Where do you draw the line? I saw many funding proposals for extremely abstract category theory justified on the basis it helps number theorists solve problems related to cryptography and protect e-commerce. Between friends everyone knows this is not true. Am not against abstract research: given how cheap it is, I don't understand why we can just allow it to be funded without having to provide justification that makes everyone feel a bit of a criminal while writing the proposal. It should be seen sort of how we public commissions art.
Another type of mini-fraud: you could say almost 80% of applied machine learning papers contain serious methodical mistakes that are almost equivalent to falsifying data, and what's worse the authors are well aware of and even the reviewers. I am talking about the kaggle-type ones where you apply some kind of PCA or neural network on some relatively small dataset.
Another type of mini-fraud: you could say almost 80% of applied machine learning papers contain serious methodical mistakes that are almost equivalent to falsifying data, and what's worse the authors are well aware of and even the reviewers. I am talking about the kaggle-type ones where you apply some kind of PCA or neural network on some relatively small dataset.