My master's thesis is in learning classifier systems (LCS), and I produced a time series classifier I called TSC.
I was attempting to using LCS to predict stocks, but it failed to produce results good enough for real-world use.
I am currently researching a much simpler approach that I hope will greatly outperform TSC.
Resources
The
The MIT Research HOWTO
(How to do Research at the MIT AI Lab),
MIT AI Lab working paper 316, October, 1988.
This is really useful to get a quick introduction to what graduate school will be like, and I wish I had read it a long time ago.
Sources
- MS&T Curtis Laws Wilson Library.
This is an MS&T-specific link, but the general idea still holds.
Go to your university's library, they will have people called
reference librarians
who do nothing but help people research stuff.
If you ask really nicely, one of them might even give you an overview of how to look for sources.
- ACM Portal [MS&T proxy]:
one of the first places to search for computer science material.
- IEEE Xplore [MS&T proxy]:
if it isn't in the ACM, it is probably here.
- CiteSeer:
this is really good for technical (mostly computer science) articles.
It often has the full article, it links to all of the references, it links to everything that links to that references the article, and it is just generally good all around.
- The Collection of Computer Science Bibliographies:
this has a BibTeX entry for almost any computer sciece article you will ever reference, so that you don't need to make it up yourself.
It also often has the article itself too.
- JSTOR [MS&T proxy]:
this is really useful if you are looking for something really old, or just feel like browsing.
- Google Scholar.
This is good for papers if you already know exactly what you want and they are recent, otherwise it kind of sucks.