M4:
Choose a tutorial for a learning problem other than image or text classification. A good
place to search is in Pytorch’s documentation.. Revise the tutorial to work with another
data set. A good place to search for data sets is Wikipedia.
Write a paper that documents your work and includes a link to the original tutorial.
Please address the following questions in that paper.
● How did you pick the dataset?
● What changes did you need to make to allow it to be used?
● What was the one or more difficulty you encountered and how were you able
to address and-or work around them?
● Once functional, did the tutorials implementation result in a model you would
feel comfortable deploying? If not, attempt to identify why and use what
you’ve learned in this course to get to a satisfactory outcome. Was it possible
and if not why?
Please submit the paper and the final notebook.
Adapting tutorials for new data sets can be a lot more difficult than it sounds, so don’t
delay getting started.
Please note that this assignment should be no longer than 1,000 words and conform to
APA guidelines. You must be succinct.
M6:
In this assignment you explore using ML services. You can use a previous assignment’s
data set and problem, or choose a new one. You also have the option of using an
existing NLP or CV service as a way to approach an unlabeled data set.
Examine the feasibility of using an existing NLP or CV service for your selected data set
and problem. Write a paper that documents your findings and addresses the following
questions.
● What ML service did you use and why?
● How well does the NLP or CV service do and is its performance sufficient?
● If the service performance isn’t sufficient, what is your recommendation for
next steps, and why? Should we use a custom built model or explore
AutoML?
Please note that this assignment should be no longer than 1,000 words and conform to
APA guidelines. You must be succinct.