A novel similarity measurement for triangular cloud models based on dual consideration of shape and distance

Abstract

It is important to be able to measure the similarity between two uncertain concepts for many real-life AI applications, such as image retrieval, collaborative filtering, risk assessment, and data clustering. Cloud models are important cognitive computing models that show promise in measuring the similarity of uncertain concepts. Here, we aim to address the shortcomings of existing cloud model similarity measurement algorithms, such as poor discrimination ability and unstable measurement results. We propose an EPTCM algorithm based on the triangular fuzzy number EW-type closeness and cloud drop variance, considering the shape and distance similarities of existing cloud models. The experimental results show that the EPTCM algorithm has good recognition and classification accuracy and is more accurate than the existing Likeness comparing method (LICM), overlap-based expectation curve (OECM), fuzzy distance-based similarity (FDCM) and multidimensional similarity cloud model (MSCM) methods. The experimental results also demonstrate that the EPTCM algorithm has successfully overcome the shortcomings of existing algorithms. In summary, the EPTCM method proposed here is effective and feasible to implement.

Keywords

• Cloud model
• Expectation curve
• Similarity measurement
• Triangular fuzzy number
• EW-type closeness
• Cloud model variance