Review of Machine Learning Techniques in the Supply Chain Management of Indian Industry: A Future Research Agenda

Introduction

The success of an industry is heavily dependent on the Supply Chain. It's a complicated integration of many different business units, each with their own unique set of technologies and resources. Despite the fact that the efficiency of a supply chain is reliant on a variety of factors, it appears to be difficult to design/develop a common platform that allows industry to respond quickly to market needs. In recent years, the world has entered a new era known as the “fourth industrial revolution,” which is characterised by the “development of digitization”, “information”, “robotics”, “communication technology”, and “Artificial Intelligence (AI)” (Baryannis et al., 2019). In this day and age, robots are becoming intelligent enough to make judgments in place of people. These techniques are referred to as machine learning (ML), and This group is focused on automated systems that are capable of “learning” through their past usage (Pournader et al., 2021). Due to machines' rising ability to deal with large amounts of data over the last two decades, and some computers' even capability of finding hidden pathways and complicated relationships to make appropriate and reliable decisions on issues where humans were incapable or unwilling, machine learning (ML) was born (Toorajipour et al., 2021). According to research, robots are capable of providing more accurate outcomes than humans in a wide range of decision-making contexts (Hartley & Sawaya, 2019; Pournader et al., 2021).

The use of machine learning and its fundamental components can improve the prediction of SCM Performance Learning algorithm, deep learning, and optimization algorithms all contribute to machine learning's ability to continually seek for the SC performance-enhancing factors. For the fourth time, using visual machine learning over a SC network, ML may examine a broad spectrum of applications in physical asset management and physical inspection. Despite the fact that the applications of machine learning have numerous advantages and despite the fact that certain research have shown that ML has made its way into Supply Chain Management (SCM), including warehouse management (Makkar et al., 2020; Ni et al., 2020), Neural Networks (Golmohammadi et al., 2009; Ni et al., 2020), Support Vector Machine (Ni et al., 2020; Prahathish et al., 2020), Logistic regression (H. Ma et al., 2018; Ni et al., 2020), Decision Tree (Estelles-Lopez et al., 2017) and Extreme Learning Machine(Martínez et al., 2020; Ni et al., 2020), One research found that just 15% of businesses had implemented ML including one or even more SC functions. The insufficient usage of machine learning in SC may be due to a lack of awareness of how ML may be implemented, a low level of acceptability in the business culture, and an inability to get appropriate data(Ni et al., 2020). An urgent systematic study is required to quantitatively examine the most recent research trends, to investigate the machine learning algorithms that are often employed in supply chain management, and to identify the SCM tasks that are most suited for ML(Sharma et al., 2021). As a result of noticing a lack of studies on the use of decision support systems for conducting supply chain management (SCM) investigations with regards to MI, we chose this study. The majority of studies have used fuzzy set theory, experience and understanding systems, multi-criteria methods, and/or minimization algorithms and situational optimization as primary methods, which we believe is insufficient. However, despite the fact that increasingly sophisticated artificial intelligence technologies such as “machine learning”, “deep learning”, and “image/text processing” are essential components of today's supply chains, the literature is still lagging behind the industry when it comes to studying their applications in manufacturing. At same time, our research showed important fragmentation in artificial intelligence (AI) research, stressing the necessity of a classification system that can guide operations and supply chain researchers in broadening their knowledge of supply chain management (SCM) AI research in the past, present, and future.