To start I should provide a brief summary as to why I am asking this. Currently I am a university student beginning the last assignment of an honors degree in Digital Media Design. I have become wildly interested in Artificial Intelligence as of late, and am looking to include some nifty things in the said assignment.
The main topic I wish to cover is emergent behaviors, which would be visually communicated back to user as they interact with this project. Anyway, I have been reading a great text by Stephen Marsland called Machine Learning. I can’t really pretend to know a lot about the topics yet, but I feel as if it is a guide to an intermediate level of many Machine Learning topics and techniques.
Machine Learning by Stephen Marsland
I have been going through the book strategically cover to cover and am about half way. I have learnt about Neural Networks like the multi-layer perceptions and radial basis networks, being used to solve linear, classification and time-series problems/situations. I then struggled through some ideas about support vector machines and, slowly, I am getting my head around high-dimensionality and kernel-tricks.
However, things get sad now, I am beginning to drag myself through some of the content and I am really struggling to hold a conscious understanding of the techniques and mathematics being presented. Further more, as I try to program the examples I am taking longer and longer to piece together the blanks (so to speak). Now, I am not looking for a simple, hard and fast way out, but I am also conscious of the time I have left to pull this project together.
If anyone else has had a similar experience, which they are now on the other side of, could they perhaps share an opinion on a couple of things with me. Would you suggest jumping right into the genetic algorithms and working back over the other topics as I gain more experience, or would this be unmanageable? Or perhaps another topic that works nicely leading into genetic algorithms? I have a reasonable grasp of mathematics and programming concepts, but am completely self taught (my degree focuses on design rather than computer sciences and software engineering).
And perhaps if so, is there any resources you might be able to share with me, or snippets of advice? I will be hugely grateful, and thank you to everyone in advance.
TL;DR Need a push in the right direction with genetic algorithms for an emergent behavior project. Please share any good resources.
(Apologies for long question, first time asker).
I recommend you Toby Segaran’s book: Programming Collective Intelligence.
Actually, it doesn’t contains much academical knowledge (with a huge scary equations and tons of math) but real word primers that could help you to intuitively feel ML algorithms.