Here's how you can approach problem solving in research from different perspectives.

Root cause evaluation is the most important aspect of the problem solving. Once you identify issues or problems in your research then only solving them will be possible. You can solve problem as it is from the project or you can breakdown the problem into components so it will be easy to fix you the issues. Sometimes you need to investigate a lot to find cause so you need to approach with some strategies to solve it.

The think which I have experienced that nowadays, people go with others research problem and try to solve in a different way, that's how, they do something called research paper and all, which is good but try to solve the issue with an optimized way if you are solving existing problem or with completely different approach. Identify the key issues, like how the problem has been solved earlier and how we can solve it in a different and better way.

Reviewing the existing literature is a way that can help looking at a research problem with an open and flexible mind. Read papers, reviews and analyses.

If you are working with ML and DL, then data is the fuel to build something. If the data is not cleaned or relevant then we must find that. Try to do some surveys in big range cuz we need data in big amount and try to preprocess the data and clean the data, drop something/columns which are not required in the project. Try to increase the quality of the data, cuz more the data is pure, the accuracy you would get.

Analyzing the data is very important. The Data we need and making pipelines and giving the most required the relevant information and looking at statistics and feature engineering is the most important part which is also considered in kind of in the analyzing part because it has to analyze the data and give most relevant columns and remove others which are not required.

I do see a lot of faulty decisions by starting with an answer and trying or needing to apply it to a problem. Do not give away the previous important work by concluding with what you expected without an in depth evaluation. Is the preferred option really the best option or are you starting out with a debit difficult to make up with a below par solution choice? Thoroughly vet alternatives to appreciate up front what advantages and challenges will be incurred on the path to realizing any potential solution.

One can also have brainstorming sessions with the research team or other experts in the field. Sometimes we may face a mind block and hence may not be able to solve the problem, or include different perspectives ourselves.