Adding a few overarching comments/questions regarding the paper in this issue. There's several other inline comments in the .tex file following the tag: "%% JBCOMMENT."
Also, if you choose to stick with LaTeX, I would suggest working out of Overleaf to facilitate real-time collaborative editing, change tracking, GitHub integration and easier compilation: https://www.overleaf.com/
The title of the paper refers to LSTMs for intensity and trajectory forecasting, but in the text, we reference intensity, wind, wind speed, and barometric pressure, presumably all referring to the same quantity. It would probably be helpful to standardize on terminology.
Do the top-level sections (i.e. scope, tropical storm background, etc.) conform to AMS requirements? Some conferences expect specific sections or a subset of sections to be present, others are more open-ended.
Related to the above question, but is the "Scope" section part of the AMS template? If so, how is it typically written? I'm used to the ACM/IEEE format, which usually begins with an introduction that provides a compressed background, scope of research and description of results (basically an expanded abstract).
There's some past/present/future tense mixing in the paper. I would suggest going with present tense throughout (e.g., "we will demonstrate" and "we demonstrated" become "we demonstrate").
The paper mixes references to predict and forecast - I think the latter is probably more common for the given domain and for time series in general, and is also more specific (forecasting could be considered a subset of prediction). Suggest replacing prediction with forecast throughout.
Not sure where the references are located. In LaTeX, they would be referenced in the text as "\cite{reference ID}", which will be replaced with the numeric reference when compiled.
There's a few sections that I don't think would be included in the paper - maybe they're a carry-over from the design exam:
Definitions and Acronyms (might be in an appendix)
Applicable documents (these would typically be cited in the references section)
Assumptions and Dependencies (not sure what this is)
Virtual Environment Dependencies (not sure what this is)
Requirements
At least a portion of the assumptions section reads like a description of the experiment setup. Would suggest describing it as such. There's probably no need to call out things like Docker and the exact run command - that's a pretty low-level implementation detail that would be captured on the GitHub README. The reader would likely be more interested in things like data source/provenance, data processing steps and train/val/test splits.
The design alternatives section, especially the portions that justify the selection of the Bi-LSTM model vs. alternatives, would probably be most logically packaged with the description of the ML approach. Discussions regarding the data processing steps would go in experiment setup.
Adding a few overarching comments/questions regarding the paper in this issue. There's several other inline comments in the .tex file following the tag: "%% JBCOMMENT."
Also, if you choose to stick with LaTeX, I would suggest working out of Overleaf to facilitate real-time collaborative editing, change tracking, GitHub integration and easier compilation: https://www.overleaf.com/