An Autonomic Deep Learning Artificial Neural Network based Algorithm for Effective Re-generation

Corruption of data due to data transmission errors from sensors and malicious users, accidental data loss is a very critical issue while applying the data set for ML algorithms. Most of the work in this field is concerned with data pre-processing and related activities. Our present work lay emphasis on the completeness of the data by analyzing data corruption and recovery from the same dataset using Deep learning techniques. Deep learning algorithms are known to have higher degree of prediction accuracy as compared to traditional ML algorithms. Further deep learning algorithms capabilities are boosted when the amount of training proportionally increases. The present study extends the capabilities of deep learning algorithms to recover corrupted or compromised data with high degree of accuracy. Additionally it generalizes by testing the algorithm across range of data sets. The application of deep learning algorithm resulted in recovery of data with average accuracy of 2%. Also the replaced values had an error rate of 2–3% as compared to the original values.