All information about Neural Network [closed] - neural-network

As it currently stands, this question is not a good fit for our Q&A format. We expect answers to be supported by facts, references, or expertise, but this question will likely solicit debate, arguments, polling, or extended discussion. If you feel that this question can be improved and possibly reopened, visit the help center for guidance.
Closed 10 years ago.
I heard about Neural Network but there are so many resources and i want to know concrete use of it and if possible some small code source with comment. ^^

you might find the following questions to be of use:
What are some good resources for learning about Artificial Neural Networks?
Open-source .NET neural network library?
Looking for a Good Reference on Neural Networks
also you can find two examples of a neural network in this article which also has source code availible:
http://www.codeproject.com/KB/recipes/aforge_neuro.aspx
if these are not helpful then please make your question more specific

For a rather quick start you could check:
the Wikipedia article on Artificial Neural Networks which provides further references
the brief introduction to Neural Networks by D. Kriesel or maybe
Jeff Heaton's website where he also published parts of his book Introduction to Neural Networks (with examples in Java resp. C#)
And of course you could search stackoverflow ... ;)

They are used for example to program AI, speach synthesizers, to recognize fingerprints.
And also for many non programming based topics, like checking on airports baggage if it has something dangerous inside, selection of employees :-)
It is so huge topic, and theres no one concrete use.

A good (free!) place to start is the FAQ for Usenet comp.ai.neural-nets:
http://www.faqs.org/faqs/ai-faq/neural-nets/part1/
If you want to code your own, I suggest considering Murray Smith's "Neural Networks for Statistical Modeling".

Related

Which is good for Video Processing applications opencv or matlab? [closed]

As it currently stands, this question is not a good fit for our Q&A format. We expect answers to be supported by facts, references, or expertise, but this question will likely solicit debate, arguments, polling, or extended discussion. If you feel that this question can be improved and possibly reopened, visit the help center for guidance.
Closed 9 years ago.
I am working on Multiple Object Tracking algorithms. Which tool is most suited to develop my own algorithm OpenCV or Matlab ?
Matlab is a faster way to develop algorithms and testing existing systems. However, when it comes down to producing a commercial and portable application you need open source libraries like openCV. Especially some heavy weight algorithms like object tracking can be quite slow in Matlab while openCV has optimized builtin methods and algorithms about your subject already
I have been working with OpenCV for quite some time and I have to say it is a great tool. The new documentation is much, much better than the old one and the ease of opening multiple camera streams is awesome. I have no experience with Matlab so by bias is pretty high.
OpenCV can be used with many different languages (Again, I don't know about Matlab) including C, C++ and Python. I have found that the Python interface is really nice as they use Numpy arrays for manipulating data.
Here is a simple example of some Python code some students wrote to track both white Frisbees and rectangular targets.
https://github.com/Team3574/2013VisionCode

How to learn modelica? [closed]

As it currently stands, this question is not a good fit for our Q&A format. We expect answers to be supported by facts, references, or expertise, but this question will likely solicit debate, arguments, polling, or extended discussion. If you feel that this question can be improved and possibly reopened, visit the help center for guidance.
Closed 10 years ago.
I'm completely new to Modelica and was wondering what is the best way to learn on my own? Also I will be using it with MapleSim 5 (or 6 when it will be available) so any information regarding that would be helpful too.
I'll need Modelica to create a few components for a thermal model, then simulate with MapleSim and hopefully I will also manage to co-simulate with another software (EnergPlus) using FMI when it will be available for MapleSim and BCVTB.
Sorry for not being specific with my question, really any reference to a good Modelica guide would be great, thanks.
Clearly, I am biased, but I like to think that my book "Introduction to Physical Modeling with Modelica" is a great way to explore the features of Modelica in the context of engineering problems. I noticed that Wolfram seems to recommend the book as well.
But in the interest of balance, there are two books by Peter Fritzson on the topic of Modelica as well.

The best tool for visualizing ontologies? [closed]

As it currently stands, this question is not a good fit for our Q&A format. We expect answers to be supported by facts, references, or expertise, but this question will likely solicit debate, arguments, polling, or extended discussion. If you feel that this question can be improved and possibly reopened, visit the help center for guidance.
Closed 10 years ago.
I am looking to visualize ontologies for understanding and making others understand. I would prefer to have top-down hierarchy of classes. I have used OntoGraph which comes with Protégé, but I am not happy with the figure :(. I have also used Microsoft Visio and stencils available for Ontology, but that too has not impressed me.
Through this question I would like to know which is the best tool for visualizing an ontology for general purpose, and/or for specific purpose (like writing an academic paper).
Thanks.
~Codera
I do not have complete answer for your specific problem, but you may try to review the Mike Bergman's list of tools (for ontology visualization).
Note, RDF is a graph, so you may be able to use any graph-visualization software. "Only" need to map RDF to tool-specific model.
While not an ontology specific tool, thejit is a very nice javascript visualization toolkit. I've used it in the past for custom semweb-related visualizations. You won't get much out of the box support but if you don't mind putting in some coding time, you can get some nice looking graphs.
I can highly recommend you to utilize the graphical editor of TopBraid Composer Maestro Edition. The graphics are not the most beautiful ones. However, it's interactive ;-)

Code generation with Machine learning [closed]

As it currently stands, this question is not a good fit for our Q&A format. We expect answers to be supported by facts, references, or expertise, but this question will likely solicit debate, arguments, polling, or extended discussion. If you feel that this question can be improved and possibly reopened, visit the help center for guidance.
Closed 11 years ago.
I am very interested in machine learning. i have an idea to develop a program which can generate a source code according to a given standards. for example if i have trained the system to how to write something to console and the standard way of using variables then it should be able to generate a source code for anything which i ask to print something to the console, likewise. Is there any resources source codes or existing open source projects?
You might be interested in Genetic Programming. It's a genetic algorithm that operates directly on programs.
The demo I saw used Lisp (Common Lisp, I think), which is a natural fit, but any language with eval should be decently easy to use.
The Wikipedia page lists a whole pile of implementations, many of them open source, so you should be able to start there.
This sounds like this is would be a Constraint Satisfaction [CSPs] or Search Problem, these approaches would probably be your best bet:
Genetic Algorithm
Some sort of goal seaking heuristic [with negative/positive reenforcements]
The algorithms that are under the CSP category.

Neural Networks for Mobile Robot Navigation [closed]

As it currently stands, this question is not a good fit for our Q&A format. We expect answers to be supported by facts, references, or expertise, but this question will likely solicit debate, arguments, polling, or extended discussion. If you feel that this question can be improved and possibly reopened, visit the help center for guidance.
Closed 10 years ago.
When designing mobile robot navigation using Artificial Neural Networks - there is a preference to use Back Propagation Methods instead of Feed Forward Methods, Why ?
Like Eric said, back-prop isn't an alternative to a feed forward networks -- its an addition. With just a feed forward network, your left with the task of figuring out all of the weights yourself, which very rarely makes sense.
Now, back-prop isn't a necessary addition, as there numerous other learning methods ranging from reinforcement learning to evolutionary methods like genetic programming. But you pretty much have to add some learning algorithm to your network to achieve any sort of decent task performance.