Respected Researchers,
i planning to do research on predicting the student performance using artificial neural networks.
could u suggest me ,
which software tool/package is best one?
and what are the best algorithms in ANN i may use for my research ?
Thanking you.
Check Wolfram Deep learning neural nets, also P5.JS ML library.
https://blog.wolfram.com/2018/06/14/launching-the-wolfram-neural-net-repository/
https://ml5js.org/
How many data points do you have per student? ANN is the right method only if you have a lot, otherwise other methods are a more natural choice.
It depends on the number of datapoints. I meant, NN works, if you have tons of inputs. But if I were you, I try to start with small range of data and design a baseline for classification or clustering.
So for the first step, you have to choose your classifier which it is supervised or unsupervised? It varies based on your inventory and existence of your data.
Secondly, check out your data dependency for prediction. If you need long dependencies so Recurrent Neural Networks like LSTM are suitable for your network, on the other hand maybe a simple feed forward works very well. You can even try some architectures like BERT.
Thirdly, for getting the maximum advantage of your NN, you can tune its hyperparameters (hidden layers, epoch and ...).
Moreover, For implementing your NN and leverage of GPU, Tensorflow or PyTorch are great.
Try with Multi-Layer Perceptron. It is simple and can be a good point of starting for the ANN field.
- Badges
- Science topic