In the program of CS, in which I am currently working, we have a group of students that for ending their Honour´s degree they should submit and present a research in the field; limited to their undergraduate knowledge. The process is that they get an adviser, read some papers, ending up with an article that describe their experiments and discuss their results. After that, they should make a small dissertation about their research findings. Before going to this stage, their research work is revised by two lecturers to give their insights about this work. Regarding this procedure I have found, that in some occasions, the following occur:
- The adviser of the student agrees that the current research work is adequate to the educational level, that is an undergraduate level. It is not a master´s or a PhD.
- One of the reviewers asks for modifications in the research made by the student, that in the majority of the cases, would imply to make deep modifications in the research work that would take more effort and, of course, time allotted for the student.
- In some cases, and this usually occurs in the dissertation, the questions asked from some of the other lecturers are very biased to their expert field. Making the student to feel discouraged of their work.
I mean how to deal with these situations, in this cases the adviser feel bad for the student also, but there are so many details that in my personal opinion is hard for the student to cope at this level. For example, imagine that a student X has chosen the topic of using neural networks for predicting diabetes treatment outcome. The student has made a literature review, obtained some datasets and perform some tests with the build model. Then, according to the adviser the paper is fine, and in the dissertation this happens:
- Lecturer X suggest to use GA for feature selection instead of PCA. If this observation is not raised the student can pass, but with a minimum grade.
- Lecturer Y, to test the knowledge of the student, starts to ask specific details of the architecture of the neural network. Lets suppose that Y is a mathematician, then he will start asking about what is the reason of the vanishing gradient issue and how to solve it. So this was a detail that the student has not considered, and according to this lecturer the student should fail.
So, what can we do to ameliorate this situation? Because at the end is bringing discomfort to either students, advisers and lecturers.