|
|
how to write c codings for ant colony algorithm by using floorplanning
|
|
|
|
|
You could start by checking out Google.[^]
It's truly amazing what information can be found.
I wasn't, now I am, then I won't be anymore.
|
|
|
|
|
Hello Experts,
I have a road map where i have a starting point A and Destination Point Z. There are multiple ways From A to Z.
Then How can i Get How many Number of Possibilities to Move from A to Z.
Thanks
If you can think then I Can.
|
|
|
|
|
This is a well-explored problem in graph theory. Google for something like 'number of paths in graph'. That should get you lots of references ranging from trivial to PhD material.
Cheers,
Peter
Software rusts. Simon Stephenson, ca 1994.
|
|
|
|
|
|
Without your being more specific about your goals, and the problem, I don't think much can be said.
Most real-world examples of combinatorial optimization have constraints ... without which you have potentially infinite recursion.
For example ... assuming by "road map" you mean, literally, a map of highways, streets ... you would certainly not want to find combinations that are possible, but that involve doubling-back and repeating segments of the journey.
Are you trying to determine the shortest possible route ?
You may want to look at the infamous "travelling salesman" problem[^], where an attempt is made to find the shortest route which passes once through each city of a list of cities.
best, Bill
"Is it a fact - or have I dreamt it - that, by means of electricity, the world of matter has become a great nerve, vibrating thousands of miles in a breathless point of time? Rather, the round globe is a vast head, a brain, instinct with intelligence!" - Nathanial Hawthorne, House of the Seven Gables
|
|
|
|
|
Hi,
this is my first post on this forum.
First a Short intro to the Self Organizing Map
A Self Organizing Map (SOM) ,-sometimes called a Kohonen network-, maps multidimensional data onto a two dimensional map. A classic example is to classify colors by the Red, green and Blue (RGB) values. So if we have a large (10 000) set of randomly chosen colors, with different shades of, greens, reds, blues, pinks, etc. the SOM can map them on a 2-dimensional map where similar colors tend to end up on the same node on the map. If we chose a map with the dimensions 5 by 5 we get a map with 25 nodes. As an example the pink shades might end up in one corner of the map close to violet shades and the violets close to blue shades and so on. So similar colors tend to be close to each other. So if we wanted to classify those 10 000 colors in our dataset into 25 classes we could look at each colors node and use the information to se what class each of the colors belong to.
But how can we create clusters of several nodes based on their similarity to each other?
modified on Thursday, August 18, 2011 9:19 AM
|
|
|
|
|
SharpSim wrote: this is my first post on this forum.
And, as it seems, most probably the last ! 
|
|
|
|
|
Sometimes I get the impression that CodeProject is a forum of uneducated people.
|
|
|
|
|
WOw, are you stuck somewhere in the time-space of last week ?
|
|
|
|
|
This question may be more suited in this forum.
<a
href="http: www.codeproject.com="" forums="" 326859="" algorithms.aspx"="">http://www.codeproject.com/Forums/326859/Algorithms.aspx[^]
Regards
[Edit]
Fix link.
It was broke, so I fixed it.
modified on Thursday, August 18, 2011 7:58 AM
|
|
|
|
|
S Houghtelin wrote: This question may be more suited in this forum.
<ahref="http: www.codeproject.com="" forums="" 326859="" algorithms.aspx"="">http://www.codeproject.com/Forums/326859/Algorithms.aspx[^]
Can we move the question to that forum?
|
|
|
|
|
|
I knew it. I did not, but I knew someone would !
|
|
|
|
|
You forgot to chill it using Liquid Nitrogen.
Panic, Chaos, Destruction. My work here is done.
Drink. Get drunk. Fall over - P O'H
OK, I will win to day or my name isn't Ethel Crudacre! - DD Ethel Crudacre
I cannot live by bread alone. Bacon and ketchup are needed as well. - Trollslayer
Have a bit more patience with newbies. Of course some of them act dumb - they're often *students*, for heaven's sake - Terry Pratchett
|
|
|
|
|
Seconded. Motion is passed.
|
|
|
|
|
What an interesting question, well I think that a Self Organizing Map should ............ oh shiny! or in my case .... hmm chocolate! 
Ali
|
|
|
|
|
Welcome to the algorithms forum.
I'm not sure what triggered such negative responses to your question (originally posted to another forum which it wasn't appropriate for ? then moved here with all its baggage of snipes ?), but it seems to me your question is related to algorithms.
I'm not familiar with SOM, or Kohonen networks, but I come here, to this forum, to share my own algorithmic interests, and to learn, and be stimulated by, others interests, so I'm happy when someone introduces me to a concept I haven't encountered before.
One way I think you can make your question here more enjoyable for other participants is to share the context of your question: why you are interested in this type of algorithm, what this problem or question is in the context of your work, or your study.
If you share what you have done so far, in code, or in theory, or where you find your progress in implementing this type of algorithm/solution is blocked, as specifically as possible, then, I think you'll get responses that will be most helpful.
Your description of Kohonen networks, and the example given of organizing a color-space, reminds me of the wonderful program 'Visual Thesaurus'[^] which, when it came out several years ago, was a real 'revelation' ... to me.
VT 'maps' semantic affinities between words in a thesaurus using font-size and font-attributes to represent a virtual third dimension, and its 'dynamic:' the 'cloud' of associated words is being continually updated.
And, yes, the 'word-tag-clouds' seen on many website these days are a form of 'static' VT, imho.
I wonder if VT type algorithms might be considered an algorithmic problem of the type you are describing ?
Where I get interested in the issue I think is raised by your question is: the how of ... once you establish the evaluative criteria by which a multi-dimensional data set is 'distilled' into some useful summary representation ... you then express that representation in a form which you can use a compiler (or lexer-parser) to create : as visual interface, as meta-language (DSL), etc.
While some data-sets, I think, have an 'easier,' perhaps statistics-based, answer to how you represent summary information in a way that facilitates communicating essential 'meaning,' and enables sophisticated exploration through 'drill-down' in hierarchical, or 'flat,' visual interfaces ...
Other 'problem spaces,' ones I think are much more interesting, are going to have internal 'semantic' networks based on ? Bayesian probabilities ?
An example: a personal interest of mine is the evolution of Buddhist sacred iconography in S.E. Asia, where I live. There is a kind of an algorithmic component in the sense of a 'guiding meme' of the 32 lakshanas originating in Sri Lanka which express a sense of the normative visual attributes of representation of the teacher now known as 'the Buddha.'
And then, there's the whole complex chaotic history of how styles of representation of 'the Buddha' evolved through conquest and assimilation (capture and relocation of artisans, and or images, and supplies of raw materials, being a key goal in S.E. Asian geo-political warfare).
And then, there unique events, as when King Tilokkaraja of Chiang Mai, in the 15th. century, as part of establishing more centralized control of the nascent proto-polity of a more cohesive northern Thailand meta-state (Lanna), promoted the image of 'the Buddha' in 'royal raiment' as part of a larger campaign to have himself recognized as a divinely sanctioned ruler (dhammaraja, chakravartin).
So you have a leit-motiv (the lakshanas), you have historical mutation and change, you have top-down visual innovation and 'orthodoxy' imposed at times, and you have serendipitous changes, sometimes enabled by a changed supply of raw materials, or contact with other cultures, or innovation by particular craftspeople, or sudden availability of new technology
And then, it gets even more complex: at times 'retrograde' change occurs where the visual style of an older, or imagined archaic, style becomes fashionable, or required, or is re-vivified by some individual innovator.
And at the maximum 'edge' of change ... fractal ? ... you have meta-cultural differences which can be quite profound. To the 'western mind,' shaped so profoundly by ideas of absolute historical origins, a non-recurrent sequence of time, cumulative progress, Aristotelian syllogism, the division of art and science (trivium, quadrivium), mind and body, Psyche and Techne: a culture where time is experienced as an 'eternal return' of a boundless cycle of dynamic manifestations of opposites (what Jung would have called 'enantidromia') is 'alien' ground.
That's the kind of 'data-space' that I am interested in. How to make an 'interface' to that kind of data ... mmm ... there's the rub.
best, Bill
"In the River of Delights, Panic has not failed me." Jorge Luis Borges
modified on Thursday, August 18, 2011 9:04 PM
|
|
|
|
|
BillWoodruff wrote: originally posted to another forum which it wasn't appropriate for ? then moved here with all its baggage of snipes
Yup. Started life in the lounge. Didn't actually get stomped as hard as I expected.
Cheers,
Peter
Software rusts. Simon Stephenson, ca 1994.
|
|
|
|
|
It's rated low because it was posted in the Lounge and then moved here with the low votes intact. It seems like an interesting enough algorithmic post.
|
|
|
|
|
Thank you Bill
BillWoodruff wrote: If you share what you have done so far, in code, or in theory, or where you find your progress in implementing this type of algorithm/solution is blocked, as specifically as possible, then, I think you'll get responses that will be most helpful.
Well, the color sorting problem is one of the examples often used to describe the Self Organizing Map (SOM) algorithm.
There is several implementations of the SOM algorithm. There exist task specific SOM algorithms and general purpose SOM algorithms. Here is three general purpose implementations:
SOM Toolbox 2.0 (Matlab 5) By Tevo Kohonen and his research team , sort of open source
http://www.cis.hut.fi/somtoolbox/[^]
SOM_PAK (ansi c) also By Tevo Kohonen and his research team, sort of open source
http://www.cis.hut.fi/research/som_lvq_pak.shtml[^]
There is also an Octave/Matlab compatibel developed based on the above SOM Toolbox 2.0 called
Melikerion , open sourcehttp://www.finndiane.fi/software/melikerion/[^]
The Melikerion research is done by for the Finnish diabetic nephropathy study. But the software is general use.
Another good SOM/Kohonen software package is Ron Wehrens "kohonen: Supervised and unsupervised self-organising maps". It is written in R http://cran.r-project.org/[^]
Kohonen package for R
http://cran.r-project.org/web/packages/kohonen/index.html[^]
There exist also several SOM/Kohonen projects here on codeproject.com
I am studying individual persons career paths wich include many person level variables, such as, age, sex, education, graduation, family related variables, income, neighborhood, house, etc. I know how to classify my data on to a 2-d map with SOM_PAK and Melikerion. As the first step the data is scaled and the nodes of the SOM is set to initial values usually based on a random sample of the input data. Then one member of the input dataset is compared to all of the nodes by calculating the multidimensional euclidian distance. The node wich is closest to the input data "winns" and are updated with the values from the one input data member. Also the surrounding nodes are adjusted as a gradient where far nodes are adjusted less than the closest. This is repeated for the whole dataset several times until the map converges. The convergens is accomplished by decreasing the adjustment of the nodes. When the input data is clustered the SOM algorithm can reveal the clustering and that can be observed by visual inspection of the map. My basic problem is how two find the clusters made of several of the map nodes. The SOM creates a map where the nodes that are close to each other is more alike than nodes far apart.
I would like to find clusters of similar nodes and use those as the classification of my data. That informatin could be used in the analysis of the differences and similarities between different individuals careers and income levels related to their educational history, work career and other family characteristics.
|
|
|
|
|
It seems to me that this problem boils down to selecting the best projection of your problem space onto two axes. Do you already know how to create such a map? Are you just looking for cluster analysis algorithms? A search should turn up some 2D cluster analysis algorithms, it must be a solved problem.
|
|
|
|
|
You sent a shiver down my spine when I read your post. I've just being doing this as part of my Uni AI course, although we were doing it with analysis of birds contaminated by an oil spill, and had 17 inputs going to a 8x8 SOM grid.
We have been using JavaNNS for all the course work.
What you basically need to do is set up your network with each of the 3 input nodes (RGB) connected to every single node in your SOM.
You haven't said what language you are trying to create this with, but searching google there are heaps of results with libraries to use.
Here is one example article;
Implementing the Kohonen Neural Network[^]
|
|
|
|
|
I will try to formulate my question about Self Organizing Map (SOM) that I have. My research problem is that I want to classify the entries of a large dataset so that similar entries end up in clusters where the the variation within clusters are less than between the clusters. I know this can be accomplished with the SOM algorithm. Now to the point:
If the SOM is started with a large map lets say 10 by 10 nodes (100 nodes in total). So I map my data onto the 100 node map. Then I visually inspect the result and find that there is lets say 7 distinct clusters. Is it then approperiate to reduce the number of nodes to a map 3 by 3 nodes (9 nodes in total) and again run the SOM onto this much smaller map. And then use the nodes of this smaller map as my clusters?
|
|
|
|
|
General 
News 
Suggestion 
Question 
Bug 
Answer 
Joke 
Praise 
Rant 
Admin 
Submit an article or tip
