Frequently Asked Questions

Can I get your implementation of DWM?

Sure! The file dwm.jar contains an implementation of DWM based on WEKA. The archive also includes a simple instance generator for the Stagger Concepts. The main method of the DWM class builds an ensemble of weighted naive Bayesian classifiers using the Stagger Concepts. For each time step, it prints the percent correct. At the end of the trial, it prints the ensemble.

Will you make available your collection of malicious and benign executables?

The malicious executables we used are available from VX Heavens. We can't distribute our benign executables because of licensing and copyright restrictions.

Can I get the data set you produced from the malicious and benign executables?

Sure: malware.arff.

Can I get your rooftops data set?

Sure. Please send me e-mail.

Can I get your data set for recidivism prediction?

I obtained the data set for recidivism prediction from the University of Michigan's Inter-university Consortium for Political and Social Research. Under the terms of my agreement with them, I cannot redistribute the data set. However, if you meet their criteria for distribution, you can retrieve the data set from Schmidt and Witte's study Predicting Recidivism in North Carolina.

How can I get AQ-PM?

The Machine Learning and Inference Laboratory at George Mason University distributes AQ-PM.

Can you give me some advice about graduate school?

If you're planning on going to graduate school in CS, then during your junior or senior year, try to get involved in a professor's research. You should probably take an upper-division course from the professor, so you'll have a basis for doing a project. Also read

In addition, each year the department holds information sessions about graduate school. We typically have one in the spring, so the seniors heading to graduate school can share their experiences.

Can I work with you as an undergraduate research assistant?

Chances are good, but you should probably take my AI class. You should also have an interest in machine learning. Although I find lots of things interesting, I won't be much help if you want to study computational geometry or parallel processing. Working arrangements can include credit or non-credit, paid or unpaid, and structured and unstructured. For instance, I'd be happy to give you a learning algorithm to implement for fun, or you might want to work on a project funded through, say, GUROP. Naturally, expectations will vary depending on the arrangement.

Can I work with you as a graduate research assistant?

Perhaps. If you are a student in our Master's program or Ph.D. program, then the first step is to take my class on machine learning. If you like what I do and I like what you do, then we can talk about projects and support. If you are a student at another university, then it would be best to work with someone at your institution with research interests similar to yours. I apologize for not being able to respond to external requests for admission, summer positions, and research assistantships.

Will you write a letter of recommendation for me?

Generally, if you've taken one of my classes, the answer will be yes, but here is some advice. You want professors to be able to say more than, "This student was in my class last fall and earned an A." Get to know your professors outside of class. We often have to comment on aspects of your personality (such as creativity!) that are difficult to assess if there's little or no interaction outside of class. Ask professors from whom you've taken classes recently. If you were in my 071 class three years ago, I won't be much help. If you're planning on going to graduate school in CS, try to get involved in a professor's research. This also will help us write a strong letter.

Do you have any advice for writing a personal statement?

Writers must always know their topic and their audience. As Mor Harchol-Balter observes in Applying to Ph.D. Programs in Computer Science, a personal statement is really a research statement. The best way to know about the topic of research is to do it. If you haven't done any research, then it shouldn't prevent you from applying to graduate school, but keep in mind that the members of the admissions committee try to determine if you'll be successful at conducting research. Clearly, you will be able to write a much stronger statement if you describe not only the research you plan to pursue in graduate school, but also the research you've actually done. Doing research with a professor also results in a stronger letter of recommendation.

The audience is the professors of the admissions committee. While they are interested in you as a person, they're most interested in your research ideas, the research you've done, and the research you want to conduct in graduate school. As professors, they read a lot of precise, technical writing. As a result, your research statement should be precise and detailed. Don't be afraid of getting technical. After all, you're writing for professors who conduct research, not for your peers or for a general audience.

I have seen two main problems with research statements. The first is they aren't written at the appropriate level of detail. The second is they aren't written precisely. More coming...

Copyright © 2019 Mark Maloof. All Rights Reserved. This material may not be published, broadcast, rewritten, or redistributed.