About

Professor Mark Kon is a member of the Applied Mathematics and Probability and Statistics research groups at Boston University. He is a faculty member in the Department of Mathematics & Statistics, where he contributes to research and teaching in these areas. Professor Kon's office is located in CDS 523, and he holds office hours on Tuesday from 11:00 to 12:15 pm and Thursday from 2:00 to 3:15 pm. For more information about him, one can visit his personal webpage. His work focuses on applied mathematics, probability, and statistics, supporting the department's mission to advance knowledge and education in these fields.

Research topics

• Computer Science
• Artificial Intelligence
• Algorithm
• Mathematics
• Data Mining
• Statistics
• Mathematical analysis
• Pure mathematics
• Combinatorics
• Geometry

Selected publications

• Coupled VAE: Improved Accuracy and Robustness of a Variational Autoencoder

  Entropy · 2022 · 17 citations

  Senior authorCorresponding
  • Artificial Intelligence
  • Computer Science
  • Artificial Intelligence

  We present a coupled variational autoencoder (VAE) method, which improves the accuracy and robustness of the model representation of handwritten numeral images. The improvement is measured in both increasing the likelihood of the reconstructed images and in reducing divergence between the posterior and a prior latent distribution. The new method weighs outlier samples with a higher penalty by generalizing the original evidence lower bound function using a coupled entropy function based on the principles of nonlinear statistical coupling. We evaluated the performance of the coupled VAE model using the Modified National Institute of Standards and Technology (MNIST) dataset and its corrupted modification C-MNIST. Histograms of the likelihood that the reconstruction matches the original image show that the coupled VAE improves the reconstruction and this improvement is more substantial when seeded with corrupted images. All five corruptions evaluated showed improvement. For instance, with the Gaussian corruption seed the accuracy improves by 1014 (from 10-57.2 to 10-42.9) and robustness improves by 1022 (from 10-109.2 to 10-87.0). Furthermore, the divergence between the posterior and prior distribution of the latent distribution is reduced. Thus, in contrast to the β-VAE design, the coupled VAE algorithm improves model representation, rather than trading off the performance of the reconstruction and latent distribution divergence.

  DOI

• Wavelet matrix operations and quantum transforms

  Applied Mathematics and Computation · 2022 · 8 citations

  Senior authorCorresponding
  • Computer Science
  • Artificial Intelligence
  • Mathematics
  DOI

• Anomaly detection: A functional analysis perspective

  Journal of Multivariate Analysis · 2021 · 6 citations

  Senior authorCorresponding
  • Computer Science
  • Data Mining
  • Mathematics
  DOI

Recent grants

• Stochastic Dynamic MOdeling of Cellular Protein Interactions

  NIH · $108k · 2018–2021

• AMPS: Uncertainty Quantification for Stochastic Analysis of Electrical Power Networks

  NSF · $229k · 2017–2022

• Stochastic Dynamic MOdeling of Cellular Protein Interactions

  NIH · $216k · 2018–2022

Frequent coauthors

• Louise A. Raphael

  34 shared

• Charles DeLisi

  17 shared

• Max Diem

  12 shared

• Julio E. Castrillón-Candás

  12 shared

• Snežana Milanović

  Sumitomo Dainippon Pharma (United States)

  12 shared

• Xinying Mu

  11 shared

• Yue Fan

  Xi'an Jiaotong University

  11 shared

• Parker Kuklinski

  11 shared

Similar researchers at Boston University

• Alexa S. Beiserh-188
• David Boash-156
• Benjamin Sovacoolh-143
• Kathryn L. Lunettah-132
• Ranga Mynenih-130