Simon Kato

Simon Kato

Mentor

Dr. Sara Pollock

College

College of Liberal Arts and Sciences

Major

Mathematics and Statistics

Minor

Computer Science

Organizations

N/A

Academic Awards

Bright Futures, HSF Scholar, Graduate Fellowship, Deans/Presidential's all semesters.

Volunteering

N/A

Research Interests

Deep Learning, Algorithms, Compact Data Structures, Eigenvector Extrapolation, Numerical Analysis

Hobbies and Interests

Gardening, cooking, and playing video games with friends

Research Project

Acclerating PageRank Convergence

The PageRank algorithm has famously been used by Google to rank web pages by assigning a reputation which is based on the quantity and quality of links to said web page. The underlying process in ranking the web pages is solving an eigenvalue problem with a Markov matrix.

Given the physical problem, there is belief that knowledge of the Markov structure and the sparsity of the matrix can be exploited with hopes to accelerate the convergence of the eigenvalues. Research into the structure of the Markov chain structure will be conducted with particular attention to properties which can lead to hyper-tuning the parameters for the eigenvalue problem. Sep Kamvar’s “Numerical Algorithms for Personalized Search in Self-organizing Information Networks” will be read to develop an understanding of the PageRank problem as well as develop intuition. Beyond this, modern methods for accelerating acceleration will be tested both parameterized independently of Markov chain properties and dependently on Markov structure.

As the world-wide-web continues to expand in size and importance, the importance for an accelerated convergence cannot be easily overlooked. The effects of a small increase can translate to a massive difference when the sheer scale of web users is taken into consideration.

  • Dr. Sara Pollock
  • Mathematics and Statistics
  • Computer Science
  • Deep Learning, Algorithms, Compact Data Structures, Eigenvector Extrapolation, Numerical Analysis
  • Bright Futures, HSF Scholar, Graduate Fellowship, Deans/Presidential's all semesters.
  • N/A
  • N/A
  • Gardening, cooking, and playing video games with friends
  • Acclerating PageRank Convergence
  • The PageRank algorithm has famously been used by Google to rank web pages by assigning a reputation which is based on the quantity and quality of links to said web page. The underlying process in ranking the web pages is solving an eigenvalue problem with a Markov matrix.

    Given the physical problem, there is belief that knowledge of the Markov structure and the sparsity of the matrix can be exploited with hopes to accelerate the convergence of the eigenvalues. Research into the structure of the Markov chain structure will be conducted with particular attention to properties which can lead to hyper-tuning the parameters for the eigenvalue problem. Sep Kamvar's "Numerical Algorithms for Personalized Search in Self-organizing Information Networks" will be read to develop an understanding of the PageRank problem as well as develop intuition. Beyond this, modern methods for accelerating acceleration will be tested both parameterized independently of Markov chain properties and dependently on Markov structure.

    As the world-wide-web continues to expand in size and importance, the importance for an accelerated convergence cannot be easily overlooked. The effects of a small increase can translate to a massive difference when the sheer scale of web users is taken into consideration.