algosel-rl

Source code for my Master's dissertation entitled Algorithm Selection for Subgraph Isomorphism Problems: A Reinforcement Learning Approach.

Running the scripts

To start, download and install R (version 3.4.4+) from CRAN. This installation contains the R interpreter and a simple GUI app for creating R scripts. This is sufficient to run the scripts in this repo; however, if you are planning to debug or modify the scripts, I highly suggest to use a full-featured IDE like RStudio.

Installing prerequisite packages

Run source('install_packages.R') on R command line to install all the necessary packages.

Rendering R Markdown (.Rmd) files

The rendered contents of the .Rmd files can be readily viewed at RPubs. Check out the following links:

• ntbk_eda_graphs2015.Rmd
• ntbk_eda_hard.Rmd
• ntbk_asresults_graphs2015.Rmd
• ntbk_asresults_reinforce.Rmd

These files can be rendered locally, and the easiest way to do this is through RStudio. Check out this guide.

Dissertation paper

The paper was written in LaTeX using TeXStudio software on Windows. Typesetting files are taken from utmthesis (v5.1) GitHub repository.

Useful Links

R Packages

• Algorithm Selection Library (aslib). RDoc | GitHub
• Leveraging Learning to Automatically Manage Algorithms (llama). RDoc | BitBucket
• R interface to TensorFlow. link

Recommended Reads

• Kotthoff, L. (2016). Algorithm selection for combinatorial search problems: A survey. In Data Mining and Constraint Programming (pp. 149-190). Springer, Cham. paper
• Kotthoff, L., McCreesh, C., & Solnon, C. (2016, May). Portfolios of subgraph isomorphism algorithms. In International Conference on Learning and Intelligent Optimization (pp. 107-122). Springer, Cham. paper
• Bischl, B., Kerschke, P., Kotthoff, L., Lindauer, M., Malitsky, Y., Fréchette, A., ... & Vanschoren, J. (2016). Aslib: A benchmark library for algorithm selection. Artificial Intelligence, 237, 41-58. paper
• Kotthoff, L. (2013). LLAMA: leveraging learning to automatically manage algorithms. arXiv preprint arXiv:1306.1031. paper
• Lindauer, M., van Rijn, J. N., & Kotthoff, L. (2017, December). Open Algorithm Selection Challenge 2017: Setup and Scenarios. In Open Algorithm Selection Challenge 2017 (pp. 1-7). paper
• Smith-Miles, K. A. (2009). Cross-disciplinary perspectives on meta-learning for algorithm selection. ACM Computing Surveys (CSUR), 41(1), 6. paper
• Sutton, R. S., & Barto, A. G. (1998). Introduction to reinforcement learning (Vol. 135). Cambridge: MIT press. book
• Policy Gradients
  • I still find Sutton & Barto's Introduction to Reinforcement Learning the easiest to understand regarding this topic. It might be understood better if complemented with readings from other sources. Check these slides: 1 2 3

Others

• ASlib website. link
• GRAPHS-2015 dataset source. GitHub
• Reinforcement Learning study plan. introductory | deep RL