MAGDM for selecting machine learning techniques: A novel approach utilizing the 2-tuple linguistic cubic q-rung orthopair fuzzy set

Abstract

Machine learning is a powerful technology for both current and future information management and it is already being used in a variety of domains. The use of machine learning in cyber security is still in its early stages, emphasizing the significant gap between research and practice. The q-rung orthopair fuzzy set (q-ROFS) has been demonstrated to be a valuable tool in describing decision-makers (DMs) assessment values in multi-attribute group decision-making (MAGDM). In this paper, we introduce a new tool called the 2-tuple linguistic cubic q-rung orthopair fuzzy set (2TLCq-ROFS) based on the combination of q-ROFS with interval-valued q-ROFS under the 2-tuple linguistic scenario in order to capture DMs assessment information in the complex MAGDM problems in a more efficient way. We investigate machine learning based MAGDM problems in cyber security in which the DM’s preference information is expressed as the 2TLCq-ROF numbers (2TLCqROFNs). Based on the combination of power average and power geometric operators with Muirhead mean, we propose a family of new aggregation operators including: the 2TLCq-ROF power Muirhead mean (2TLCq-ROFPMM), the 2TLCq-ROF dual power Muirhead mean (2TLCq-ROFDPMM), the 2TLCq-ROF weighted power Muirhead mean (2TLCq-ROFWPMM), and the 2TLCq-ROF weighted dual power Muirhead mean (2TLCq-ROFWDPMM) operators under the 2TLCq-ROF environment, which are more precise than the current aggregation operators and take into account the interconnection of the 2TLCq-ROFNs. Then, a versatile 2TLCq-ROF MAGDM approach is established and a case study on the evaluation of machine learning techniques in cyber security is given to show the method’s effectiveness and validity. Moreover, a parameter analysis is carried out to analyze the influence of parameters on ranking results. The comparative analysis further confirms the effectiveness and feasibility of the proposed approach to show why DMs should select our suggested strategy over several others. In the end, some conclusions of this paper are determined and future directions are demonstrated.