Ankit Siva

Seattle, Washington, United States Contact Info
740 followers 500+ connections

Join to view profile

About

Working on Privacy, Fairness and Explainability in AI engineering and research at AWS…

Activity

Join now to see all activity

Experience & Education

  • Amazon Web Services (AWS)

View Ankit’s full experience

See their title, tenure and more.

or

By clicking Continue to join or sign in, you agree to LinkedIn’s User Agreement, Privacy Policy, and Cookie Policy.

Patents

  • Generating Relaxed Synthetic Data Using Adaptive Projection

    Issued US 11487765

    This patent is related to generating synthetic versions of private data. The synthetic data may be used to answer large numbers of statistical queries, subject to differential privacy, including queries other than those that have been explicitly answered using the private data. In one aspect, this solution relies on the adaptive use of a continuously relaxed projection mechanism that answers queries on a private dataset and perturbs the answers to the queries with “noise,” e.g., by a Gaussian…

    This patent is related to generating synthetic versions of private data. The synthetic data may be used to answer large numbers of statistical queries, subject to differential privacy, including queries other than those that have been explicitly answered using the private data. In one aspect, this solution relies on the adaptive use of a continuously relaxed projection mechanism that answers queries on a private dataset and perturbs the answers to the queries with “noise,” e.g., by a Gaussian perturbation technique. When evaluating queries, a synthetic dataset (e.g., a set of relaxed synthetic data) identified according to a gradient optimization technique results in answers that most closely match the “noisy” answers. The synthetic dataset may be generated in an iterative manner, such as by identifying queries for which differences between “noisy” answers determined using the private dataset and answers determined using a previous incarnation of the synthetic dataset are maximized, and optimizing the synthetic data to account for queries with the maximum differences. In this manner, a synthetic dataset that returns answers to queries that are closest to “noisy” answers obtained using a private dataset may be identified effectively and efficiently, and with a determined use of a privacy budget.

    Other inventors

  • Generating Relaxed Synthetic Data Using Adaptive Projection

    US 11841863

    Other inventors
    See patent

Courses

  • Advanced Algorithms

    CS 4540

  • Honors Algorithms

    CS 3511

  • Honors Probability and Statistics

    MATH 3225

Honors & Awards

  • Dean's List

    Office of the Institute President

  • Faculty Honors

    Office of the Institute President

    Faculty Honors awarded for obtaining perfect GPA (4.0) in the Spring 2017 Semester.

  • Faculty Honors

    Office of the Institute President

    Awarded Faculty Honors for obtaining a perfect GPA (4.0) in the Fall 2016 Semester.

  • National Talent Search Scholar

    National Council of Educational Research and Training (NCERT)

    One of 1000 students selected at the national level for a scholarship recognizing talent in academics. Chosen out of roughly 1,000,000 students who appeared at the State and National levels.

Languages

  • English

    Native or bilingual proficiency

  • Sanskrit

    Limited working proficiency

  • Hindi

    Full professional proficiency

  • Tamil

    Native or bilingual proficiency

More activity by Ankit

View Ankit’s full profile

  • See who you know in common
  • Get introduced
  • Contact Ankit directly
Join to view full profile

Other similar profiles

Explore collaborative articles

We’re unlocking community knowledge in a new way. Experts add insights directly into each article, started with the help of AI.

Explore More

Others named Ankit Siva

1 other named Ankit Siva is on LinkedIn

See others named Ankit Siva

Add new skills with these courses

See all courses