Dear all,
following the question of Shiva Gorjian (you can find as attachment), I was wondering the utility of Artificial Neural Network. I've read a lot of work based on ANN and I published some paper based on them (you can find the link below).
But if for example, we construct an architecture of only 1 layer and only 20 neurons with 5 inputs there are 100 weights to report for a possible formula.
So is this formula usable?
Or we can only state that the input parameters can represent the output variable?
But my question can go deeper. ANN is an powerful instruments so, if I change my input parameters, that I think are the best ones, and I put inside some other parameters or I change some of them and I train my network for a long time probably I can obtain the same results.
Sorry If my thinking is not so clear or my english is not so good.
https://www.researchgate.net/post/How_can_ANN_models_be_practically_used_by_designer_engineers_for_solar_radiation_prediction?amp%3Bch=reg&%3BloginT=ceKi2cQPVoAqo6A_iw14jroCteaMMhEpvgMxmWYVTic&%3Bpli=1
Article Surface tension of alcohols: A scaled equation and an artifi...
Dear Humbul Suleman,
thanks a lot for your opinion. Your is an interesting point of view.
Thanks!
- Badges
- Science topic
- Similar topics
- Engineering
- Instrumentation Engineering