so i was idle lately (or must i say most of the time) and i have always wanted something to do that involves software development that is worth my while. no i’m not planning to develop an enterprise level application for profit, at least not at the moment. i simply want to test if i could make something good out of my learning. as we all know good programming skills is just as precious as having the capability to generate new ideas for an application. and that’s why i’m practicing.
anyway back in college i worked on an online implementation of feedforward artificial neural networks for my thesis or what we call special problem (SP). it was pretty successful. if it was not, i wouldn’t have graduated because a fully accomplished SP was a requirement to graduate. however, i of course have a number of disappointments. i guess i was on a rush those days that i wasn’t able to plan the system well. it was done and it passed scrutiny but behind it was quite a list of wrong decisions.
now that i have a lot more time to work on whatever i want, i would love to create a similar application. only this time it will be a windows application so that i can have the network training process run on the background as a service as oppose to running it on the browser like what was in my SP.
this application i will call BotFactory. by that i don’t mean spam bots but really virtual robots that can be trained on certain tasks and assist their creators. i’m planning to use the same learning algorithm i used in my SP but this time i will employ the methods for speeding up the learning process. i wanted to do that in my SP before but i didn’t have enough time to test them so i just dropped them from the things i wanted to include in the system. i’m also planning to put into this little project all the things i know about software design and optimization at least just for practice. this can be good. who knows? maybe by the end of 2012 i’m already being assisted by a number of virtual robots in decision making haha. running the application should not be a problem. the processes involved particularly the training process is very computationally intensive but will only use very little resource. time complexity and is the thing that should get me worried. hopefully i’ll be able to think of a great optimization approach. wish me luck folks.
the details are still sketchy but if you would like to collaborate that will be cool. just leave a reply. (: