American Journal of Management Science and Engineering

2018; 3(5): 38-43

http://www.sciencepublishinggroup.com/j/ajmse
doi: 10.11648/j.ajmse.20180305.11
ISSN: 2575-193X (Print); ISSN: 2575-1379 (Online)

Review Article
Deep Learning Applications in Business Activities
Zhiqiao Zhong1, Xu Zhuang2
1
Guangdong Country Garden School, Foshan, China
2
School of Science, Tianjin University, Tianjin, China

Email address:

To cite this article:

Zhiqiao Zhong, Xu Zhuang. Deep Learning Applications in Business Activities. American Journal of Management Science and Engineering.
Vol. 3, No. 5, 2018, pp. 38-43. doi: 10.11648/j.ajmse.20180305.11

Received: September 14, 2018; Accepted: October 9, 2018; Published: October 24, 2018

Abstract: With huge improvement of computer calculation abilities, deep learning method have great potential applications in
wider business fields. With the data provided by many companies, deep learning method has achieved great success in the aspect
of reducing expense of companies’ activities, and brought unexpected profits. This article explains the basic principles of deep
learning, introduce its main scope of application, and explore its application in business. This article provides a more pertinent
assessment by querying the data and relevant reports of the enterprises engaged in this work. This article introduces and explain
the mathematical equations for the deep learning, and discuss about different types of Neural Network including Feed-forward
Neural Networks and Recurrent Neural Networks. Based on the types of deep learning model, this article demonstrates the
applications of deep learning method in business activities based on concrete examples. The applications include Customer
Service, Sales, Marketing, Daily Operation and Risks Management. Through the relevant queries, this article indicates a lot of
convincing data and examples to prove that deep learning in business activities has a good effect. This is instructive and helps
business practitioners to consider a new and more effective way to increase revenue or save costs. Through the relevant queries,
this article found a lot of convincing data and examples to prove that deep learning in business activities has a good effect.
Studying from the principle of deep learning to the applications in real business situation, deep learning is coherently introduced
to the audience.
Keywords: Deep Learning, Business, Neural Networks

outcomes can be created in many fields utilizing its virtues.

1. Introduction As mentioned above, deep learning can be applied in
Deep learning is a form of machine learning that enables complicated models which cannot be achieved by any other
computers to learn from experience and understand the world learning algorithms, it is widely used in fields like image
in terms of a hierarchy of concepts. Since the term of “Deep identification, Pilotless driving and natural language
Learning” was first introduced by Rina Dechter to the field of recognition. [3] The main ideas in the applications of the deep
machine learning in 1986, it received extensive attention from learning are all similar — use a large scale of data as simple to
many fields. [1] train the model then applied the finished model to perform
As a part of machine learning, deep learning has its own predictions. [4] Therefore, it is obvious that deep learning has
distinct demerits. One of its outstanding advantages is that it a hopeful application prospect in many fields.
can commendably dealing with complex work that may be In modern business activities, companies always face vast
hard to achieve by other techniques. On the other hand, it and irregular data in their daily operation while huge business
requires massive sets of data and advanced computer opportunities may be hidden behind. Either companies do not
hardware to operate. [2] While in many fields nowadays, big know how to extract useful information out of such data, or
data is easy to obtain, meanwhile the development of employees waste their time on some repeated work. Both of
technology provides more powerful computers. Therefore, the issues can be solved by deep learning. It can analyze
overcoming the shortcoming of deep learning, exceptional historical data to conceive new perspective or replace human
39 Zhiqiao Zhong and Xu Zhuang: Deep Learning Applications in Business Activities

in reduplicative work to lower human cost. In other words, achievements can already be seen. [5]
deep learning injects vigor into business activities, and some

Figure 1. Deep Learning Category.

when it comes to more complex problems, other operations

2. Deep Learning Frameworks may be needed, XOR gate for instance. At this time, one layer
of neurons may not be able to solve it, and therefore, multiple
To understand the frameworks of deep learning, artificial layers of neurons are required in order to fit arbitrary complex
neural network (ANN) is definitely starting point. [6] ANN is continuous functions. This is so-called “deep learning”, the
one of the main tools to achieve machine learning. ANN is a system calls for multiples layers of neurons. [8]
brain-inspired system created by human to imitate the way
that animals learn, containing mainly three layers: input data 3. Types of Neural Network
for the first layer, hidden layer transforming the input data into
the output data, while output data is presented in the third layer. Deep neural network with numerous layers of neurons
Consisting of these three layers, neuron is the functional comes with different ways of combinations of layers.
foundation. [7] Different types of neural networks are applied to different
tasks and complexity.
Let’s first consider the simplest case. Each layer is fully
interconnected to its adjacent layers and has no peer or
cross-layer connections. The data only flows from input to
output directly. This artificial neural networks are called
“Feed-forward Neural Networks” (FNN). In such case, an
input x passes through equation f x and gets the output y.
Weight and threshold of each neuron are adjusted to fit the
situation. [9, 10]

Figure 2. Deep learning Model.

The neuron model receives n input signals

x , x , x … x , which is multiplied by a corresponding
weight ω , ω , ω … ω and summed. The result is then
processed by the activation equation, Sigmoid function in the
typical case. The result can be compressed in the interval (0,
1), then outputted based on its unique parameter, threshold θ.
The process can be expressed by the equation below:

y f ωx θ

In the process of neural network training, the two

parameters are adjusted to fit the continuous inputted data.
A single neuron model can be applied to simple problem
Figure 3. Feedforward Neural network.
operating only AND gates, OR gates and NOT gates. However,
 American Journal of Management Science and Engineering 2018; 3(5): 38-43 40

When feedback is extended to FNN, it turns to a more Nowadays, there are some companies which have already
prevailing type, “Recurrent Neural Network” (RNN). [11] developed the deep learning model and applied it in the real
Different from FNN, RNN memorizes the previous data life. For instance, DigitalGenius founded in 2013 focuses on
inputted and makes prediction for future data. While applying deep learning model to customer service.
sometimes the initial data maybe wrong and therefore makes DigitalGenius developed the "the first solution to enable
wrong prediction, this will affect the parameters to change. To end-to-end case resolution of your common repetitive
avoid such conditions, neural network can back propagate journeys without human intervention." which is called
such data to the previous layer and adjust the weight for it to "AutoPilot". During customer service, replying repeated
minimize the effect brought by it for the final result. This questions is always the one that occupies the most time of
process is called gradient descent. With such system, RNN can staffs, while AutoPilot is the perfect tool for it. Companies
be applied to more situations. which employ AutoPilot for their platform can train it by
inputting samples. This enables AutoPilot to deal with the
simple questions raised by consumers, and let staffs to spend
their time on consumers who have individual requirement.
Based on the development of AutoPilot, DigitalGenius later
launched a more advanced new product -- CoPilot. After
receiving massive historical data, CoPilot can basically be
qualified for the job of human staffs. In practice, after
consumers raising a question, CoPilot will first analyze the
problem and give out an answer. Then it will calculate the
confidence level of the answer. If such confidence level
exceeds a threshold, it can be directly outputted to customer.
Otherwise, the answer will be sent as an advice for human
staff who can choose either to approve the answer or to reedit
Figure 4. Gradient Descent.
it. During the process of giving answers and getting feedback,
the model becomes more and more accurate itself. Again,
4. Application of Deep Learning in CoPilot improves efficiency, lower the cost and average
handling time (AHT).
Business Numerous Companies choose to employ products from
The presence of huge amounts of data in daily business DigitalGenius for their own, like TravelBird, StarOfService
activities is a perfect opportunity to utilize deep learning and KLM Royal Dutch Airlines. Take TravelBird for instance,
models. As showed above, there are different types of Neutral after the Interference of DigitalGenius, the CSAT (Customer
Networks which can be applied to different cases in business Satisfaction Score) soared to 90% after nine months, AHT
area. The following sections will explain from five (Average Handling Time) dropped by 30%, and the sales
perspectives of how deep learning can overcome some increased by 230% in a year. There is an optimistic prediction
problems in business and other applications in business by on the future applications of such deep learning models.
some concrete examples.
4.2. Sales
4.1. Customer Service
Selling job is closely connected to the profit of a company.
It is known that customer service is a vital section in In order to let more consumers to pay for the product, besides
providing customers great experience of purchasing and improving products quality and customer service, appropriate
increasing company’s profit. Inversely, the cost of a terrible sales effort is also essential. Outstanding sales tools can aim at
customer service is expensive. U. S. companies lose more than consumers who are more willing to purchase, and increase the
$62 billion annually due to poor customer service. While total profit. While in real life, there are some obstacles for
improving the level of customer service is not an easy work. human staffs to recognize the best sales tool for consumers,
Businesses spend $1.3 trillion on 265 billion customer service introducing deep learning models can help out.
calls each year. Such big amounts of money are mainly for the Data of potential can be easily obtained by companies, such
salary and training fee for customer service staffs. But such as income level, education background and hobbies. [13] At
job is actually mechanical, staffs spend great mass of time in the same time, the historical record of successful sales is
answering repeated and simple questions. With all the available. Crucial information is actually hidden behind those
understanding of deep learning, it makes people think that massive data. For a normal sales person, it is hard to combine
training deep learning model can replace customer services messy data together and extract the useful part. Therefore, the
staffs. In this way, companies are able to lower the cost, and assist of deep learning model is necessary in order to filter the
artificial model has greater efficiency than human which potential customers. In today’s marketplace, one company is
improve consumers’ experience and in turn increases doing exactly the same job.
companies’ profit. [12] Salesforce developed intelligence -- Sales Cloud which
assists salesman for better achievement. Its features consist of
41 Zhiqiao Zhong and Xu Zhuang: Deep Learning Applications in Business Activities

three different parts: Einstein Lead Scoring, Einstein excellently. Finally, deep learning model will both save the
Opportunity & Account Insights and Einstein Activity cost the companies invest into marketing and improve
Capture. Firstly, Einstein Lead Scoring plays an important role: efficiency. Hundreds of corporations “use Persado to optimize
determine the customers. It obtains and analyzes the past data marketing campaigns, resulting in a 49.5% average lift in
of the companies to build a model of identifying the potential conversions” According to Forrester Analyst Shar VanBoskirk,
consumers. Based on the result provided by Einstein Lead “By 2021, CMOs will spend nearly $119 billion on search
Scoring, salesman can select the potential consumers with marketing, display advertising, online video, and email
higher priority which increases the success rate. After setting marketing.” Increasing conversion rate definitely brings more
up the customer list, company can continue using Einstein business opportunity to companies. While with the appearance
Opportunity & Account Insights which also analyzes based on of Persado, the cost will be largely reduced.
past data. It finds the sales way that is best fitted to the
companies and provides instant news, like customer sentiment 4.4. Daily Operation
or competitor involvement. Finally, Einstein Activity Capture There are many fields in daily operations, the following
connects to salesmen’s emails and calendars and adds emails paragraphs will mainly focus on human resource management
and events directly which saves their times. and internal communications. These jobs can be minor for
Customers of Salesforce report 39% increase in sales small companies, while with the expansion of companies,
productivity. Such accurate prediction of potential customers these jobs become harder and harder. Therefore, deep learning
brings convenience to salesman and profit to corporations at model can lower the error rate and human cost. Three
the same time. examples will be raised to illustrate the application of deep
4.3. Marketing learning model in daily operations. [15]
The first task is the arrangement of regular meetings. Dealing
In modern business activities, the major problem is the with complicated meetings is a tedious work, while a product
selection of marketing methods. One product always faces (called x. ai) based on the theory of deep learning solves the
different groups of target audience, while single marketing problem effectively. x. ai provides trained deep learning model
method does not work well for all the people. The presence of as virtual assistant. It can understand the email content of clients
massive unstructured trading records will be an ideal place for and help them to do the schedule. The processes behind it is that
deep learning model. A company named Persado has a great when clients sending email to others to arrange work, they need
progress in this field. [14] to copy one to the virtual assistant and provide assess rights at
Persado focuses on building Language Cloud which is to the same time. x. ai can then arrange all the meetings for clients
design language that resonates with every customers and with detailed places, time and people and reminds clients when
achieves better marketing effect. In October 2017, Persado the times come. In other words, x. ai as a personal virtual
released their product Persado One for the first time. Persado One assistant does not replace other’s jobs while it offers
learned from the past data to deconstruct both language elements convenience for people who need to communicate a lot with
and emotions for customers. Persado One assigns profiles for others but without a personal assistant.
every customer to record individuals’ situation, like age, income Recruitment event is the second task for deep learning
and education background. After gathering these data, Persado model, including reading resumes and having interviews. Mya,
One will choose the most appropriate one for every customers. a A. I. recruiting assistant, can help to pre-screen candidates. It
Then, the model will be modulated based on the real response is able to collect the basic information about and preliminarily
from customers to certain language and emotions. assess candidates by communicating with them. Meanwhile, it
Persado One can provide more benefits compared to the also answers some questions from candidates like arranging
traditional way of marketing. First, this brand-new way of interview. Based on Neuro Linguistic Program (NLP), it
marketing focuses more on customers’ personal emotion effectively lower work intensity of HR people. According to
which can lead to higher engagement of customers. For the the official website of Mya, there is a 79% decrease in the time
same product, because of the physical restriction, traditional to hire when a recruiter works with Mya.
way can only emphasize one perspective or a brief over-all The final one is the documenting during meetings. For
introduction of the product when facing multiple groups of business, meetings is one of the main parts of the works.
potential customers. While different groups of people value Human forces are always needed to record the content of
different aspects of the product, it is difficult to attract all the meetings, while committing faults is inevitable for humans.
groups. On the other hand, Persado one takes aim at every However, Clarke, as a professional assistant trained by deep
individual to customize marketing language which closely learning, can deal with this problem. Only by emailing or
connected with every customers who will be more willing to calling Clarke to invite it to attend the meeting, it can record
purchase. On top of that, Persado One can insert photos with the audio of meetings and compile a note automatically and
their language. Once the model combines words and pictures, silently. This helps the business workers to a great extent.
an integral marketing content will be displayed. Second, the
marketing language created by Persado One can be applied to 4.5. Risks Management
different area, like advertising with emails, website and social Risk management is the vital part in business activity as it
media. This can replace people generated advertisements
 American Journal of Management Science and Engineering 2018; 3(5): 38-43 42

goes through all processes when a project is carrying on. Deep learning model has more applications in business
Because business activity is always accompanied by risks, activities besides from five perspectives mentioned above.
knowing and mitigating risks is significant. In traditional More importantly, this article discusses the basic theory of
commercial pattern, risk management includes identifying the deep learning, which refers to the best application of deep
condition and possibility of the occurrence of risk and seeking learning in what kind of data sets and actual scenarios. It is
the optimal method to avoid it. The uncertainty of task makes pointed out that deep learning is very suitable for business
it difficult for human to analyze even with experience and activities. At the same time, this article discusses the
specific economic theory. However, deep learning model can successful application of deep learning in business activities,
still do a good job dealing with risks. [16, 17] which will give every practitioner inspiration. In their work,
One typical example of risks is customer defection. people can use the tool of deep learning very well and get
Predicting and controlling the customers lost is vital as those good results.
customers is possible to flow into its competitors’ hand. It is clear that deep learning will has a considerable effect in this
Studies by Bain & Company, along with Earl Sasser of the “Big Data Time”. [19] It not only can replace human in some
Harvard Business School, have measured that even a 5 percent repeated or time-wasting work, but also can analyze the trend that
increase in customer retention can lead to an increase in profits human work cannot achieve. Deep learning has only been
somewhere between 25 and 95 percent. developed through the recent years, but the achievements are
Deep learning can manage abundant non-structural data. incredible. On the other hand, it is inevitable that deep learning
Using this feature, Federico Castanedo published a paper creates unemployment. It is hard to predict that what will deep
called “Using Deep Learning to Predict Customer Churn in a learning bring us in the future since in a fast-changing time, but it
Mobile Telecommunication Network” on Wise Athena using will definitely have great impacts on us – with both positive and
the theory of deep learning. Castanedo points out a custom negative effects. People have to find a balance point between the
churn prediction model. When massive past information of usage of deep learning.
clients is inputted, this model can predict the actions of
clients in every process of business activity, including future Acknowledgements
actions.
Deep learning model can analyze information involved We thank Prof. Chen for providing this research topic, and
multiple layers of information and statistics, which can hardly instructive discussions during the entire project.
be done by human work. Mentioned by Castanedo in the
article that “We use billions of call records from an enterprise
business intelligence system and present our current work References
towards using deep learning for predicting churn in a prepaid
mobile telecommunication network. To the best of our [1] Rina Dechter, National Conference on Artificial Intelligence.
Philadelphia, 178-185, 1983
knowledge this is the first work reporting the use of deep
learning for predicting customer churn. On average, our model [2] Lecun Y, Bengio Y, Hinton G, Nature, 2015, 436-444, 2015.
achieves 77.9% AUC1 on validation data, significantly better
[3] Deng, Li, and D. Yu, Foundations & Trends in Signal
than our prior best performance of 73.2% obtained with Processing, 197-387, 2014
random forests and an extensive custom feature engineering
applied to the same datasets.” [4] Schmidhuber J, Neural Networks, 85-117, 2015
Although deep learning model like a black box operation, [5] Howard, Jeremy, ACM SIGKDD International Conference on
which cannot give out reasons for the result, it still has a Knowledge Discovery and Data Mining, 1135-1135, 2013
significant meaning in real applications. Moreover, inspired
by this problem of customer churn, we can imagine that many [6] Yao, Xin, Proceedings of the IEEE, 1423-1447, 1999
business risks can be controlled by introducing deep learning. [7] Hassoun, Mohamad H, Proceedings of the IEEE, 84 (6) :906,
For example, if the partnership between enterprises breaks 2002
down and has a huge impact on both sides, can a deep learning
[8] Widrow, Bernard, and M. A. Lehr, Proceedings of the IEEE,
model control it? [18] 1415-1442, 2002

[9] Ahn, B. S., S. S. Cho, and C. Y. Kim, Expert Systems with

5. Conclusions Applications, 65-74, 2000
Deep learning can be applied to various areas of business [10] Sanger, Terence D. Neural Networks, 459-473, 1989
activities, in customer service can help practitioners more
reasonable and efficient communication with customers, in [11] Williams, Ronald J., and D. Zipser, MIT Press, 270-280, 1989.
marketing and sales can improve sales, increase profits, in [12] Xu, Anbang, et al. CHI Conference on Human Factors in
daily operations, can improve efficiency, save human Computing Systems ACM, 3506-3510, 2017
resources, facilitate each practitioner, in the wind, in terms of
[13] Agrawal, et al. Computer Engineering & Applications,
risk management, there are also suitable models for user 619-624, 2000
behavior prediction.
43 Zhiqiao Zhong and Xu Zhuang: Deep Learning Applications in Business Activities

[14] Glorot, Xavier, A. Bordes, and Y. Bengio, International [17] Kvamme, Havard, et al, "Predicting Mortgage Default using
Conference on International Conference on Machine Learning Convolutional Neural Networks." Expert Systems with
Omnipress, 513-520, 2011 Applications 102, 2018
[15] Ghaphery, Jimmy, and S. Stearns. "Making the Most of Your [18] Singh, Kulwant, and W. Mitchell, Strategic Management
Data: Embedding Business Intelligence into Daily Operations." Journal, 99-115, 1996
2013.
[19] Waller, Matthew A., and S. E. Fawcett, Journal of Business
[16] Spanoudes, Philip, and T. Nguyen, "Deep Learning in Logistics, 77-84, 2013
Customer Churn Prediction: Unsupervised Feature Learning on
Abstract Company Independent Feature Vectors.", 2017