Research on Image Processing and Analysis

A brief summary of the problems under investigation:

  • How to defend against adversarial attacks on artificially intelligent systems

    • In the adversarial attacks, legitimate inputs (e.g., images) are altered by adding small, often imperceptible by humans, perturbations to the inputs

    • This increases the probability that a learned classifier makes classification error

  • Further details are provided here

Student

  • Mr. Shubham Malaviya