry 2020, 12, 1783; doi:10.3390/sym12111783 www.mdpi.com/journal/symmetry

ral possibilities to address the problem but not a satisfactory solution to many e-comm
lems [3–6] as many e-commerce platforms use a recommender system and mainly rely on
d searching approaches. It takes the user input and bases it on the word tokenization pro
recommender system
Symmetry 2020, 12,is to bring the possible matching products to the users.2 However,
1783 of 18 this
pproach had some limitations and missed many features such as colors, patterns, texture,
the problem but not a satisfactory solution to many e-commerce problems [3–6] as many e-commerce
e in the product images [7,8]. In this text-based search, there is a need to specify the descrip
platforms use a recommender system and mainly rely on text-based searching approaches. It takes
ch product,thewhich
user input may notitdescribe
and bases the entireprocess.
on the word tokenization product. Here, we
The recommender come
system up with a pos
is to bring
the possible matching products to the users. However, this kind of approach had some limitations
ion, which is to search for the relevant product bases on the given input image, as show
and missed many features such as colors, patterns, texture, and shape in the product images [7,8].
re 1. We try In
tothis
train the model
text-based foris the
search, there various
a need features
to specify the through
descriptions the feature
of each product, which may engineering
not pro
describe the entire product. Here, we come up with a possible solution, which is to search for the
as colors, texture, and the shape of the objects in the image, which further can be proce
relevant product bases on the given input image, as shown in Figure 1. We try to train the model for the
ugh variousvariousmachine learning
features through dimensionality
the feature reduction
engineering process, such techniques,
as colors, texture, and the shape ofsuch
the as Prin
objects in the image, which further can be processed through various machine learning dimensionality
ponent Analysis (PCA) [9], Expectation-Maximization (EM) PCA [10], Linear Discrimi
reduction techniques, such as Principal Component Analysis (PCA) [9], Expectation-Maximization
ysis (LDA) [11], and
(EM) PCA Probabilistic
[10], Linear DiscriminantPCA [12,13],
Analysis forand
(LDA) [11], the properPCA
Probabilistic recommendations
[12,13], for the proper from the g
recommendations from the given input image [14]. Once the features are extracted, it is necessary
t image [14]. Once the features are extracted, it is necessary to compute the similarity measu
to compute the similarity measure in terms of distance to be computed from the origin to the set of
s of distanceimages
to beincomputed
the database [15].from the origin to the set of images in the database [15].

Figure 1. Outcome
Figure 1. Outcome of the
of the proposed final
proposed recommender
final model.
recommender model.
Several clustering-based unsupervised learning approaches have been proposed by researchers
for a similar image recommender system where the ground truth labels are known. For measuring the
Several clustering-based unsupervised learning approaches have been proposed by researc
distance and analyzing the quality of the recommender model, the following measures [16] are to be
similar image
usedrecommender
for the unsupervised system where the ground truth labels are known. For measu
learning approach.
distance and •analyzing
Adjusted RandtheIndex
quality
(ARI) isof the
used recommender
to measure model,
the similarity between thethe following
clusters measures [16
where ground
truth labels are known
used for the unsupervised [17]. Theapproach.
learning ARI can be computed using Equations (1) and (2) [16].

Random Index − Expected Random Index

Adjusted Rand Index (ARI) isARIused
= to measure
Maximum the−similarity
Random Index between
Expected Random Index the clusters
(1) where gro
truth labels are known [17]. The ARI can be computed x + y using Equations (1) and (2) [16].
Random Index = nsamples (2)
S2

𝑅𝑎𝑛𝑑𝑜𝑚 𝐼𝑛𝑑𝑒𝑥 − 𝐸𝑥𝑝𝑒𝑐𝑡𝑒𝑑 𝑅𝑎𝑛𝑑𝑜𝑚 𝐼𝑛𝑑𝑒𝑥

𝐴𝑅𝐼 =
𝑀𝑎𝑥𝑖𝑚𝑢𝑚 𝑅𝑎𝑛𝑑𝑜𝑚 𝐼𝑛𝑑𝑒𝑥 − 𝐸𝑥𝑝𝑒𝑐𝑡𝑒𝑑 𝑅𝑎𝑛𝑑𝑜𝑚 𝐼𝑛𝑑𝑒𝑥
𝑥+𝑦
Symmetry 2020, 12, 1783 3 of 18

where x, y are the number of elements with the same and different set in S of the cluster.
• Homogeneity (hg), Completeness (cp) are the measures for the same class and same cluster
predictions [18], and harmonic mean can be computed using V-measure [19] as described in
Equation (3) [16].
(1 + β) × hg × cp
Vmeasure = (3)
(β × hg + cp)
where β is the random value to be randomly considered as less than 1, the harmonic mean value
ranges between 0 and 1, and the highest value will be the best to consider.
• The Fowlkes–Mallows score (FMS) [20,21] is used to compute the geometric mean of the similarity
of the clusters where the ground truth labels are known where the FMS is lying between the 0 and
1, and a greater value is a better similarity among the clusters. The FMS uses the True Positives
(TP), False Positive (FP), and False Negative (FN) for the similarity measure analysis is shown in
Equation (4) [16].
TP
FMS = p (4)
(TP + FP)(TP + FN )

In our research work, we proposed of approach through unsupervised clustering algorithms

where the ground truth labels are unknown. This kind of clustering problem can be computed with
performance measures such as the Silhouette Coefficient (SC), the Calinski-Harabasz (CH) index,
and the Davies–Bouldin (DB) index, and these measures are discussed briefly in the results section in
the paper.
Further, we organized this research work into five major sections, where related works are
discussed in Section 2, our proposed method of a similar image recommender system is presented in
Section 3, dataset, pre-processing stage, and various experimental approaches are presented in Section 4,
the final results and comparisons are presented in Section 5 and the conclusion and limitations of this
research work are presented in Section 6.

2. Related Works
Image-based recommendations through clustering play a vital role in many application areas,
and profound research works are done mostly on the medical images for various analyses using
artificial intelligence and other machine learning techniques. Hancer et al. [22] proposed an artificial
bee colony approach for the three benchmarks Lena, Remote Sensing, and Brain MRI images for the
clustering approach compared with the particle swarm optimization, and K-means cluster approaches.
M. Gong et al. [23] had come with a new clustering approach using the Kernal metric for the fuzzy
c-mean clustering process and tested on the various image datasets like synthetic, natural, and medical
images for the performance evaluation. Karthikeyan and Aruna [24] proposed a probabilistic text and
image-based semi-supervised clustering approach. They used the topic modeling, comparing image
features such as color sets and image block signature computing the similarity distance measure.
Their method further compared with the K-means and Dbscan unsupervised clustering algorithms.
Matrix-factorization-based image clustering was proposed by K. Zeng et al. [25], and the objective
of their work to recognize the basis on the parts of the objects in the images which are computed
through the non-negative matrix factorization technique with the hypergraph regularization process.
The authors used the USPS handwritten digit dataset, ORL, and Yale, which are the grayscaled
face image datasets used for the performance analysis, and compared the results with pre-existence
clustering algorithms such as graph regularized Non-negative matrix factorization (NMF), PCA,
and K-means. An improved Artificial Bee Colony (ABC) clustering approach had been tested on seven
benchmark images compared to the fitness values, objective functions along with quality measures
such as DBI and XBI values with particle swarm optimization, genetic algorithms, and K-means
clustering [26].
Symmetry 2020, 12, 1783 4 of 18

Younus et al. [27] proposed a content-based similar image retrieval system using Particle swarm
optimization (PSO) and K-means by extracting various features in the color images like histograms,
co-occurrence matrix, both the color and wavelet moment for the clustering process. For their
experimentation, the Wang dataset is used, which consists of 10 label classes with every 100 images.
Precision and recall measures are computed on the query and retrieval images and listed out the
comparison results with existing methods. The hierarchical and flat clustering approach used to cluster
for the cell phone images was proposed [28,29] using sensor pattern noise. In this approach, a total of
1350 images are used to process and cropped and converted to grayscale for applying wiener filter
and applied the hierarchical and flat clustering approach. Then, to validate the clustering process,
a silhouette coefficient is computed and further applied true positive rate to assess the degree of
certainty of the clusters.
Convolutional neural networks have been used to train and perform image clustering by
continuously applying the forward and backward propagation to improve the clustering process
for representation learning, object loss function, and interpreted through agglomerative clustering
process [30]. The work is extensively tested on the benchmark multiple handwritten image datasets and
facial image datasets and made a performance comparison with several other algorithmic approaches
by using NMI metric. Pandey and Khanna [31] had proposed the Content-based image retrieval (CBIR)
approach using agglomerative clustering and applied on labeled multiple datasets and computes
various cluster measure for similarity. Vantage Point (VP) trees are a popular approach for the faster
indexing of the images in a CBIR approach and in [32] come up with a new distance index measure
called DCIVI for the nearest and furthest neighbors by applying the VP-Tree concept for faster image
retrieval on a remote sensing image dataset compared with a sequential scanning algorithm in terms
of indexing, feature extraction and query response time.
Biradar and Ahmed [33] developed visual CBIR using edge and corner detection of the image
using the Harris corner detector and feature extraction of the images done using the Voronoi tree
algorithm. Further authors using a support vector machine (SVM) as a classifier of final image
classification achieved 90% accuracy on their approach. In [13], the PCA-based unsupervised learning
approach was implemented for the segmentation of brain tumor cells from T1 weighted MRI scan
images. To this process, authors tested various PCA dimensionality reduction techniques and found
that Expectation-Maximization (EM) PCA and Probabilistic PCA (PPCA) are the best fit for their process
through FCM clustering and K-means clustering with various image dimensions. A graph-based
probabilistic dimensionality reduction using manifold hashing techniques was proposed for image
similarity search, which uses the K-nearest neighbor approach for landmark representation and
further analysis [34]. Authors had tested their model on benchmark datasets such as CIFAR, MNIST,
NUS-WIDE, and GIST and compared with other popular hash methods, such as LSH, ITQ, MDSH,
AGH, etc., and shown improvement using their proposed method.
A similar fruit query system was proposed by Fachrurrozi et al. [35] using fuzzy-based feature
extraction using a color histogram, and the color moment invariant approach then applied the K-means
clustering for the extracted features to compute the similarity distance between the query image and
the cluster images done using the K-nearest neighbor search algorithm. In this process, the authors
used various single and multi-object fruits to test the accuracy of their model and achieved 92.5%
and 90%, respectively. Many researchers have been using converting RGB (Red, Green, and Blue)
images to greyscale for feature extraction and further apply various approaches for similar search
recommendations. Instead of converting into greyscale, PCA with a smaller dimension of the images can
be utilized. Fleece fabric-based image classification has developed using PCA dimensionality reduction
and further applies K-nearest neighbors and Naive Bayes classifiers for the accuracy comparison [36].
Various textile image retrieval was proposed using local PCA-based features descriptors like color,
orientation for localized color feature extraction and classification done on the join feature criterion for
better image retrieval and experimentation carried out on various textile images with different patterns
and measured the precision and recall [37].
Symmetry 2020, 12, 1783 5 of 18

Chen et al. [38] proposed a novel approach to high dimensional large scale datasets for image
classification and retrieval using the local neighboring approach and clustering through NQ-Dbscan
and achieves the O(n ∗ log(n)) and O(n) for some cases. Singh and Srivastava [39] had come up with
an improved image classifier using Random Forest classifier and LBP feature extraction for text, color,
moment, and histogram features. Benchmark Wang dataset is used for performance analysis in terms
of accuracy, precision-measure, and MCC are compared with K-NN, SVM, Naive Bayes with the
proposed approach. In [40], an artificial neural-network-based approach, along with various image
descriptors for color, edge detection using the YCbCr method and discrete wavelet transformation,
was applied for better image retrieval, and experiments have been carried out on the Wang dataset and
computed the accuracy measures. Later, Recall measures were compared with the existing methods.
Jian et al. [41] used the direction patch extraction approach based on the perception and color histogram,
and the Gabor filter for texture feature, employed for its process to avoid the complex segmentation,
further uses Dbscan clustering for various regions and was compared with other researchers work in
terms of precision and recall.
Jafarzadegan et al. [42] proposed the PCA-based combined hierarchical cluster approach and used
the cophenetic correlation score and Wilcoxon hypothesis measure to compare the quality and efficiency
of the cluster approach. Various datasets have been taken and compared with other existing methods
and shown improvement using their methodology. Mateen et al. had proposed VGG-19-based feature
extraction along with dimensionality reduction using PCA, SVD and optimized through Gaussian
Mixture Model for the Fundus image classification and carried out the experimentation on the Kaggle
image dataset consisting of 35,126 images and achieved more than 92% accuracy at various levels [43].
The authors of [44] presented keyword-based image recommendations for e-commerce products using
the Markov-chain-based method along with the annotations for the image and results are compared
with the existing methods. D. sha et al. had proposed a fashion clothing feature extraction for analyzing
various attributes, such as cloth pattern, sleeves, and collar using GIST, Fourier and Pyramid Histogram
of Oriented Gradients (PHOG) feature extraction and experimented using the tmall fashion dataset with
8000 images [45]. In [46], the authors proposed a text-based image recommendation using bag-of-words
and the Term frequency–Inverse document frequency (TF-IDF) approach to get similar product image
recommendations. For this approach, the authors experimented on the amazon fashion image dataset
and computed the Euclidean distance measure to fetch the top-N similar recommendations.
Many researchers have worked on various machine learning approaches for similar image retrieval
systems for various domains with supervised and unsupervised approaches. The majority of the
researchers performed computing performance metrics such as precision, recall, and F1, where ground
truth labels are known. In this paper, we have addressed a similar image recommender system for
the e-commerce domain using various machine learning unsupervised clustering algorithms and
computed the cluster performance metrics where ground truth labels are unknown.

3. Similar Image Recommendation

In this section, we discuss our proposed research method using Machine learning statistical
unsupervised dimensionality reduction techniques using Principal Component Analysis through
singular value decomposition (PCA-SVD) on a fashion image dataset and further applied unsupervised
clustering algorithm using K-means(++) to find out the similar image clusters and computed the
distance measures for the similar top-N product image recommendations and in Figure 2 shown our
proposed research method.
unsupervised dimensionality reduction techniques using Principal Component Analysis through
singular value decomposition (PCA-SVD) on a fashion image dataset and further applied
unsupervised clustering algorithm using K-means(++) to find out the similar image clusters and
computed the distance measures for the similar top-N product image recommendations and in Figure
Symmetry 2020, 12, 1783 6 of 18
2 shown our proposed research method.

Figure Figure 2. Proposed

2. Proposed visual
visual recommender system
recommender system using
usingmachine
machinelearning approaches.
learning approaches.
3.1. PCA through Eigen Value Decomposition (PEVD)
Dimensionality reduction is helpful to reduce the image dimensions and initial image matrix
represented with the RGB (Red, Blue, and Green) color space values. Initially, the standardization
process was for scaling and centering the data on a d-dimensional apparel image dataset. In this
process, we can compute mean and standard deviation on the given data points as follows
Equations (5)–(7) [9,47].
Smean = Mean(si )i=1n (5)

σ = Sd_dev(si )i=1n (6)

0si − smean
si = (7)
σ
where si are the feature values of the given data matrix; further, the covariance data matrix is constructed
to see how far the features are changing together by using the following covariance Equation (8),
where A, B are the features and ai, bi are the feature values, and the mean is represented by µ.
n
1X
cov(A, B) = (ai − µA ) ∗ (bi − µY ) (8)
n
i=1

Further, the covariance matrix is used for calculating and building eigenvalues and eigenvectors
for measuring the spread of the variance on projected vector points. Computing the eigenvalues and
eigenvector from the following Equation (9):

λn vn = Uvn (9)

where λ1 , λ2 , λ3 , . . . λn andv1 , v2 , v3 , . . . vn are the eigenvalues and eigenvectors of U covariance matrix

and further generates the new, transformed dimension vector.

3.2. PCA through Singular Value Decomposition (PSVD)

Principal components (PC) computation through singular value decomposition is a better approach
in terms of numerical precision and stability. SVD [48,49] relies on the divide-and-conquer strategy,
and eigenvalue decomposition mostly relies on the QR algorithm [50], which is less stable, and a loss
 Symmetry 2020, 12, 1783 7 of 18

of precision can cause the forming of the covariance matrix in the EVD approach. PCA through SVD is
also computationally efficient when compared to EVD in terms of dealing with the high dimensional
datasets. SVD for principal
 components can be computed using Equation (10), where the given Matrix
X singular values σ1 ≥ σ2 ≥ σ3 ≥ σ4 . . . .σp ≥ 0 are represented by the Σ, and orthogonal values are
represented by the Uand V, respectively.

T
X = Un×m Σn×p Vp×p (10)

Furthermore, in our approach, we have performed the dimensionality reduction on our e-commerce
dataset. The total dimensions in our dataset before PSVD [51] transformation is 14,400 and after
applying the SVD and reduced by 90.01% and retained the 144 components to transform the e-commerce
images. The following Figure 3 shows the reconstructed images from the original image after retaining
the major principal components, and Figure 4 shown the 2-Dimensional visualization image projection
0, 12, x FOR PEER REVIEW
for major principal components of the e-commerce product dataset.

Figure 3. Sample Principal Component Analysis (PCA) through Singular Value Decomposition (PSVD)
3. Sample Principal Component
transformed reconstructed images. Analysis (PCA) through Singular Value Decompos

) transformed reconstructed images.

Symmetry 2020,3.12,Sample
Figure 1783 Principal Component Analysis (PCA) through Singular Value Decomposition8 of 18
(PSVD) transformed reconstructed images.

Figure
Figure 4. PCA-SVD
PCA-SVD Transformed
Transformed images using 2-D visualization.

3.3. K-Mean(++)
3.3. K-Mean(++)Clustering
ClusteringApproach
Approach
Clustering is
Clustering is an
an unsupervised
unsupervised learning
learning technique
technique toto find
find out
out the
the K-Patterns
K-Patterns inin the
the given
given image
image
dataset. For
dataset. For our
our image
image similarity
similarity approach,
approach, we we have
have taken
taken aa Fashion
Fashion image
image dataset
dataset with
with 40,000
40,000
images for analysis. We have computed K-Mean clustering [52] by computing ten iterative timestimes
images for analysis. We have computed K-Mean clustering [52] by computing ten iterative with
with different
different centroid
centroid positions,
positions, and theandK-Means++
the K-Means++ [53] initialization
[53] initialization methodmethod
has beenhasused
beenfor
used for
better
better convergence. An earlier version of K-Means consists of the initialization method,
convergence. An earlier version of K-Means consists of the initialization method, assignment of data assignment of
data points, updating the centroid, and repeats the stages until it finds out the
points, updating the centroid, and repeats the stages until it finds out the convergence. Hence, convergence. Hence,
choosing the
choosing therandom
random k-centroids
k-centroids is also known
is also as theas
known Initialization sensitivity
the Initialization procedure.
sensitivity To overcome
procedure. To
the initialization sensitivity procedure in our approach, we have adapted the K-Means
overcome the initialization sensitivity procedure in our approach, we have adapted the K-Means ++ ++ approach
for findingfor
approach out the better
finding initial
out the selection
better initial of K-centroids
selection for PCA transformed
of K-centroids images using
for PCA transformed imagestheusing
SVD
approach
the on our e-commerce
SVD approach 40K dataset.
on our e-commerce 40KHere, in the
dataset. Here, in the++
K-Means approach
K-Means initially picks
++ approach a centroid
initially picks
point C1 arbitrarily and computes the distance to all available data points from the arbitrarily initialized
centroid, as shown in Equation (11).

2
di = max( j:1−m) xi − C j (11)

where m is the number of centroids used and picked for each iteration and xi becomes the new centroid
depends on the proportion of higher probability to distance di and we repeat these steps until finding
the K-centroids. Further, we find the Euclidean distance for each data point from the K-centroids using
Equation (12).
q
d(p, q) = (q1 − p1 )2 + (q2 − p2 )2 (12)

Closest points with respect to the e-commerce images are assigned to the clusters and repeats
this process until we reach the convergence and all the image points (Si ) are assigned to ith Cluster by
using Equation (13).
centroid depends on the proportion of higher probability to distance 𝑑 and we repeat these steps
until finding the K-centroids. Further, we find the Euclidean distance for each data point from the K-
centroids using Equation (12).

𝑑(𝑝, 𝑞) = (𝑞 − 𝑝 ) + (𝑞 − 𝑝 ) (12)
Symmetry 2020, 12, 1783 9 of 18
Closest points with respect to the e-commerce images are assigned to the clusters and repeats
this process until we reach the convergence and all the image points (𝑆 ) are assigned to 𝑖 Cluster
by using Equation (13).
1 1 X
𝐶 Ci= 𝑥 x (13)
(13)
|𝑆 | |Si | ∈ xi ∈Si i
Further,
Further, we wehave
havegenerated
generated 16 16 different
different clusters
clusters fromfrom sampled
sampled imagesimages
from thefrom40Kthe 40K dataset
image image
dataset by computing inertia, as shown in Figure 5, and the silhouette coefficient
by computing inertia, as shown in Figure 5, and the silhouette coefficient to understand the goodness to understand the
goodness of our
of our cluster cluster
model. model. (14)
Equation Equation
is used (14)
to is used tothe
measure measure
goodness the metric
goodness metricthe
through through the
silhouette
silhouette coefficient [54] on various clusters formed
coefficient [54] on various clusters formed using k-Means++. using k-Means++.
𝑎(𝑖)
⎧1 − 𝑖𝑓𝑎(𝑖) < 𝑏(𝑖)
 𝑏(𝑖)
a(i)
⎪
⎪ 
 1 − b(i) i f a ( i ) < b ( i )
0 𝑖𝑓𝑎(𝑖) = 𝑏(𝑖)

𝑆(𝑖)S(=i) = 

0 i f a(i) = b(i) (14)
(14)
⎨𝑏(𝑖)

 b−(i)1 𝑖𝑓𝑎(𝑖) > 𝑏(𝑖)
− 1 i f a(i) > b(i)

⎪𝑎(𝑖)

⎪

a(i)
⎩
where (i), b𝑏(𝑖)
where a𝑎(𝑖), (i) are
arethe
theaverage
average and
andlowest average
lowest averagedistances between
distances i and𝑖 all
between andother
all data
otherpoints
data within
points
and different clusters, respectively.
within and different clusters, respectively.

Figure 5.
Figure 5. Optimal
Optimal clusters
clusters finding
finding using
using K-means++
K-means++ inertia.
inertia.

3.4. Similar
3.4. Similar Distance
Distance Measure
Measure
Fetching the
the relevant
relevantsimilar
similarimages
imagesfrom
fromthe
the given
given input
input imageimage from
from the the fashion
fashion dataset,
dataset, we
we
havehave computed
computed thethe Manhattan
Manhattan distance
distance measure
measure [15]
[15] between
between any anytwo
twogiven
givenvectors
vectorsXXand
and Y
Y to
compute the lowest distance on PCA through SVD transformed image clusters generated using the
K-Means++approach
K-Means++ approachfor fore-commerce
e-commerceproduct
product image
image clusters
clusters by
by using
using the distance metric function
as shown in Equation (15) and fetches the top-5 most similar images for the given input image as
user recommendations.
Xn
Manhattan(X, Y) = |Xi − Yi | (15)
i=1

4. Dataset and Experimentation

We have collected 40K fashion product image dataset [55] along with the major category and
subcategory of the products. There is a total of six major categories of products, such as apparel,
accessories, footwear, personal care, sporting goods, etc., and 44 subcategories of the various fashion
products top wear, shoes, bags, watches, wallets, belts, etc. Out of these major categories and
subcategories, most items are listed in apparel, accessories, footwear, and personal care. Further, as in
accessories, footwear, personal care, sporting goods, etc., and 44 subcategories of the various fashion
products top wear, shoes, bags, watches, wallets, belts, etc. Out of these major categories and
subcategories, most items are listed in apparel, accessories, footwear, and personal care. Further, as
in the pre-processing data
Symmetry 2020, 12, 1783 stage, we have sampled out into our training dataset with107632 of 18 images of

16 subcategories for our experimentation purpose with each category of equal distribution to avoid
the unbalance
the data distribution
pre-processing problems
data stage, we havethat affect
sampled out the
into final recommendations.
our training dataset with 7632 Figure
images of6 shows the
16 subcategories for our experimentation purpose with each
category product image distribution from our fashion product image dataset. category of equal distribution to avoid
the unbalance data distribution problems that affect the final recommendations. Figure 6 shows the
category product image distribution from our fashion product image dataset.

Figure 6. Total category distribution from the Kaggle fashion product dataset.

Experimentation
Figure 6. Total category distribution from the Kaggle fashion product dataset.
In this section, we performed our experimentation using the PSVD approach for dimensionality
Experimentation
reduction and fetched the important principal components. Initially, we have set the number of
components to 500 and
In this computed
section, the PCA
we performed singular values
our experimentation usingthrough
the PSVD the PSVD
approach forprocess on our sampled
dimensionality
reduction and fetched the important principal components. Initially, we have set the number of
dataset dimensionally reduced to 90.01% generates 144 final principal components. Figure 7 shows
components to 500 and computed the PCA singular values through the PSVD process on our sampled
cumulative dataset
variance by 500
dimensionally PCAto 90.01%
reduced components and
generates 144 finalconvergence achieved
principal components. Figure 7atshows
144 principal
components.cumulative variance by 500 PCA components and convergence achieved at 144 principal components.

FigureFigure 7. PCA
7. PCA throughSVD
through SVD cumulative
cumulativevariance.
variance.
Symmetry 2020, 12, 1783 11 of 18

Symmetry
Symmetry 2020,
2020, 12,
12, xx FOR
FOR PEER
PEER REVIEW
REVIEW 10
10 of
of 17
17
Further, we have computed the K-means++ clustering approach on the PSVD transformed images.
Here, we Further,
Further, we
we have
have computed have computed
computed the
the K-means++
K-means++
various initializations clustering
numberapproach
clustering
for the of clusteron
approach on the
the PSVD
PSVD transformed
identifications transformed
ranging from
images.
images. Here,
Here, we
we have
have computed
computed various
various initializations
initializations for
for the
the number
number of
of cluster
cluster
8 to 16. For each iteration, we tried to determine the better convergence for the better number of cluster identifications
identifications
ranging
ranging from
formation from
of our88 to
to 16.
16. For
fashion For each
each iteration,
products.iteration, we tried
we we
Further, tried to
to determine
have determine
calculated the
the better
better
each convergence
inertia tofor
convergence
cluster for the
the better
maximizebetterthe
number
number of cluster
of clusterof formation
formation of our
of our fashion products. Further, we have calculated each cluster inertia
quality estimation the number offashion products.
clusters Further,
for further we have
analysis. Wecalculated
have found each clustervalues
optimal inertia at
to
to maximize
maximize the
the quality
quality estimation
estimation of
of the
the number
number of
of clusters
clusters for
for further
further analysis.
analysis. We
We have
have found
foundthe
cluster 16 with a lower inertia value, which converge after 35 iterations. In Figures 8 and 9, showed
optimal
optimal values
values at at cluster
cluster 16
16 with
with aa lower
lower inertia
inertia value,
value, which
which converge
converge after
after 35
35 iterations.
iterations. In In Figures
Figures
high dimensional cluster segregation for 16 subcategories of images using t-distributed stochastic
88 and
and 9,
9, showed
showed the the high
high dimensional
dimensional cluster
cluster segregation
segregation forfor 1616 subcategories
subcategories of of images
images using
using t-t-
neighbor embedding (t-SNE) [56] for the K-means++ approach.
distributed
distributed stochastic
stochastic neighbor
neighbor embedding
embedding (t-SNE)
(t-SNE) [56]
[56] for
for the
the K-means++
K-means++ approach.
approach.

Figure
Figure 8. 8.
Figure 8. T-distributed
T-distributed stochastic
T-distributedstochastic neighbor
neighbor embedding
stochasticneighbor embedding (T-SNE)
embedding (T-SNE)visualization
(T-SNE) visualization
visualization on PSVD
onon transformed
PSVD
PSVD transformed
transformed
using
using K-means++.
K-means++.
using K-means++.

Figure 9.
Figure9.
Figure T-SNE
9.T-SNE visualization
visualization on
T-SNE visualization onthe
on theoriginal
the originaldataset.
original dataset.
dataset.
Symmetry 2020, 12, 1783 12 of 18
Symmetry 2020, 12, x FOR PEER REVIEW 11 of 17

Finally, we computed the distance measure using the Manhattan distance measure for the given
input image to the cluster
cluster images
images and
and fetched
fetched the
the top-K
top-K similar
similar images,
images, and
and we
we have
have further
further analyzed
analyzed
and compared with other standard clustering algorithms for better results. All our experimentations
are carried out using a desktop computer with Intel Core i7-8700K Processor/16 GB of RAM/64-bit
Ubuntu operating system/Python 3.6 Environment.

5. Results
5. Results
We have
We have carried
carried out
out various
various cluster
cluster performance metrics to
performance metrics to analyze
analyze and and achieve
achieve better
better
recommendation results
recommendation results for
for our
our approach. Apart from
approach. Apart from inertia
inertia for
for K-means
K-means clustering,
clustering, wewe have,
have,
furthermore, computed various cluster performance metrics, such as the silhouette
furthermore, computed various cluster performance metrics, such as the silhouette coefficient, as coefficient,
as computed
computed using
using Equation
Equation (14)(14)
withwith results
results rangesranges
fromfrom the values
the values −1where
−1 to 1, to 1, where the the
the higher higher
SC
the SCthe
value, value, theefficient
more more efficient the cluster.
the cluster. In our approach,
In our approach, we have weachieved
have achieved
a 0.1414a 0.1414 coefficient
coefficient value,
value, which
which is a higher
is a higher value value from
from the the other
other algorithms.
algorithms. The comparison
The comparison results
results for SC
for the thecoefficient
SC coefficient
for
for various unsupervised clustering algorithms are shown
various unsupervised clustering algorithms are shown in Figure 10.in Figure 10.

Figure 10.
Figure Comparison of
10. Comparison of the
the silhouette
silhouette coefficient
coefficient of
of various
various clusters
clusters shown
shown in
in (a–f).
(a–f).

To analyze the criteria for the variance ratio for checking the clusters, we computed the
To analyze the criteria for the variance ratio for checking the clusters, we computed the Calinski-
Calinski-Harabasz (CH) score. The CH score tells us whether or not the clusters are well separated
Harabasz (CH) score. The CH score tells us whether or not the clusters are well separated and if the
optimal value is higher than any other clustering approach. Equations (16)–(18) shows the calculation
of the CH score:
Symmetry 2020, 12, 1783 13 of 18

and if the
Symmetry optimal
2020, value
12, x FOR isREVIEW
PEER higher than any other clustering approach. Equations (16)–(18) shows the
12 of 17
calculation of the CH score:
𝐵(𝑘)
"
B(k×) (𝑛 −(𝑘)
# "
n − k)
#
𝐶𝐻(𝑘) =
CH (k) 𝑊(𝑘)
= × (16)
(16)
W (k) (𝑘 −(1) k − 1)
Xk X   2
𝑊(𝑘)W=(k ) = (𝑥 − 𝑐 x )(𝑥
− cq− x𝑐 −) cq (17)
(17)
q=1 x∈Cq
∈
Xk X   2
W (k ) = cq − cE cq − cE (18)
𝑊(𝑘) = q=1 x∈C−
(𝑐 q 𝑐 )(𝑐 − 𝑐 ) (18)
∈
where n𝑛 is
where is the
the number
number of of data
data points, (K), 𝑊(𝐾)
points, B𝐵(𝐾), W (K) are
are between
between and
and within
within the
the cluster
cluster variation,
variation,
respectively, k is total clusters, c is the set of cluster points, and the total data size E. CH
respectively, 𝑘 is total clusters, 𝑐 q is the set of cluster points, and the total data size is 𝐸.isThe Thescore
CH
score
we gotwehere
got for
hereour
forapproach
our approach is 669.4,
is 669.4, whichwhich an optimal
an optimal valuevalue compared
compared to other
to other algorithms.
algorithms. The
The comparison results for various CH scores are shown
comparison results for various CH scores are shown in Figure 11. in Figure 11.

Figure 11. Comparison

Figure 11. Comparison of
of the
the Calinski-Harabasz
Calinski-Harabasz score
score of
of various
various clusters
clusters shown
shown in
in (a–f).
(a–f).

Finally, we have computed the Davies–Bouldin index score for the average cluster similarity,
which computes that an optimal value is a high score among the clusters, and zero is the lowest score.
The DB index score can be computed using Equations (19) and (20):

1
𝐷𝐵(𝑘) = 𝑚𝑎𝑥 𝑀 (19)
𝑘
Symmetry 2020, 12, 1783 14 of 18

Finally, we have computed the Davies–Bouldin index score for the average cluster similarity,
which computes that an optimal value is a high score among the clusters, and zero is the lowest score.
The DB index score can be computed using Equations (19) and (20):

k
1X
DB(k) = maxi, j Mij (19)
k
i=1
Symmetry 2020, 12, x FOR PEER REVIEW 13 of 17
s + sj
+ 𝑠i
Mij𝑠 = (20)
𝑀 = dij (20)
𝑑
where 𝑠s is the mean distance between cluster points and 𝑀 M is
is to
to compute
compute the similarity between
clusters i𝑖and
centroids of the clusters andj.𝑗.The
TheDB
DBscore
scorewe
weachieved
achievedisis1.8538,
1.8538,which
which isis
slightly
slightlylower compared
lower compared to
other
to algorithms.
other The The
algorithms. comparative performance
comparative of the DB
performance index
of the DBscore for score
index various forclustering
various algorithms
clustering
is shown inisFigure
algorithms shown 12.in Figure 12.

Figure 12.
Figure Comparison of
12. Comparison of Davies–Bouldin
Davies–Bouldin (DB)
(DB) index
index score
score of
of various
various clusters
clusters shown
shown in
in (a–f).
(a–f).

Tables 11 and
Tables and 22 show
show the
the PSVD
PSVD transformed
transformed cluster
cluster evaluations
evaluations for
for various
various standard
standard algorithms
algorithms
and computational time comparisons for various clustering algorithms.
and computational time comparisons for various clustering algorithms.

Table 1. Performance comparison of PSVD transformed clustering algorithms.

Clustering Algorithm SC Coefficient CH Score DB Score

MiniBatch 0.131669 637.60 1.9229
K-Mediod 0.124709 609.61 2.0946
 Symmetry 2020, 12, 1783 15 of 18

Table 1. Performance comparison of PSVD transformed clustering algorithms.

Clustering Algorithm SC Coefficient CH Score DB Score

Symmetry 2020, 12, x FOR PEER REVIEW 14 of 17
MiniBatch 0.131669 637.60 1.9229
K-Mediod 0.124709 609.61 2.0946
Table 2. Computational
Agglomerative wall 0.108739
time of various clustering
590.66 algorithms.
1.9856
Brich 0.109466 600.36 1.8893
GMM 0.067083Computation
460.55 2.2714Time
Wall
K-Means++ 0.141421 669.44 1.8538
Clustering Algorithm (Milliseconds (ms)/Seconds (s)/Minutes (min)
Cluster Fitting
Table 2. Computational SC Coefficient
wall time of various clustering algorithms. CH Score DB Score
MiniBatch 2.51 s 9.09Time
Computation Wall s 51.2 ms 120 ms
Clustering (Milliseconds (ms)/Seconds (s)/Minutes (min)
K-Mediod Algorithm 10.9 s 9.08 s 60. 4 ms 108 ms
Cluster Fitting SC Coefficient CH Score DB Score
Agglomerative 40.7 s 11.6 s 79.1 ms 131 ms
MiniBatch 2.51 s 9.09 s 51.2 ms 120 ms
Brich K-Mediod 35.4
10.9 s s 9.08 s 9.0560.s 4 ms 10848.2
ms ms 108 ms
GMM Agglomerative 1 min
40.7 s 4 s 11.6 s 10.279.1
s ms 131 ms1.16 s 1.12 s
Brich 35.4 s 9.05 s 48.2 ms 108 ms
K-Means++ GMM 1 21.3
min 4 ss 10.2 s 48.5 1.16
s s 52.5
1.12 s ms 1.04 s
K-Means++ 21.3 s 48.5 s 52.5 ms 1.04 s

Figure 13 shown the final similar product recommendations using PSVD transformed K-
Figure 13 shown the final similar product recommendations using PSVD transformed K-means++
means++ approach with Top-5 recommendations fetched using the Manhattan distance measure.
approach with Top-5 recommendations fetched using the Manhattan distance measure.

Figure 13. Top-5

Figure final
13. Top-5 recommendations
final from
recommendations from the proposed
the proposed approach.approach.

6. Conclusions
6. Conclusions
The purpose of our proposed model is to fetch similar images for e-commerce portals when the
The purpose
user of the
selects ourdesired
proposed
productmodel is to unsupervised
image using fetch similar images
statistical for e-commerce
machine portals when the
learning techniques.
Here, in our approach, we initially performed PSVD dimensionality reduction on the fashion
user selects the desired product image using unsupervised statistical machine learning techniques. product
image dataset and retained the 144 principal components and achieved a 90.01% variance from the
Here, in our approach, we initially performed PSVD dimensionality reduction on the fashion product
image dataset and retained the 144 principal components and achieved a 90.01% variance from the
original dimensions, which are a total of 14,400 dimensions. Further, we performed the K-means++
clustering on the PSVD transformed images. We have compared PSVD transformed with other
Symmetry 2020, 12, 1783 16 of 18

original dimensions, which are a total of 14,400 dimensions. Further, we performed the K-means++
clustering on the PSVD transformed images. We have compared PSVD transformed with other
clustering algorithms, such as MiniBatch, Agglomerative, Brich, the Gaussian Mixture Model (GMM),
and K-Mediod. Performance metrics are evaluated on all these algorithms for the SC coefficient
for similarity, CH score for variance ration, and DB score for average similarity. Out of all the
measures, PSVD-K-means++ has scored well on the SC coefficient and CH score and is lagging in
the DB index score. Where we observed in the final similar image recommendations, most of the
recommendations work well, but, in a case like if the input image orientation changes, it is possible
to get the mixed product suggestion instead of all the similar images. To overcome these limitations
further, we suggested applying image augmentation techniques and deep learning approaches for
more accurate image feature extraction and training for similar image recommendations.

Author Contributions: Conceptualization, methodology, validation, writing—original draft preparation, S.K.A.;

Supervision, review and guidance, A.A. All authors have read and agreed to the published version of
the manuscript.
Funding: This research received no external funding.
Conflicts of Interest: The authors declare no conflict of interest.

References
1. Wulansaria, I.Y.; Parwantob, N.B. Asian E-Commerce Engages Global Trade Openness: The Role of
Information and Communications Technology, Social, and Security Indicators. Int. J. Innov. Creat. Chang.
2020, 11, 12.
2. eCommerce—Asia | Statista Market Forecast. Available online: https://www.statista.com/outlook/243/101/
ecommerce/asia (accessed on 2 February 2020).
3. Salau, A.O.; Jain, S. Feature Extraction: A Survey of the Types, Techniques, Applications. In Proceedings of the
2019 International Conference on Signal Processing and Communication (ICSC), Noida, India, 7–9 March 2019;
pp. 158–164.
4. Haji, M.S.; Alkawaz, M.H.; Rehman, A.; Saba, T. Content-Based Image Retrieval: A Deep Look at Features
Prospectus. Int. J. Comput. Vis. Robot. 2019, 9, 14–38. [CrossRef]
5. Kumari, M. Content Based Image Retrieval. Available online: https://papers.ssrn.com/sol3/papers.cfm?
abstract_id=3371777 (accessed on 14 May 2019).
6. Zhou, J.; Liu, X.; Liu, W.; Gan, J. Image Retrieval Based on Effective Feature Extraction and Diffusion Process.
Multimed. Tools Appl. 2019, 78, 6163–6190. [CrossRef]
7. Pal, M.S.; Garg, D.S.K. Image Retrieval: A Literature Review. Int. J. Adv. Res. Comput. Eng. Technol. 2013, 2,
1323–2278.
8. Limitations of Text Based Image Retrieval Psychology Essay. Available online: https://www.ukessays.com/essays/
psychology/limitations-of-text-based-image-retrieval-psychology-essay.php (accessed on 15 October 2020).
9. Wold, S.; Esbensen, K.; Geladi, P. Principal Component Analysis. Chemom. Intell. Lab. Syst. 1987, 2, 37–52.
[CrossRef]
10. Roweis, S.T. EM Algorithms for PCA and SPCA. Advances in Neural Information Processing Systems.
pp. 626–632. Available online: http://papers.neurips.cc/paper/1398-em-algorithms-for-pca-and-spca (accessed on
22 October 2020).
11. Tharwat, A.; Gaber, T.; Ibrahim, A.; Hassanien, A.E. Linear Discriminant Analysis: A Detailed Tutorial.
AI Commun. 2017, 30, 169–190. [CrossRef]
12. Tipping, M.E.; Bishop, C.M. Probabilistic Principal Component Analysis. J. R. Stat. Soc. Ser. B Stat. Methodol.
1999, 61, 611–622. [CrossRef]
13. Kaya, I.E.; Pehlivanlı, A.Ç.; Sekizkardeş, E.G.; Ibrikci, T. PCA Based Clustering for Brain Tumor Segmentation
of T1w MRI Images. Comput. Methods Programs Biomed. 2017, 140, 19–28. [CrossRef]
14. Geng, X.; Zhang, H.; Bian, J.; Chua, T.-S. Learning Image and User Features for Recommendation in Social
Networks. In Proceedings of the IEEE International Conference on Computer Vision, Santiago, Chile,
7 December 2015; pp. 4274–4282.
Symmetry 2020, 12, 1783 17 of 18

15. Tyagi, V. Similarity Measures and Performance Evaluation. In Content-Based Image Retrieval; Springer:
Berlin/Heidelberg, Germany, 2017; pp. 63–83.
16. Clustering Scikit-Learn 0.23.2 documentation. Available online: https://scikit-learn.org/stable/modules/
clustering.html#clustering-performance-evaluation (accessed on 2 February 2020).
17. Steinley, D.; Brusco, M.J.; Hubert, L. Properties of the Hubert-Arable Adjusted Rand Index. Psychol. Methods
2004, 21, 261. [CrossRef]
18. Rosenberg, A.; Hirschberg, J. V-Measure: A Conditional Entropy-Based External Cluster Evaluation Measure.
In Proceedings of the 2007 joint conference on empirical methods in natural language processing and
computational natural language learning (EMNLP-CoNLL), Prague, Czech Republic, 28–30 June 2007;
pp. 410–420.
19. Hirschberg, J.B.; Rosenberg, A. V-Measure: A Conditional Entropy-Based External Cluster Evaluation. 2007.
Available online: http://www.aclweb.org/anthologyy/D07-103 (accessed on 14 August 2020).
20. Gholamian, M.; Jahanpour, S.; Sadatrasoul, S. A New Method for Clustering in Credit Scoring Problems.
J. Math. Comput. Sci. 2013, 6, 97–106. [CrossRef]
21. Lu, Y.; Wu, Y.; Liu, J.; Li, J.; Zhang, P. Understanding Health Care Social Media Use from Different Stakeholder
Perspectives: A Content Analysis of an Online Health Community. J. Med. Internet Res. 2017, 19, e109.
[CrossRef] [PubMed]
22. Hancer, E.; Ozturk, C.; Karaboga, D. Artificial Bee Colony Based Image Clustering Method. In Proceedings
of the 2012 IEEE Congress on Evolutionary Computation, Brisbane, Australia, 10–15 June 2012; pp. 1–5.
[CrossRef]
23. Gong, M.; Liang, Y.; Shi, J.; Ma, W.; Ma, J. Fuzzy C-Means Clustering with Local Information and Kernel
Metric for Image Segmentation. IEEE Trans. Image Process. 2012, 22, 573–584. [CrossRef] [PubMed]
24. Karthikeyan, M.; Aruna, P. Probability Based Document Clustering and Image Clustering Using
Content-Based Image Retrieval. Appl. Soft Comput. 2013, 13, 959–966. [CrossRef]
25. Zeng, K.; Yu, J.; Li, C.; You, J.; Jin, T. Image Clustering by Hyper-Graph Regularized Non-Negative Matrix
Factorization. Neurocomputing 2014, 138, 209–217. [CrossRef]
26. Ozturk, C.; Hancer, E.; Karaboga, D. Improved Clustering Criterion for Image Clustering with Artificial Bee
Colony Algorithm. Pattern Anal. Appl. 2015, 18, 587–599. [CrossRef]
27. Younus, Z.S.; Mohamad, D.; Saba, T.; Alkawaz, M.H.; Rehman, A.; Al-Rodhaan, M.; Al-Dhelaan, A.
Content-Based Image Retrieval Using PSO and k-Means Clustering Algorithm. Arab. J. Geosci. 2015, 8,
6211–6224. [CrossRef]
28. Lin, X.; Li, C.-T. Large-Scale Image Clustering Based on Camera Fingerprints. IEEE Trans. Inf. Forensics Secur.
2016, 12, 793–808. [CrossRef]
29. Villalba, L.J.G.; Orozco, A.L.S.; Corripio, J.R. Smartphone Image Clustering. Expert Syst. Appl. 2015, 42,
1927–1940. [CrossRef]
30. Yang, J.; Parikh, D.; Batra, D. Joint Unsupervised Learning of Deep Representations and Image Clusters.
In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR), Las Vegas, NV, USA,
27–30 June 2016.
31. Pandey, S.; Khanna, P. Content-Based Image Retrieval Embedded with Agglomerative Clustering Built on
Information Loss. Comput. Electr. Eng. 2016, 54, 506–521. [CrossRef]
32. L i, S.; Yu, H.; Yuan, L. A Novel Approach to Remote Sensing Image Retrieval with Multi-Feature vp-Tree
Indexing and Online Feature Selection. In Proceedings of the 2016 IEEE Second International Conference on
Multimedia Big Data (BigMM), Taipei, Taiwan, 20–22 April 2016; pp. 133–136. [CrossRef]
33. Biradar, M.; Ahmed, M. Visual Based Information Retrieval Using Voronoi Tree. In Proceedings of the 5th
International Conference on Frontiers in Intelligent Computing: Theory and Applications, Beijing, China,
2–5 November 2017; pp. 603–609.
34. Zhu, X.; Li, X.; Zhang, S.; Xu, Z.; Yu, L.; Wang, C. Graph PCA Hashing for Similarity Search. IEEE Trans. Multimed.
2017, 19, 2033–2044. [CrossRef]
35. Fachrurrozi, M.; Fiqih, A.; Saputra, B.R.; Algani, R.; Primanita, A. Content Based Image Retrieval
for Multi-Objects Fruits Recognition Using k-Means and k-Nearest Neighbor. In Proceedings of the
2017 International Conference on Data and Software Engineering (ICoDSE), Palembang, Indonesia,
1–2 November 2017; pp. 1–6.
Symmetry 2020, 12, 1783 18 of 18

36. Yildiz, K. Dimensionality Reduction-Based Feature Extraction and Classification on Fleece Fabric Images.
Signal Image Video Process. 2017, 11, 317–323. [CrossRef]
37. Cui, Y.; Wong, W.K. Textile Image Retrieval Using Joint Local PCA-Based Feature Descriptor. In Applications
of Computer Vision in Fashion and Textiles; Elsevier: Amsterdam, The Netherlands, 2018; pp. 253–271.
38. Chen, Y.; Tang, S.; Bouguila, N.; Wang, C.; Du, J.; Li, H.L. A Fast Clustering Algorithm Based on Pruning
Unnecessary Distance Computations in DBSCAN for High-Dimensional Data. Pattern Recognit. 2018, 83,
375–387. [CrossRef]
39. Singh, V.P.; Srivastava, R. Improved Image Retrieval Using Fast Colour-Texture Features with Varying
Weighted Similarity Measure and Random Forests. Multimed. Tools Appl. 2018, 77, 14435–14460. [CrossRef]
40. Ashraf, R.; Ahmed, M.; Jabbar, S.; Khalid, S.; Ahmad, A.; Din, S.; Jeon, G. Content Based Image Retrieval by
Using Color Descriptor and Discrete Wavelet Transform. J. Med. Syst. 2018, 42, 44. [CrossRef] [PubMed]
41. Jian, M.; Yin, Y.; Dong, J.; Lam, K.-M. Content-Based Image Retrieval via a Hierarchical-Local-Feature
Extraction Scheme. Multimed. Tools Appl. 2018, 77, 29099–29117. [CrossRef]
42. Jafarzadegan, M.; Safi-Esfahani, F.; Beheshti, Z. Combining Hierarchical Clustering Approaches Using the
PCA Method. Expert Syst. Appl. 2019, 137, 1–10. [CrossRef]
43. Mateen, M.; Wen, J.; Nasrullah; Song, S.; Huang, Z. Fundus Image Classification Using VGG-19 Architecture
with PCA and SVD. Symmetry 2019, 11, 1. [CrossRef]
44. Sejal, D.; Rashmi, V.; Venugopal, K.R.; Iyengar, S.S.; Patnaik, L.M. Image Recommendation Based on Keyword
Relevance Using Absorbing Markov Chain and Image Features. Int. J. Multimed. Inf. Retr. 2016, 5, 185–199.
[CrossRef]
45. Sha, D.; Wang, D.; Zhou, X.; Feng, S.; Zhang, Y.; Yu, G. An Approach for Clothing Recommendation Based
on Multiple Image Attributes. In Proceedings of the International Conference on Web-Age Information
Management, Nanchang, China, 3–5 June 2016; pp. 272–285.
46. Shrivastava, R.; Sisodia, D.S. Product Recommendations Using Textual Similarity Based Learning Models.
In Proceedings of the 2019 International Conference on Computer Communication and Informatics (ICCCI),
Coimbatore, India, 23–25 January 2019; pp. 1–7.
47. Aït-Sahalia, Y.; Xiu, D. Principal Component Analysis of High-Frequency Data. J. Am. Stat. Assoc. 2019, 114,
287–303. [CrossRef]
48. Furnas, G.W.; Deerwester, S.; Durnais, S.T.; Landauer, T.K.; Harshman, R.A.; Streeter, L.A.; Lochbaum, K.E.
Information Retrieval Using a Singular Value Decomposition Model of Latent Semantic Structure.
ACM SIGIR Forum 2017, 51, 90–105. [CrossRef]
49. De Lathauwer, L.; De Moor, B.; Vandewalle, J. A Multilinear Singular Value Decomposition. SIAM J. Matrix
Anal. Appl. 2000, 21, 1253–1278. [CrossRef]
50. Parlett, B.N. The QR Algorithm. Comput. Sci. Eng. 2000, 2, 38–42. [CrossRef]
51. Wall, M.E.; Rechtsteiner, A.; Rocha, L.M. Singular Value Decomposition and Principal Component Analysis.
In A Practical Approach to Microarray Data Analysis; Springer: Berlin/Heidelberg, Germany, 2003; pp. 91–109.
52. Ding, C.; He, X. K-Means Clustering via Principal Component Analysis. In Proceedings of the Twenty-First
International Conference on Machine Learning, Banff, AB, Canada, 4–8 July 2004; p. 29.
53. Arthur, D.; Vassilvitskii, S. K-Means++: The Advantages of Careful Seeding; Stanford University: Stanford, CA,
USA, 2006; pp. 1–11.
54. Aranganayagi, S.; Thangavel, K. Clustering Categorical Data Using Silhouette Coefficient as a Relocating
Measure. In Proceedings of the International Conference on Computational Intelligence and Multimedia
Applications (ICCIMA 2007), Sivakasi, India, 13–15 December 2007; Volume 2, pp. 13–17.
55. Fashion Product Images Dataset | Kaggle. Available online: https://www.kaggle.com/paramaggarwal/
fashion-product-images-dataset (accessed on 2 February 2020).
56. van der Maaten, L.; Hinton, G. Visualizing Data Using T-SNE. J. Mach. Learn. Res. 2008, 9, 2579–2605.

Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional
affiliations.

© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access
article distributed under the terms and conditions of the Creative Commons Attribution
(CC BY) license (http://creativecommons.org/licenses/by/4.0/).