Securing sensitive manufacturing data with data mining-based differential privacy

What is it about?

Manufacturing industries are turning to new technologies to improve the way they operate. Data mining is one such technology. Advances in sensors and data collection methods have helped companies analyze their operations and improve them using computer models. But, this has increased chances of data leaks. There is a risk of competitors gaining access to the data used to build and train models based on artificial intelligence. This can include details about the operating parameters of a machine and product design. In this paper, the authors propose a data mining framework to prevent data leaks. They use an approach called “differential privacy” (DP). In this method, the input data for training is modified by adding noise to it. This way, attackers cannot simply reverse engineer the model to get hold of sensitive information. Using this method, the authors build a secure model and test it with real world data. They show that it helps reduce power consumption in computer numerical control machines.

Featured Image

Photo by Tim Mossholder on Unsplash

Photo by Tim Mossholder on Unsplash

Why is it important?

Companies today face a wide range of data privacy issues. Since data is stored in a central server, it is very easy to access. To keep these data safe, this study offers a robust method for protecting data privacy. By randomizing the input data with noise, the method hides the true nature of the data. This protects its privacy even in the case of a data leakage. KEY TAKEAWAY: A data mining framework based on DP can ensure the privacy of data and enable smart manufacturing.