Menu

Contact

ETH Zürich
Gabriel Krummenacher
Dept of Computer Science
CAB F 61.2
Universitaetstrasse 6
8092 Zurich
Switzerland

Phone
+41 44 632 38 80

E-Mail
gabriel.krummenacher [_at_] inf.ethz.ch

Large-scale Randomized Regression (LRR)

iws
This python package implements multiple algorithms to solve large-scale linear regression problems by subsampling and projecting the original data set.
  • SRHT: Uses the Subsampled Randomized Hadamard Transform (SRHT), equivalent to leverage based sampling.
  • aIWS, aRWS: Samples based on approximated statistical influence.
  • Uluru: SRHT with bias correction.

See: McWilliams, B., Krummenacher, G., Lučić, M., and Buhmann, J. M. (2014) Fast and Robust Least Squares Estimation in Corrupted Linear Models.

[download LRR] [Download working paper]

Ellipsoidal Multiple Instance Learning (eMIL)

emil
eMIL is implemented as part of the optimization toolbox OptWok.
As an example of how to use eMIL, have a look at the demo experiment code in emil_demo.zip. The code shows how to reproduce the results in Krummenacher et al. (2013).

[download OptWok] [download demo] [download paper]

Experimental Design Toolbox (EDT)

greedy optimal time points
The toolbox from the paper Near-optimal Experimental Design for Model Selection in Systems Biology (Busetto et al. 2013, submitted) implemented in MATLAB.

EDT is based heavily on the Mutual Information Toolbox and also requires the Systems Biology Toolbox 2 and the Submodular Function Optimization Toolbox.

The Mutual Information Toolbox (MIT) is documented in Appendix A of the master's thesis "Entropy-based Experimental Design for Model Selection in Systems Biology" [pdf] of Alain Hauser, 2009 ETH Zurich.

Documentation for the Experimental Design Toolbox is available here.

[download EDT] [EDT documentation]
[download MIT] [MIT documentation]