A Review On Deep Learning Applications
A Review On Deep Learning Applications
Chapter 2
A Review on Deep
Learning Applications
Chitra A. Dhawale
P. R. Pote College of Engineering and Management, India
Kritika Dhawale
Indian Institute of Information Technology, Nagpur, India
Rajesh Dubey
Mohanlal Sukhadia University, India
ABSTRACT
Artificial intelligence (AI) is going through its golden era. Most AI applications are indeed using ma-
chine learning, and it currently represents the most promising path to strong AI. On the other hand, deep
learning, which is itself a kind of machine learning, is becoming more and more popular and successful
at different use cases and is at the peak of developments by enabling more accurate forecasting and
better planning for civil society, policymakers, and businesses. As a result, deep learning is becoming a
leader in this domain. This chapter presents a brief review of ground-breaking advances in deep learn-
ing applications.
INTRODUCTION
Artificial intelligence (AI) is omnipresent, our daily life is mostly influenced by it in one way or the
other. In recent years, most of the researchers are getting attracted towards AI domain due to its major
applications including multimedia (text, image, speech, video) recognition, social network analysis,
data mining, natural language processing, driverless car and so forth are using machine learning which
leads this sub domain to the top of popularity amongst researchers and industrialist. Most AI applica-
tions are indeed using Machine learning and it currently represents the most promising path to strong
AI. Machine-learning technology has a potential which influenced our modern society in all dimensions
DOI: 10.4018/978-1-7998-1192-3.ch002
Copyright © 2020, IGI Global. Copying or distributing in print or electronic forms without written permission of IGI Global is prohibited.
A Review on Deep Learning Applications
like content mining social media networks to e-commerce websites, it has converted our simple calling
phone to smart phone, normal camera to smart cameras. Apart from this Machine-learning are used in
image identification, speech recognition and its translation to text, extraction of specific news, customer
sentiment analysis for business growth etc .
Till now machine learning was implementing all above tasks in traditional way by processing raw
data and then applying feature extraction techniques followed by matching and labeling. To execute these
steps machine-learning system needed a significant skillful domain wise expertise to construct a feature
extractor which can represent the raw data into specific feature vector which can be used by machine
learning algorithm often a classifier, which can identify and classify input data. Machine learning uses
single level at each stage of processing raw data to suitable representation for detection, recognition and
classification. Due to this single level of processing between input and output, may lead to less accurate
result, On the other hand recently to increase accuracy and speed of processing, Deep-learning methods
are used which uses learning methods but with multiple intermediate layers of representation. These
layers are incorporated by non-linear modules which modify the representation at one level into a more
conceptual representation at next level. Complex functions are built using such simple transformation
at each layer which can lead to decrease the variations and discriminations. Unlike machine learning
feature extracting layers are not designed by human experts: they are learned from data using a general-
purpose learning procedure.
Deep learning is making a significant contribution for achieving more and more strong and accurate
decision . Deep learning has changed the entire landscape over the past few years. Every day, there are
more applications that rely on deep learning techniques in fields as diverse as healthcare, finance, human
resources, retail, earthquake detection, and self-driving cars. In Overall dimensions in development DL
is at the peak and hence DL is becoming a leader in this domain (LeCun, Bengio, and Hinton, 2015).
This chapter presents a brief review of pioneering advances in deep learning applications.
Here we will take a general review of several selected applications of the deep networks.
Speech Recognition
During the past few decades, machine learning algorithms have been widely used in areas such as
automatic speech recognition (ASR) and acoustic modeling (Jaitly and Hinton, 2011 & Mohammed,
Dahl, and Hinton, 2012 & Mohamed, Yu, and Deng, 2010 & Padmanaban and Premkumar, 2015). The
ASR can be regarded as a standard classification problem which identifies word sequences from feature
sequences or speech waveforms.
Many issues have to be considered for the ASR to achieve satisfactory performance, for instance,
noisy environment, multi-model recognition, and multilingual recognition. Normally, the data should
be pre-processed using noise removal techniques before the speech recognition algorithms are applied.
Singh, Tripathy, and Anand (2014) reviewed some general approaches for noise removal and speech en-
hancement such as spectral subtraction, Wiener filtering, windowing, and spectral amplitude estimation.
22
A Review on Deep Learning Applications
Traditional machine learning algorithms, such as the SVM, and NNs, have provided promising results
in speech recognition (Hinton et. al, 2012). For example, Gaussian mixture models (GMMs) have been
used to develop speech recognition systems by representing the relationship between the acoustic input
and the hidden states of the hidden Markov model (HMM) (Baker et. al, 2009).
Since 2006, deep learning has emerged as a new research area of machine learning. As we just men-
tioned, deep learning algorithms can bring satisfactory results in feature extraction and transformation,
and they have been successfully applied to pattern recognition. Compositional models are generated using
DNNs models where features are obtained from lower layers. Through some well-known datasets including
the large vocabulary datasets, researchers have shown that DNNs could achieve better performance than
GMMs on acoustic modeling for speech recognition. Due to their outstanding performance in modeling
data correlation, the deep learning architectures are now replacing the GMMs in speech recognition
(Deng et. al, 2013 & Hinton et. al, 2012). Early applications of the deep learning techniques consist of
large vocabulary continuous speech recognition (LVCSR) and phone recognition (Dahl et. al, 2012).
Computer vision aims to make computers accurately understand and efficiently process visual data
like videos and images (Mohamed, Dahl, and Hinton, 2010 & Mohamed et. al, 2011). The machine is
required to perceive real-world high-dimensional data and produce symbolic or numerical information
accordingly. The ultimate goal of computer vision is to endow computers with the perceptual capability
of human. Conceptually, computer vision refers to the scientific discipline which investigates how to
extract information from images in artificial systems.
Recently, the development in deep learning architectures has provided novel approaches to the problem
of pattern recognition, which will be discussed in what follows.
1) Recognition: During the past few years, deep learning techniques have achieved tremendous prog-
ress in the domains of computer vision and pattern recognition, especially in areas such as object
recognition (Ballard and Brown, 1982).
2) Detection: Detection is one of the most widely known sub-domains in computer vision. It seeks
to precisely locate and classify the target objects in an image. In the detection tasks, the image is
scanned to find out certain special issues (Sonka, Hlavac, and Boyle, 2014).
3. Other Applications: Face alignment plays an important role in various visual applications such
as face recognition. However, for the extreme situations where the face images are taken, face
alignment may lead to difficulties during the analyzing process. Therefore, different models
for shape and appearance variation have been considered to solve this problem. Based on the
model used, the standard approaches can be roughly divided into three groups: the active ap-
pearance model, the constrained local model and the regression. Compared with the normal
regression based methods, the adaptive cascade deep convolution neural networks (ACDCNN),
proposed by Dong for facial point detection, have dramatically reduced the system complex-
ity (Dong and Wu, 2015). Dong et al. improved the basic DCNNs by exploiting an adaptive
manner for training with different network architectures. Experiment results showed that their
networks can achieve better performance than the DCNNs or other start-of-the-art methods.
It should be noted that the multi label image annotation is a hot topic in the field of computer
vision (Zhu et. al, 2015). Furthermore, deep learning techniques have recently been applied
23
A Review on Deep Learning Applications
to the content-based image retrieval applications (Emad, Yassine, and Fahmy, 2015). More
specifically, to address the cross-modal retrieval tasks, a correspondence AE (Corr-AE) was
introduced by correlating the hidden representations of two uni-modal AEs (Feng, Wang, and
Li, 2014).
Virtual Assistants
The most popular application of deep learning is virtual assistants ranging from Alexa to Siri to Google
Assistant. Each interaction with these assistants provides them an opportunity to learn more about your
voice and accent, thereby providing you a secondary human interaction experience (López, Quesada,
and Guerrero, n.d.). Virtual assistants use deep learning to know more about their subjects ranging from
your dine-out preferences to your most visited spots or your favorite songs. They learn to understand
your commands by evaluating natural human language to execute them (Chung, Iorga, and Lee, n.d.).
Another capability virtual assistants are endowed with is to translate your speech to text, make notes
for you, and book appointments. Virtual assistants are literally at your beck-and-call as they can do ev-
erything from running errands to auto-responding to your specific calls to coordinating tasks between
you and your team members. With deep learning applications such as text generation and document
summarizations, virtual assistants can assist you in creating or sending appropriate email copy as well
(Tulshan et. al, 2019).
Deep Learning is going to transform the worksplace, but often in ways that pose a threat to jobs.
There’s a widespread notion that many tasks will be automated, and there’ll be no place for humans.
Healthcare
According to NVIDIA, “From medical imaging to analyzing genomes to discovering new drugs, the
entire healthcare industry is in a state of transformation and GPU computing is at the heart. GPU-
accelerated applications and systems are delivering new efficiencies and possibilities, empowering
physicians, clinicians, and researchers passionate about improving the lives of others to do their best
work.” Helping early, accurate and speedy diagnosis of life-threatening diseases, augmented clinicians
addressing the shortage of quality physicians and healthcare providers, pathology results and treatment
course standardization, and understanding genetics to predict future risk of diseases and negative health
episodes are some of the Deep Learning projects picking up speed in the Healthcare domain (Miotto
et. al, 2017). Read missions is a huge problem for the healthcare sector as it costs tens of millions of
dollars in cost. But with the use of deep learning and neural networks, healthcare giants are mitigating
health risks associated with readmissions while bringing down the costs (Pitoglou, 2018). AI is also
being exceedingly being used in clinical researches by regulatory agencies to find cures to untreatable
diseases but physicians skepticism and lack of a humongous dataset are still posing challenges to the
use of deep learning in medicine.
Self-Driving Cars
Deep Learning is the force that is bringing autonomous driving to life. A million sets of data are fed to a
system to build a model, to train the machines to learn, and then test the results in a safe environment. The
Uber AI Labs at Pittsburg is not only working on making driverless cars humdrum but also integrating
24
A Review on Deep Learning Applications
several smart features such as food delivery options with the use of driverless cars. The major concern
for autonomous car developers is handling unprecedented scenarios (Bojarski et. al, 2017a). A regular
cycle of testing and implementation typical to deep learning algorithms is ensuring safe driving with
more and more exposure to millions of scenarios. Data from cameras, sensors, geo-mapping is helping
create succinct and sophisticated models to navigate through traffic, identify paths, signage, pedestrian-
only routes, and real-time elements like traffic volume and road blockages. According to Forbes, MIT is
developing a new system that will allow autonomous cars to navigate without a map as 3-D mapping is
still limited to prime areas in the world and not as effective in avoiding mishaps (Bojarski et. al, 2017b).
CSAIL graduate student Teddy Ort said, “The reason this kind of ‘map-less’ approach hasn’t really
been done before is because it is generally much harder to reach the same accuracy and reliability as
with detailed maps. A system like this that can navigate just with onboard sensors shows the potential
of self-driving cars being able to actually handle roads beyond the small number that tech companies
have mapped” (Bojarski et. al, 2016).
Extensive use of deep learning in news aggregation is bolstering efforts to customize news as per readers.
While this may not seem new, newer levels of sophistication to define reader personas are being met to
filter out news as per geographical, social, economical parameters along with the individual preferences
of a reader. Fraud news detection, on the other hand, is an important asset in today’s world where the
internet has become the primary source of all genuine and fake information. It becomes extremely hard
to distinguish fake news as bots replicate it across channels automatically. The Cambridge Analytica is
a classic example of how fake news, personal information, and statistics can influence reader perception
(Bhartiya Janta Party vs Indian National Congress), elections (Donald Trump Digital Campaigns), and
exploit personal data (Facebook data for approximately 87 million people was compromised). Deep
Learning helps develop classifiers that can detect fake or biased news and remove it from your feed and
warn you of possible privacy breaches. Training and validating a deep learning neural network for news
detection is really hard as the data is plagued with opinions and no one party can ever decide if the news
is neutral or biased (Najafabadi et. al, 2015 & Wang and Xu, 2018).
Entertainment
Wimbledon 2018 used IBM Watson to analyse player emotions and expressions through hundreds of hours
of footage to auto-generate highlights for telecast. This saved them a ton of effort and cost. Thanks to
Deep Learning, they were able to factor in audience response and match or player popularity to come up
with a more accurate model (otherwise it would just have highlights of the most expressive or aggressive
players). Netflix and Amazon are enhancing their deep learning capabilities to provide a personalized
experience to its viewers by creating their personas factoring in show preferences, time of access, history,
etc. to recommend shows that are of liking to a particular viewer. VEVO has been using deep learning
to create the next generation of data services for not only personalized experiences for its users and
subscribers, but also artists, companies, record labels, and internal business groups to generate insights
based on performance and popularity. Deep video analysis can save hours of manual effort required for
25
A Review on Deep Learning Applications
audio/video sync and its testing, transcriptions, and tagging. Content editing and auto-content creation
are now a reality thanks to Deep Learning and its contribution to face and pattern recognition. Deep
Learning AI is revolutionizing the filmmaking process as cameras learn to study human body language
to imbibe in virtual characters (Thomas et, al, 2018 & Yang, Yao, Wang, 2018 & Aznan et. al, 2018).
Fraud Detection
Another domain benefitting from Deep Learning is the banking and financial sector that is plagued
with the task of fraud detection with money transactions going digital. Autoencoders in Keras and Ten-
sorflow are being developed to detect credit card frauds saving billions of dollars of cost in recovery
and insurance for financial institutions (El Bouchti, Chakroun, and Okar, 2017). Fraud prevention and
detection are done based on identifying patterns in customer transactions and credit scores, identifying
anomalous behavior and outliers. Classification and regression machine learning techniques and neural
networks are used for fraud detection. While machine learning is mostly used for highlighting cases of
fraud requiring human deliberation, deep learning is trying to minimize these efforts by scaling efforts
(Moro, Cortez, and Rita, 2015).
Personalizations
Every platform is now trying to use chatbots to provide its visitors with personalized experiences with a
human touch. Deep Learning is empowering efforts of e-commerce giants like Amazon, E-Bay, Alibaba,
etc. to provide seamless personalized experiences in the form of product recommendations, personalized
packages and discounts, and identifying large revenue opportunities around the festive season. Even race
in newer markets is done by launching products, offerings, or schemes that are more likely to please
the human psyche and lead to growth in micro markets. Online self-service solutions are on the rise
and reliable workflows are making even those services available on the internet today that were only
physically available at one time. Robots specialized in specific tasks are personalizing your experiences
real-time by offering you the most suited services whether it is insurance schemes or creating custom
burgers (Schneider and Handali, 2019).
Autism Detection
Speech disorders, autism, and developmental disorders can deny a good quality of life to children suf-
fering from any of these problems. An early diagnosis and treatment can have a wonderful effect on the
physical, mental, and emotional health of differently-abled children. Hence, one of the noblest applica-
tions of deep learning is in the early detection and course-correction of these problems associated with
infants and children (Tariq et. al, 2019). This is a major difference between machine learning and deep
learning where machine learning is often just used for specific tasks and deep learning on the other
hand is helping solve the most potent problems of the human race. Researchers at the Computer Science
and Artificial Intelligence Laboratory at MIT and Massachusetts General Hospital’s Institute of Health
Professions have developed a computer system that can identify language and speech disorders even
before kindergarten when most of these cases traditionally start coming to light. The researchers evalu-
ated the system’s performance using a standard measure called area under the curve, which describes
26
A Review on Deep Learning Applications
the tradeoff between exhaustively identifying members of a population who have a particular disorder.
They use residual analysis that identifies the correlation between age, gender, and acoustic features of
their speech to limit false positives (Heinsfeld et. al, 2018). Autism is often detected by combining it
with cofactors such as low birth weight, physical activity, body mass index, learning disabilities, etc.
CHALLENGES
In the above section we took the review of various luring applications of deep learning. These applica-
tions are exciting people about the opportunities this field brings. Although it seems to be attracting
the industries, but researcher are facing challenging problems to handle the exponentially growing data,
overfitting issues in neural networks, minimization of hyper-parameter, lack of highly configured pro-
cessors, processing in hidden layers etc etc (Shrivataya, n.d.).
Voluminous Data
To increase the accuracy for getting desired result, Deep learning algorithms are trained on large data set
like a human brain needs a lot of experiences to learn and deduce information. Researchers feed terabytes
of data for the algorithm to learn which is really a time-consuming process and requires tremendous data
processing capabilities. The neural network becomes more complicated as the number of parameters
used in training goes on increasing.
Overfitting Problem
At times, there is a sharp difference in error occurred in training data set and the error encountered in
a new unseen data set. It occurs in complex models, such as having too many parameters relative to the
number of observations. The efficacy of a model is judged by its ability to perform well on an unseen
data set and not by its performance on the training data fed to it.
Minimisation of Hyperparameter
Hyperparameters are the parameters whose value is defined prior to the commencement of the learn-
ing process. Changing the value of such parameters by a small amount can cause a large change in the
performance of your model.
Relying on the default parameters and not performing Hyperparameter Optimization can have a
significant impact on the model performance. Also, having too few hyperparameters and hand tuning
them rather than optimizing through proven methods is also a performance driving aspect.
High-Performance Hardware
Deep learning algorithms requires a lot of data for training models. To perform a task to solve real
world problems, the machine needs to be equipped with adequate processing power. To ensure better
efficiency and less time consumption, data scientists switch to multi-core high performing GPUs and
similar processing units. These processing units are costly and consume a lot of power.
27
A Review on Deep Learning Applications
Industry level Deep Learning systems require high-end data centers while smart devices such as
drones, robots other mobile devices require small but efficient processing units. Deploying Deep Learn-
ing solution to the real world thus becomes a costly and power consuming affair.
We know our model parameters, we feed known data to the neural networks and how they are put to-
gether. But we usually do not understand how they arrive at a particular solution. Neural networks are
essentially Blackboxes and researchers have a hard time understanding how they deduce conclusions.
The lack of ability of neural networks for reason on an abstract level makes it difficult to implement
high-level cognitive functions. Also, their operation is largely invisible to humans, rendering them un-
suitable for domains in which verification of process is important.
Deep Learning models, once trained, can deliver tremendously efficient and accurate solution to a specific
problem. However, in the current landscape, the neural network architectures are highly specialized to
specific domains of application.
CONCLUSION
In this Chapter, we have touched most of the sensitive applications of deep learning including computer
vision, pattern recognition, speech recognition, virtual assistant, driverless car, personisation, etc. Along
with applications we also put light on the challenges in deep learning for the successful implementation
of these applications. With the fast development of hardware resources and computation technologies,
we are confident that deep neural networks will receive wider attention and find broader applications in
the future. Still there is vast scope for further extensive research. In particular, more work is necessary
on how we can adapt Deep Learning algorithms for problems associated with Big Data with high di-
mensionality, sparcity, streaming data analysis, scalability of Deep Learning models, improved formula-
tion of data abstractions, distributed computing, semantic indexing, data tagging, information retrieval,
criteria for extracting good data representations, and domain adaptation. Future works should focus on
addressing one or more of these problems often seen in Big Data.
REFERENCES
Baker, J. M., Deng, L., Glass, J., Khudanpur, S., Lee, C. H., Morgan, N., & Shaughnessy, D. O. (2009).
Developments and directions in speech recognition and understanding, part 1 [dsp education]. IEEE
Transactions on Signal Processing Magazine, 26(3), 75–80. doi:10.1109/MSP.2009.932166
Ballard, D. H., & Brown, C. M. (1982). Computer vision. Englewood Cliffs, NJ: Prenice-Hall.
28
A Review on Deep Learning Applications
Bojarski, Asa, Colak, & Czarkowski. (2017). Analysis and control of multiphase inductively coupled
resonant converter for wireless electric vehicle charger applications. IEEE Transactions on Transporta-
tion Electrification, 3(2), 312-320.
Bojarski, M., Choromanska, A., Choromanski, K., Firner, B., Jackel, L., & Muller, U. (2016). Visual-
backprop: visualizing cnns for autonomous driving. arXiv preprint arXiv:1611.05418
Bojarski, M., Yeres, P., Choromanska, A., & Choromanski, K. (2017). Explaining how a deep neural
network trained with end-to-end learning steers a car. arXiv preprint arXiv:1704.07911
Dahl, G. E., Yu, D., Deng, L., & Acero, A. (2012). Context-dependent pre-trained deep neural networks
for large-vocabulary speech recognition. IEEE Transactions on Audio, Speech, and Language Process-
ing, 20(1), 30–42. doi:10.1109/TASL.2011.2134090
Deng, L., Li, J., Huang, J. T., Yao, K., Yu, D., Seide, F., . . . Acero, A. (2013). Recent advances in deep
learning for speech research at Microsoft. Acoustics, Speech and Signal Processing (ICASSP), 2013
IEEE International Conference on, 8604–8608. 10.1109/ICASSP.2013.6639345
Dong, Y., & Wu, Y. (2015). Adaptive cascade deep convolutional neural networks for face alignment.
Computer Standards & Interfaces, 42, 105–112. doi:10.1016/j.csi.2015.06.004
El Bouchti, A., Chakroun, A., Abbar, H., & Okar, C. (2017). Fraud detection in banking using deep rein-
forcement learning. Seventh International Conference on Innovative Computing Technology (INTECH),
58-63. 10.1109/INTECH.2017.8102446
Emad, O., Yassine, I. A., & Fahmy, A. S. (2015). Automatic localization of the left ventricle in cardiac
mri images using deep learning. Engineering in Medicine and Biology Society (EMBC), 2015 37th An-
nual International Conference of the IEEE, 683–686. 10.1109/EMBC.2015.7318454
Feng, F., Wang, X., & Li, R. (2014). Cross-modal retrieval with correspondence autoencoder. Proceed-
ings of the ACM International Conference on Multimedia, 7–16.
Heinsfeld, A. S., Franco, A. R., Craddock, R. C., Buchweitz, A., & Meneguzzi, F. (2018). Identification
of autism spectrum disorder using deep learning and the ABIDE dataset. NeuroImage. Clinical, 17,
16–23. doi:10.1016/j.nicl.2017.08.017 PMID:29034163
Hinton, G. E., Deng, L., Yu, D., Dahl, G. E., Mohamed, A. R., Jaitly, N., ... Sainath, T. N. (2012). Deep
neural networks for acoustic modeling in speech recognition: The shared views of four research groups.
IEEE Signal Processing Magazine, 29(6), 82–97. doi:10.1109/MSP.2012.2205597
Hinton, G. E., Deng, L., Yu, D., Dahl, G. E., Mohamed, A. R., Jaitly, N., ... Sainath, T. N. (2012). Deep
neural networks for acoustic modeling in speech recognition: The shared views of four research groups.
IEEE Signal Processing Magazine, 29(6), 82–97. doi:10.1109/MSP.2012.2205597
Jaitly, N., & Hinton, G. (2011). Learning a better representation of speech soundwaves using restricted
Boltzmann machines. Acoustics, Speech and Signal Processing (ICASSP), 2011 IEEE International
Conference on, 5884–5887. 10.1109/ICASSP.2011.5947700
LeCun, Y., Bengio, Y., & Hinton, G. (2015). Deep learning. Nature, 521(7553), 436–444. doi:10.1038/
nature14539 PMID:26017442
29
A Review on Deep Learning Applications
Miotto, Wang, Wang, Jiang, & Dudley. (2017). Deep learning for healthcare: Review, opportunities and
challenges. Briefings in Bioinformatics, 1–11.
Mohamed, A., Dahl, G. E., & Hinton, G. E. (2012). Acoustic modeling using deep belief networks. IEEE
Transactions on Audio, Speech, and Language Processing, 20(1), 14–22. doi:10.1109/TASL.2011.2109382
Mohamed, A., Sainath, T. N., Dahl, G., Ramabhadran, B., Hinton, G. E., & Picheny, M. A. (2011). Deep
belief networks using discriminative features for phone recognition. IEEE International Conference on
Acoustics, Speech and Signal Processing (ICASSP), 5060–5063. 10.1109/ICASSP.2011.5947494
Mohamed, A., Yu, D., & Deng, L. (2010). Investigation of full-sequence training of deep belief networks
for speech recognition. INTERSPEECH, 2846–2849.
Mohamed, Dahl, & Hinton. (2009). Deep belief networks for phone recognition. Nips workshop on deep
learning for speech recognition and related applications, 1(9), 39.
Mohamed, Yu, & Deng. (2010). Investigation of full-sequence training of deep belief networks for speech
recognition. INTERSPEECH, 2846–2849.
Moro, S., Cortez, P., & Rita, P. (2015). Business intelligence in banking: A literature analysis from
2002 to 2013 using text mining and latent Dirichlet allocation. Expert Systems with Applications, 42(3),
1314–1324. doi:10.1016/j.eswa.2014.09.024
Najafabadi, Villanustre, Khoshgoftaar, Seliya, Wald, & Muharemagic. (2015). Deep learning applica-
tions and challenges in big data analytics. Journal of Big Data, 2(1). doi:10.118640537-014-0007-7
Nik, K. N. A., Bonner, S., Connolly, J., Al Moubayed, N., & Breckon, T. (2018). On the Classification
of SSVEP-Based Dry-EEG Signals via Convolutional Neural Networks. Systems Man and Cybernetics
(SMC) 2018 IEEE International Conference on, 3726-3731.
Padmanabhan, J., & Johnson Premkumar, M. J. (2015). Machine learning in automatic speech recogni-
tion: A survey. IETE Technical Review, 32(4), 240–251. doi:10.1080/02564602.2015.1010611
Pitoglou. (2018). Machine Learning in Healthcare, Introduction and Real World Application Consider-
ations. International Journal of Reliable and Quality E-Healthcare, 7(2), 27-36.
Schneider & Handali. (2019). Personalized explanation in machine learning. CORR Journal.
Shrivatava, P. (n.d.). Challenges in Deep Learning. Retrieved from https://hackernoon.com/challenges-
in-deep-learning-57bbf6e73bb
Singh, M. T., & Anand, R. (2014). Subjective and objective analysis of speech enhancement algorithms
for single channel speech patterns of indian and english languages. IETE Technical Review, 31(1), 34–46.
doi:10.1080/02564602.2014.890840
Sonka, M., Hlavac, V., & Boyle, R. (2014). Image processing, analysis, and machine vision. Cengage
Learning.
Tariq, Fleming, Schwartz, Dunlap, Corbin, Washington, … Wall. (2019). Works citing “Detecting De-
velopmental Delay and Autism Through Machine Learning Models Using Home Videos of Bangladeshi
Children: Development and Validation Study. J Med Internet Res., 21(4).
30
A Review on Deep Learning Applications
Thomas, J., Comoretto, L., Jin, J., Dauwels, J., Cash, S. S., & Westover, M. B. (2018). EEG CLassification
Via Convolutional Neural Network-Based Interictal Epileptiform Event Detection. IEEE Engineering in
Medicine and Biology Society (EMBC) 2018 40th Annual International Conference of the, 3148-3151.
10.1109/EMBC.2018.8512930
Tulshan, A., & Dhage, N. (2019). Survey on Virtual Assistant: Google Assistant, Siri, Cortana, Alexa.
4th International Symposium SIRS 2018, Bangalore, India. 10.1007/978-981-13-5758-9_17
Wang, Y., & Xu, W. (2018). Leveraging deep learning with LDA-based text analytics to detect automobile
insurance fraud. In Decision Support Systems. Elsevier.
Yang, J., Yao, S., & Wang, J. (2018). Deep Fusion Feature Learning Network for MI-EEG Classification.
Access IEEE, 6, 79050–79059. doi:10.1109/ACCESS.2018.2877452
Zhu, S., Shi, Z., Sun, C., & Shen, S. (2015). Deep neural network based image annotation. Pattern
Recognition Letters, 65, 103–108. doi:10.1016/j.patrec.2015.07.037
31