Mark Kon

Professor (Math/ Stat)

  • Title Professor (Math/ Stat)
  • Office Room 259, MCS Building (111 Cummington Mall)
  • Phone 617-353-9549
  • Education PhD, Mathematics, MIT

Research Interests:

  • Quantum Information and Computation
  • Machine Learning and Neural Networks
  • Computational Biology and Bioinformatics

Recent research projects:

Quantum spin systems:  This project studies the properties of high dimensional quantum systems produced by interacting spins.  It is an entre into the statistical calculation of quantum matrix elements through sampling of complex spin systems whose individual matrix elements are not all positive, something traditionally difficult to do.  The eventual goal is to study real time evolution of spin systems using heretofore unstudied sampling methods, in which complex phases are sampled in ways that allow system phase cancellations to be estimated in an accurate way.

Machine learning:  Many supervised machine learning (ML) methodologies (such as support vector machines, neural nets and random forests) are reaching limiting prediction accuracies.  The next logical step in their development is a better integration of ML with data outside our experiments and in the outside world (prior information).  In this project new statistical learning theorems have been proved for improving ML inference via improvement of the data used, using prior information (for example protein-protein interactions for genetic information).  This information can be used to impose graph and metric structures on data indices.  Applications have included improved classification of cancer types using gene expression data.

Computational biology:  The most challenging current problem in cancer research is a computational one:  how can we separate different cancer subtypes, and identify them through genetic and other molecular testing?  Subtyping in turn allows the identification of new drugs that may work only on some subtypes of a cancer (though not on others) – this is in fact providing the leading headlines in cancer research these days. Toward a similar end, we are working on statistical methods of variable selection that can be applied to genetic data to allow for more efficient subtyping of cancers.  These methods are in some cases producing excellent differentiation between cancer types, promising to allow such subtyping to produce distinctions between cancers that till recently could only be identified in laboratory work.

Abbreviated CV:

Mark Kon is a professor of Mathematics and Statistics at Boston University, and is affiliated with the Bioinformatics Program and the Computational Neuroscience program.  He received a PhD in Mathematics from MIT, and has Bachelor’s degrees in Mathematics, Physics, and Psychology from Cornell. He has had appointments at Columbia University as Assistant and Associate Professor (Computer Science, Mathematics), and at Harvard and MIT as a Visiting Professor.  He has also served as his department’s director of graduate studies. He has approximately 100 publications in mathematical quantum physics, statistics, machine learning, neural networks, and complexity, including one book.  He is on the editorial board of Neural Networks, and has been on the organizing committee of the World Congress on Neural Networks twice. He has had research grants and contracts from the American Fulbright Commission, National Science Foundation, National Institutes of Health, and the U.S. Air Force. He has given around 120 talks in 15 countries.

View all profiles