Detecting the Security Level of Various Cryptosystems Using Machine Learning Models

Detecting the Security Level of Various Cryptosystems Using Machine Learning Models

H.K Shashikala¹, Nallani Ashok kumar², K Jayanth³

²³⁴7 B.Tech 4th Year, CSE, Jain University, Bengaluru 562112, Karnataka, India.

¹Associate Professor, CSE, Jain University, Bengaluru, 562112, Karnataka, India.

Abstract
Recent advances in multimedia technology have increased the need for digital data security. Researchers typically focus on changing current practices to address system security vulnerabilities. However, the security of critical data is seriously compromised by a number of proposed encryption algorithms that have proven weak over time. Using the right encryption technique is essential, but the type of data you want to protect will determine which algorithm is best for your particular situation. However, comparing different cryptosystem solutions individually can be a time-consuming process. Our technique uses support vector machines (SVMs) to assess the security level of image encryption. Choose the right encryption method quickly and securely. This effort also produces a dataset containing standard cryptographic security metrics such as entropy, contrast, uniformity, peak signal-to-noise ratio, mean squared error, energy, and correlation. These features were assembled from a series of scrambled images. Record labels are divided into three classes based on the level of security they offer.

Their accuracies were calculated to assess the performance of acceptable, well-performing, and bad recommendation models, and the results showed how effective his SVM-based system was.

Key Words: security analytics, image encoding, support vector machines (SVM), and cryptosystems