EFFICIENT LIFELONG LEARNING ALGORITHM (ELLA) Version 1.0 (July 3, 2013) Copyright (c) 2013 Paul Ruvolo & Eric Eaton The copyright extends to all implementation files for ELLA under this distribution. This program is free software; you can redistribute it and/or modify it under the terms of the GNU General Public License Version 3 BUILDING The ELLA code is written in pure matlab and thus does not need to be built. However, ELLA does have a dependency on the Sparse Additive Modeling Toolbox of Julien Mairal et al. (http://spams-devel.gforge.inria.fr/). In order to use ELLA one must first go to the subdirectory 'externalLibs/spams-matlab' and follow the instructions for building this toolbox. RUNNING THE CODE We have included three demos for utilizing the different aspects of the ELLA software. The demo files are called: (1) runExperiment (2) runExperimentActiveTask (3) runExperimentActiveTaskTargeted CITING OUR WORK If you use or adapt this code, please cite the following papers: Paul Ruvolo & Eric Eaton. (2013). ELLA: An efficient lifelong learning algorithm. In Proceedings of the 30th International Conference on Machine Learning. Atlanta, GA, June. Paul Ruvolo & Eric Eaton. (2013). Scalable lifelong learning with active task selection. In Proceedings of the 27th AAAI Conference on Artificial Intelligence. Bellevue, WA, July. DATA This release includes the landmine data set for use by the demonstration scripts. The landmine data was originally described in: Xue, Y., Liao, X., Carin, L., & Krishnapuram, B. (2007). Multi-task learning for classification with Dirichlet process priors. The Journal of Machine Learning Research, 8, 35-63. Please cite this paper if you use the landmine data set in your own work.