Data Re-Balancing using Fuzzy Clustering and SMOT Mechani

• Published in: 2023 4th International Conference on Electronics and Sustainable Communication Systems (ICESC) pp. 1714 - 1718
• Main Authors: Parmar, Gaurav, Gupta, Rimi, Bhatt, Tejas, Sahani, G.J., Panchal, Brijeshkumar Y., Patel, Hiren
• Format: Conference Proceeding
• Language: English
• Published: IEEE 06.07.2023
• Subjects: Bias; Clustering algorithms; Communication systems; Data models; Data Re-Balancing; Ensemble learning; Fuzzy Clustering; Machine Learning; Measurement; Merging; Protected Attributes; Support vector machines; Synthetic Minority Over-Sampling Technique
• DOI: 10.1109/ICESC57686.2023.10192964

In this research, "the authors present an innovative method for rebalancing data to eliminate bias in confidential features. It is possible for individuals with protected attributes, such as race, gender, or age, to be subjected to discrimination in the fields of job selection and loan approval. The suggested method utilizes fuzzy clustering in conjunction with the Synthetic Minority Over-sampling Technique (SMOTE) to rebalance the data set and reduce the amount of bias that is there. Fuzzy clustering is applied to discover groups of similar instances based on specific characteristics, while SMOTE is utilized to construct fake examples in the underrepresented group. Fuzzy clustering is also utilized to locate groups of comparable cases based on certain characteristics. The method is assessed in contrast to other rebalancing strategies that already exist and based on many standard data sets. The findings indicate that the suggested approach outperforms the methods that are state-of-the-art in terms of accuracy, precision, recall, and f1-score metrics, demonstrating its effectiveness in reducing bias in the protected" attributes.