I am a new PhD student. If I use existing methods to solve existing problems, but no one has applied these methods to solve these problems, is such a procedure a sufficiently new contribution for a paper?. For example if i use vgg16 (convolution neuron network) model to classify one problem that belong to Biology field.

  • 3
    By "existing problems" do you mean already solved problems or do you mean open problems? Also in what field? – JoshuaZ Nov 19 '18 at 13:46
  • If you add a little more specificity (perhaps field) and maybe an example so it is more clear what you mean, I'd think this question could be re-opened. For instance if you want to use an already existing statistical method on a new data set - in many fields that's the vast majority of published papers! In most of technology and social sciences, in fact, methods papers are the minority, and using an existing method is the norm. But just using a method for a new problem alone is not always a contribution if you can't say speak to motivation, what you learned, and why it matters. – BrianH Nov 20 '18 at 4:58
  • for example if i use vgg16 (convolution neuron network) model to classify one problem that belong to Biology field. – Mohamed Bakr Nov 20 '18 at 6:11
  • 1
    The concept of scientific contribution is not binary and it depends a lot on the domain: in your case (CNN for biology), maybe it's not a contribution for ML but it is for biology. It's actually very common to see valuable contributions which apply a known ML technique to a new domain, but it completely depends: is it a trivial adaptation? how does it help the target domain? etc. This is why you should ask this question to your PhD advisor, it's exactly their role to guide you thanks to their experience of the field. – Erwan Nov 20 '18 at 12:23
  • You should ask this question to your advisor. It is literally part of their job to advise you on what problems you could work. – mlk Nov 20 '18 at 15:01

In mathematics and similar fields, new solutions to previously solved problems may be extremely valuable.

The statements of theorems aren't really the most important thing in mathematics. They are markers that we use to simplify the proofs of yet more advanced theorems. It is the proof methodologies that are the truly stunning advances in mathematics. If you can provide a new way of looking at a problem, or, better, a class of problems, you have made an important advance and can/should publish it.

But make sure that your paper points out that the advance is in the methodology, not the theorem.

Some "hard sciences" may have the same characteristics. How you look at a problem can be as important as a solution.

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