On Wed, Mar 08, 2017 at 11:06:27AM +0530, Vivek Bhave wrote: > Hi, > > I am Vivek Bhave, a sophomore undergraduate student from India. I went > through all the project ideas, for Google Summer of Code 2017, and I > got fascinated with the implementation for k-center problem.
Hi Vivek, Thanks for getting in touch. I think the k-centers project could be an exciting way to spend a summer. Answers below... > I have read and understood exactly what the k-center problem is. Then > I went on to learn about dual-tree algorithms. I got a few resources, > mostly research papers. I got Ryan's thesis work, which I actually > liked because it explains many things in an order I can understand. I > have read the first two chapter in that, but it would be great if you > could tell me what should I read now. Should I keep reading the thesis > in order so that I have a clear understanding of dual tree algorithms, > or I could just stop somewhere in the middle, and look at other > resources. How proficient should I be with dual tree algorithms to do > this project? Since you would be designing and implementing a dual-tree algorithm, you would need to have a pretty complete understanding of them. You are welcome to read as much or as little of the thesis (and related papers) as you like, but at the end of the day, you should be sure that you are able to understand the dual-tree algorithms that are implemented in mlpack, both from an algorithmic and implementational standpoint. > Could I ask these specific questions on the IRC channel? So that once, > I am familiar with the code base, I could start contributing to > mlpack. Apart from the folder I mentioned, could you please suggest > any other specific part in the code base I should read. Sure, feel free to ask on IRC or on the mailing list. But in either case it is better to ask publicly so that the set of people who could respond is larger. I would also recommend, after you understand the tree code, to spend some time with the actual dual-tree algorithm implementations found in src/mlpack/methods/. > I just wanted to know, is there an algorithm, which you have designed > for this problem, and is the challenge about implementing it, or my > job would be to modify an existing algorithm, or design and implement > that. Personally, I would love to design and implement an algorithm, > but I am just fine even if it's just about implementing it as well, I > am really interested in working on the problem. I just want to know > what exactly will be the task. Yes, this project would be to design an algorithm, and then spend the summer implementing it. Unfortunately, Google funds students for implementation and not for research, so any proposal submitted would have to have the basic approach for the algorithm clearly outlined, so that when the summer started the path to implementation would be clear. However, if it works out and we get good results, it is likely that the work will be publishable in a conference, workshop, or journal. Let me know if I can clarify anything. Thanks, Ryan -- Ryan Curtin | "I like this game." [email protected] | - Coach _______________________________________________ mlpack mailing list [email protected] http://knife.lugatgt.org/cgi-bin/mailman/listinfo/mlpack
