I didn't find a lot articles with making artificial neural network on FPGA, only just reviews. So my question: is it possible? If it is - is it hard? why there is a few articles about it - may be it is too easy?
thank you for responces
Danii,
have you had a look on document paper already?
FPGA Implementations of Neural Networks
Edited by AMOS R. OMONDI
Flinders University, Adelaide,
SA, Australia
and JAGATH C. RAJAPAKSE
Nanyang Tecnological University,
Singapore
available at :
http://lab.fs.uni-lj.si/lasin/wp/IMIT_files/neural/doc/Omondi2006.pdf
might be useful for you.
regards.
Yes, it is possible rather a better and suited approach to ANN implementation where you can change neural parameters.
Yes, it is possible.
I recommend you to see this:
https://www.researchgate.net/publication/235418031_A_NEW_METHOD_FOR_FPGA_IMPLEMENTATION_OF_ARTIFICIAL_NEURAL_NETWORK_USED_IN_SMART_DEVICES
Conference Paper A NEW METHOD FOR FPGA IMPLEMENTATION OF ARTIFICIAL NEURAL NE...
You can also check this one
http://www.cee.uma.pt/morgado/ANGE/index.html
http://www.cee.uma.pt/morgado/ANGE/index.html
Yes, it's possible.
you have to respect some parameters:
1) you must choose a representation of data for implementationbecause in artificial neural networks you work with real numbers.
2) you must choose a representation of the sigmoid function because it is the difficult part in the implementation of the artificial neural network.
you can see my publications, and if you have any questions do not hesitate to ask them to me.
Not only it is possible, but probably using FPGAs for ANNs is the best way because of their extremely high level of parallelism which may outperform any MCU/DSP on the market.
I vote for what was recommended by Mohamad Hanif Md Saad - he pointed to a valuable resources.
Regards.
you can refer to my work in the subject. it is very simple design\
https://www.researchgate.net/publication/278246291_Two_Novel_Generic_Reconfigurable_Neural_Network_FPGA_Architectures
Conference Paper Two Novel Generic, Reconfigurable Neural Network FPGA Architectures
Yes... it is the best option if you want to implement an ANN on any hardware. Due to parallelism in FPGA (i.e. FPGAs can process the inputs at a time unlike the sequential i.e. one-by-one execution behavior of Microprocessors and Microcontrollers), ANN can be built up with more ease with faster and efficient outputs.
Yes.. Easy Solution.. Check out this
>> https://ieeexplore.ieee.org/stamp/stamp.jsp?arnumber=8651457
- Badges
- Science topic
- Similar topics
- Engineering
- Computer Engineering