I am working in a sub-field of computer science (deep learning), more specifically a particular application of LSTMs to learn simple functions. My work is building on the work of another researcher to an extent that I am using the same model, parameters and problem definition. I am writing a research paper based on my findings.
My work is different in the sense that I am introducing some different concepts from another area and trying to make a generalisation about the very nature of ANNs (using his work as an intermediary). And since his work is the only such publication till now, I am taking some things directly in my work eg. Problem Definition or Model.
So the question is whether it will count as plagiarism? If yes then how can I circumvent it as I have to write many things as it is, there’s no other way to get around that as I am using them as it is.