Can we use Convolutional Neural networks for a non image dataset for prediction?

The dataset is a record of student academic details

I know that CNN is mostly used in computer vision and image processing for analyzing visual imagery. And it is also used in natural language processing for text classification.

I am new to this please guide me through

Thank you!!


Short answer:

Yes, absolutely

Longer answer:

As a rule-of-thumb, you can always use any technique on any problem, as long as you define it properly. CNNs are meant to work on signal data. 1D signals can be sound waves, financial records, etc. 2D signals are usually images. 3D signals can be videos (with time being the 3rd dimension). And obviously, you can have as many dimensions as you'd like.

So as long as you can turn your input data (let's say a single training instance) into a signal, then be my guest. However, I fail to see how this dataset can be viewed as a signal.

| improve this answer | |
  • $\begingroup$ Thank you for the information can you please help me with some information as to which neural networks can be used for prediction purpose with such dataset.I searched but couldn't get any which explains it $\endgroup$ – Sachin Yadav Oct 23 '19 at 10:11
  • 3
    $\begingroup$ @SachinYadav you do not need to go into deep learning for this. Realistically bog standard ML algorithms like random forest or boosted options (e.g. XGB) are designed for these kinds of task and will less computationally intensive. If you insist LTSTM models are probably more appropriate for the task you have in mind. $\endgroup$ – Fnguyen Oct 23 '19 at 13:34
  • $\begingroup$ @Fnguyen thank you for the information I will surely refer it $\endgroup$ – Sachin Yadav Oct 23 '19 at 15:00
  • 1
    $\begingroup$ @Valentin Calomme Many might consider a sound wave as a 2D signal defined by time and amplitude, in the same way you use video as a 3D example, with time as an additional dimension. $\endgroup$ – image_doctor Oct 24 '19 at 15:05
  • $\begingroup$ Is it a valid assumption that there is a local effect in these datapoints? $\endgroup$ – S van Balen Oct 28 '19 at 10:59

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.