In October 2012, I joined SVBio (Silicon Valley
Biosystems, Inc.) a early-stage genomics startup backed by Sequoia Capital that is focused on
clinical-grade NGS-based genetic testing. At SVBio, I've been building out the
core genomics infrastructure, developing genomics and machine learning
algorithms for processing of NGS data and diagnostics, and engaging in
relationships with clients and partners.
In September 2013, I completed a Master's in Computer Science at Stanford
University, where I was funded by the NSF
Graduate Research Fellowship Program. At Stanford, I did research in the
lab of Professor Serafim Batzoglou, working on developing machine learning
algorithms for predictive modeling of gene expression. During my Master's, I
also took a number of interesting classes, especially enjoying visiting
professor Kevin Murphy's class on advanced
methods for probabilistic graphical models, and professor Boyd's class on
convex optimization. At Stanford, I was also fortunate to have the opportunity
for interdisciplinary research with members of the Biomedical Informatics
I did my undergraduate studies in Electrical Engineering and Computer
Science at UC Berkeley, where I worked with Assistant Professor Pieter Abbeel on probabilistic
robotics and computer vision for robotics. My most recent work in this area
introduced a novel method for simulating cloth objects that is suitable for
robotic applications, a method of planning manipulations with cloth
objects, and a method for probabilistically tracking the state of a cloth
object over a sequence of robotic manipulations.
In general, I am interested in applying cutting-edge machine learning and
probabilistic modeling methods to data-rich molecular-biological systems, such
as gene regulation. I am also interested in statistical machine learning and
probabilistic graphical models from a theoretical perspective. I've also been
a long-time fan of unsupervised feature learning and "deep" hierarchical
representations of data.
Journal Publications / Conference Proceedings
A social network of hospital acquired infection built from electronic medical record data
Cusumano-Towner M, Li DY, Tuo S, Krishnan G, Maslove DM. Journal of the American Medical Informatics Association. 2013 May 1;20(3):427-34
We used EMR data from Stanford Hospital to construct a probabilistic infection
network which we used for simulation experiments.
Bringing clothing into desired configurations with limited perception (pdf, video)
Cusumano-Towner M, Singh A, Miller S, Abbeel P. In the proceedings of the International Conference on Robotics and Automation (ICRA), 2011.
We formulate a cloth simulator suitable for robotics
applications as a convex optimization problem, as well as a HMM-based
framework for tracking the state of cloth objects through a sequence of
manipulations. We apply these methods to autonomously and reliably perform
the challenging task of bringing an unidentified crumpled article of clothing
into a desired configuration.
An optical flow-based approach for cloth grasp point detection with application to robotic towel folding (pdf, video)
Maitin-Shepard J, Cusumano-Towner M, Lei J, Abbeel P. In the proceedings of the International Conference on Robotics and Automation (ICRA), 2010.
We present a novel vision-based cloth grasp point detection algorithm and
demonstrate it on the end-to-end task of autonomously folding towels. The
system successfully folds 50 out of 50 previously-unseen towels.
Posters and Presentations
Modeling context specific gene regulatory interactions
Kyriazopoulou-Panagiotopoulou S, Cusumano-Towner M, Kasowski M, Grubert
F, Snyder M, Kundaje A, Batzoglou S. To appear in the proceedings of the 2013
RECOMB/ISCB Conference on Regulatory and Systems Genomics.
Modeling context-specific gene regulation
Kyriazopoulou-Panagiotopoulou S, Cusumano-Towner M, Batzoglou S, Kundaje A. 2013 Genome Informatics CSHL Meeting.
Modeling context-specific gene regulation with multi-task boosting
Kyriazopoulou-Panagiotopoulou S, Cusumano-Towner M, Batzoglou S, Kundaje A. 2013 NIPS Workshop on Machine Learning in Computational Biology (MLCB).
Projects, Ideas, and Writeups
Inference in Dynamic Infection Network (pdf)
I had just taken CS228T at Stanford and was excited to try out some of the
methods. In this project I used blocked Gibbs sampling to do inference in a
dynamic infection network with SIR (susceptible-infectious-resistant) dynamics.
Introduction to Conjugate Gradient (report, presentation)
Final project for scientific computing course at Stanford.
Generalizing Boosting Algorithms (pdf)
Boosting with log-loss (pdf)
For use in the research on predictive modeling of gene expression, I made the
following writeups and derivations to help myself understand the relationship
between boosting and optimization, and derived some efficient algorithms for our particular problem.
Efficient K-Nearest-Neighbors for Feature Spaces with Separable Distance Function (pdf)
An algorithm for efficiently finding the K-Nearest-Neighbors of a test example,
when the training and test data lie on some discrete lattice in feature space.
This was used to efficiently run KNN on the genes x conditions lattice of
examples used in the predictive modeling of gene expression research.
Exploring the Functional Landscapes of Gene Sets with Interactive Multidimensional Scaling (pdf)
Stanford CS 448 (Data Visualization) final project in which I visualized gene
sets using interactive dimensionality reduction for a visual version of
Epigenetic Reduction (pdf)
Stanford CS 273a (Computational Tour of Human Genome) final project in which we
learned to predict epigenetic marker signals across the genome and across cell
lines from a small set of these markers. We investigated how well we could
predict epigenetic states using a subset of the markers to impute the missing
Analyzing Gene Expression Time Series (pdf)
Stanford CS 229 (Machine Learning) final project in which I applied several
clustering algorithms to time-series gene expression data.
Sparse convolutional restricted Boltzmann machine with application to trajectory classification (paper, poster)
Berkeley CS281a (Fall 2009, Statistical Learning Theory) final project in which
we trained a 1-dimensional CRBM to perform unsupervised feature learing of
helicopter trajectory features, for use in segmentation of helicopter
trajectories into different maneuvers.
Just For Fun
Interactive Distance Metric Learning (pdf)
In this brief project I was experimenting with iteratively learning a distance
metric that agreed with a user's implicit internal metric, based on continuous
user feedback in a dimensionality reduction setting. The algorithm takes
gradient descent steps in a (diagonal) metric learning objective simultaneously
with an interactive MDS.
Digital Universe (video1, video2)
A program written in C++ that superimposes a human figure onto a
nondeterministic version of Conway's Game of Life, and modifies the
probabilities based on location in the image.
Last updated October 28, 2013.