Deep Enhancement in Supplychain Management with Adaptive Serial Cascaded Autoencoder with Long Short Term Memory and Multi-layered Perceptron Framework

Abstract

Recognizing and reducing risk is a major part of Supply Chain Management (SCM). Several companies are invested in Supply Chain Risk Management (SCRM) and they have the knowledge about the procurement occupancies within their companies and take steps to ensure this potent source of strategic value. Additionally, these types of companies yield the highest returns with the lowest amount of financial risk. Moreover, reducing financial risk in the SCM network requires thoughtful analysis and a proactive strategy. Hence, this task aims to make a financial risk assessment in SCM with deep learning techniques based on big data. Financial risk-related big data is collected from the Kaggle database and utilized in the data transformation phase. The transformed data is employed for evaluating the financial risk with the support of an Adaptive Serial Cascaded Autoencoder with Long Short-Term Memory and Multi-Layered Perceptron (ASCALSMLP). Here, the parameters for the deep learning techniques like LSTM and MLP were tuned by the hybrid Sandpiper Galactic Swarm Optimization (SGSO) algorithm to enhance the efficacy of the offered approach. From the results analysis, the accuracy of the developed model is 91.12% better than DHOA, 92.5% more than COA, 93.75% improved than GSO, and 94.62% superior to SOA models. Therefore, the results from the developed approach demonstrate effective prediction of financial risks.