which programming language is best for dynamic growth of inputs at run time for neural networks design?
I think it depends on wether you prefer writing fast code or writing code fast. To write fast code try assembler, C or C++. If you prefer to write code fast try C#, Java#, Python or functional languages like Haskell, D, Ocaml or lisp.
I am not sure of your goal, but the D programming language can be a good choice. It's a language with C-like syntax that shows a good compromise between efficiency and modelling flexibility and productivity.
If you are not so much interested in efficiency, Python is perhaps one of the best choices as there are many packages available to kickstart right away. Some useful libraries to begin with: Theano, NetworkX and Caffe
I think it depends on wether you prefer writing fast code or writing code fast. To write fast code try assembler, C or C++. If you prefer to write code fast try C#, Java#, Python or functional languages like Haskell, D, Ocaml or lisp.
I am not sure of your goal, but the D programming language can be a good choice. It's a language with C-like syntax that shows a good compromise between efficiency and modelling flexibility and productivity.
If you are not so much interested in efficiency, Python is perhaps one of the best choices as there are many packages available to kickstart right away. Some useful libraries to begin with: Theano, NetworkX and Caffe
To start with I would say python. However, myself being matlab programmer, you can use matlab with inbuilt functions to program ANNs. There are some libraries for C#, C and other languages.
It solely depends on purpose.
As you mentioned the best, I think it must be HDL(Hardware Description Language). The framework of CPU has limited the performance of neural networks, it's so bad in parallel calculation. As the neural networks is highly parallel, ASIC ,FPGA and GPU will do a better job than CPU.
thanks you all for all of your suggestion . OK here is the basic idea of me is .. i want to design a neuron and it's having a basic input characteristics and a collection of different kinds of transfer functions which gives optional to a neural network designer.. and in default mesh topological network is created in contrast . the network is being redesigned by itself during the learning phase of the network.
at last but not least . the network should be ideal to the what type of learning process it undergoes.
Matlab Neural Network tool is useful for your case. It is flexible for your adjustment. Thiano in Python is another option if you are familiar with Python. If you are planning to design a network from scratch with your own learning algorithm. Then, the best option is C++ due to its fast computing power.
I have practically used Matlab for neural networks. It works fine. If you have a consideration on speed, better have a try with C++.
i too used matlab for NN. but it will not let me to design my own network design and it will not allow me to implement my own learning algorithm so that is the reason why i have asked this question. presently i am designing a generic neuron and a generic network in a typographical nature (theoretically) and with the help of LAB VIEW software soon i will present those things ...
- Badges
- Science topic