Crossover Approach utilizing Ant Colony Optimization (ACO) and Neural Network for Anonymization in Big Data

  • U. Selvi, S. Pushpa
Keywords: Anonymization . k-Anonymity . Privacy aware Machine Learning . Big Data . Neural Network . Map Reduce . Interactive Machine Learning . Ant colony Optimization

Abstract

With the expansion of distributed computing environment, many learning problems now need to affect with distributed input data. Applying Machine Learning techniques to guard privacy is termed as Privacy aware Machine Learning which can be simply accomplished by first anonymizing a dataset before liberating it for the determination of data mining or knowledge extraction. To reinforce support in learning, it is significant to deal with the privacy concern of every data holder by extending the privacy preservation notion to original learning algorithms.  Hence, a hybrid algorithm variant that incorporates back propagation algorithm with Ant colony Optimization (ACO) in Map Reduce framework is proposed. The proposed work permits a neural network to be trained without demanding either party to reveal data to the opposite. Human background is applied via interactive Machine Learning to the method of anonymization; this is often done by eliciting human preferences for preserving some attribute values over others within the light of specific tasks. The effect of interactive learning within the sector of anonymization will greatly depend upon the experimental setup, like an appropriate choice of application field and selection of suitable test focuses. Further, human knowledge in proposed MapReduce computer cluster yield a measurably better classification result for big data application.  We offer complete correctness and security analysis of our algorithms. Our experiments show that proposed work results in significantly improved classification effects than a rigid automatic approach.

Published
2021-07-05
How to Cite
S. Pushpa, U. S. (2021). Crossover Approach utilizing Ant Colony Optimization (ACO) and Neural Network for Anonymization in Big Data. Design Engineering, 1752-1767. https://doi.org/10.17762/de.vi.2492
Section
Articles