Introduction on models, applications and managerial implications
The following paper presents a critical evaluation on the research conducted by Kraus et. al. (2020) on deep learning applicability to business operations. The examined article investigates field development from a technological standpoint. It shows a theoretical review and in advance different data analysis methods to clearly establish the operational performance that deep neural networks have towards organizational development. The research examines deep learning as a progressive optimization routine that can scale more efficiently in data analysis and corporate operations. Finally, this progression towards deep learning benefits is defined through technological advantages, rather than economic benefits.
Industrial applicability
The article defines deep learning as the new industrial standard in predictive analytics and business operations. Regarding chapter 2 and chapter 3 this paper investigates on a very technical level and utilises terminology appropriate for computer scientists or analysts. This is due to its detailed analysis on deep neural networks, their representational layers and activation functions. Considering a convex optimisation, the research recognises problems such as high numbers of free parameters in analysis processes. The paper describes deep neural networks in a more technical analysis and goes into modelling and illustrating the whole process from input to output.
However, considering that the topic is introduced as an investigation in business conducts utilising DNN, the paper may be more technical than expected. This paper provides very few examples on how such systems apply in practical examples and are being utilised. Some of the examples provided in table 2 include image recognition, speech recognition, text classifier, text mining etc.
In chapter 3 the research provides satisfactorily analysis and good examples that motivate business analytics utilising deep neural networks. In my opinion, this part is also more technical than may be needed and may not be easily understood by everyone.
Personal reflection
In terms of technical analysis this paper is very detailed and thorough. However, it is difficult to understand. The paper utilised terminology associated to computer science and data analytics. Thus, it may be difficult to follow without having experience in the field or having experience in the processes described. Thus, detailing these processes in more collaborate examples or storytelling may have increased the readers understanding. A possible improvement could be to connect practical implementations in real life scenarios that demonstrate business strategies utilising such functions.
Also, there are parts that feel needles to the research objective and not completely accurate. For example, the comparison between neural networks and human brain does not corelate in value with a business investigation. Also, my personal experience in machine learning have led me to understand that there is a general misunderstanding on how deep neural networks mimic brain activity. Since the paper does not specify the gap that exists in understanding human brain and applying its complexity to machine learning, and since this information is not required to understanding the DNN applicability in business strategies, this part could be avoided.
Conclusion
This paper utilizes very interesting literature on the development of deep neural networks from a technical perspective. However, in the abstract we are led to believe that its content is specific to business analytics and utilizes the technical aspect for business purposes. However, even though there is appropriate motivation on the usefulness of deep neural networks, there is little association to operational practices and practical implementations. Finally, this paper can be considered correct in terms of content, and it can also be said that it does fulfill its purpose. However, it is not written in a way that it can be easily understood and applied to many different disciplines. For example, it may demonstrate how engineers construct and utilise DNN, but does not provide good understanding framework for a business specialist.
References
Mathias Kraus, Stefan Feuerriegel, and Asil Oztekin. Deep learning in business analytics and operations research: Models, applications and managerial implications. European Journal of Operational Research, Volume 281, Issue 3,
2020, Pages 628-641.