Authors:

Chloe Yoder

Date Created:

Not specified

Course Title:

Professor:

Not specified

About Paper:

The modern age has become increasingly reliant on artificial intelligence and machine learning (AI/ML) to analyze their systems. This has led to the rise of some serious security concerns. The confidentiality of data across various industries necessitates a secure data protection technique that allows data to be securely masked while maintaining its utility. State-of-the-art encryption and data masking techniques employed in industry either limit analysts' capability or can be reversed-engineered. These methods also require a level of trust between industries and analysts because of potential data misuse and legal concerns. Most techniques do not require this level of trust and can impose limitations on the analyst, and therefore the data. Different industries capture a multitude of data ranging from time series to audio files all of which need to be protected against potential misuse from a collaborator or adversary. Additionally, when working with and encrypting audio data, the sound can become noticeably distorted to collaborators and adversaries. This encryption also means that the data is not preserved for AI/ML applications, like voice recognition. Using the deceptive infusion of data (DIOD) method solves the issues of encryption methods by allowing for the inferential capabilities to be preserved while protecting the data. This allows for the most important features needed for AI/ML tools to be concealed without compromising performance or proprietary information.

Source:

Purdue University / 2023

Topics:

No topics listed

Co-authors:

Chloe Yoder