CSCI-335 (Research) Final

Introduction to Machine Learning Research Project (150 points)

Due dates: Slides due by 4:10pm, May 1 (Friday) & Final Report/Materials due May 5 (Tuesday) 8:00am
Presentations: Finals Slot (5/1/2026 4:15pm-6:45pm)

For this work, you will work in groups of 2-3. The goal of this assignment is for your group to read one paper on fundamental research or an application in machine learning (details below). You should carefully read this paper and discuss amongst yourselves what the paper is about, the details of the statistical learning used in the paper, how the paper is related to the class, criticisms of the paper's experiments and setup, how the paper can be improved, and present a careful analysis of the contributions and results of the paper. Note that your choice must relate to one of the topics covered in this class (or an advancement/extension of one of the topics). For example, you can present a work on a graphical model (one we have not seen, such as the hidden Markov model) or something related to the support vector machine, for example, drilling down on how a kernel-based SVM is developed and applied to a medical problem (for example). The key is that you are able to teach the class each and every detail clearly (which means you might need to look materials elsewhere to aid you. Do not assume any knowledge from your classmates except what we have covered in lecture and the basics of linear algebra, probability/stats, and differential (vector) calculus. For example, if you present a topic in Computer Vision using convolutional networks, you will need to teach the class the concept of convolution and how it fits in the context of backpropagation of errors.

Paper selection

Your papers should come from reputable sources (conferences or journals) related to the topic at hand. Your topic should be related to the course content. You can use the library database to find papers - look in the ACM Digital Library or IEEE Explore databases for best results. Another place to consider is the end of each chapter in the textbook, where they discuss related works. For those papers, you might also use Google Scholar or CiteSeerX to find forward citations (i.e. those papers that cite the mentioned paper) which will be newer and may be more interesting. The paper must be at least 6 pages long, and the paper must be working on a topic or problem using machine learning, particularly any one of the methods we discussed in class or an extension of one thereof.

Deliverable: A final write-up, presentation slides, and a short presentation given to the class on the final exam slot day.

Writeup (Due May 5, 8 am)

The writeup should carefully explain the papers referring to the current state of the art as relevant from the papers you have read. You should make sure to summarize as well as describe the specific advances from the paper in your own words, and to contrast them to the basic topics we have learned in the lecture. You are required to use LaTeX for the writeup, and must specifically use the conference document class / format provided here (use: this, and here is a basic starter template file you can modify, or use the exact equivalent on Overleaf). I expect the write-up to be about 2-3 pages of text content in length in this format (if you use figures, this means you will have more than 3 pages, but while you may have figures, you must a sufficient amount of meaningful text). NOTE: label your report file via convention: lastname1_lastname2_lastname3_report.[file_suffix]

We will be looking for (at least) a section with meaningful insight/content w.r.t. to the following:

An intuitive, high-level overview of the paper. Why is the paper important? What does it do that had not been before it (what is novel)?
A clear, well-written summary of the specific advances made in the paper, present key related math (equations/algorithms/proofs/theorems/lemmas) with careful explanation of each variable/operation and the general point of the equation (do NOT just throw/copy math from the paper -- you must digest and re-present it to demonstrate that you deeply understand it).
A discussion of the related state-of-the-art (sota) as related/relevant to the paper.
A discussion/analysis of how this paper relates and is different from the topics we cover this semester
Clear/intuitive & mathematical description of any concepts in the paper that were NOT cover this semester but needed to undestand the paper (for example, if your paper uses convolution inside an artificial neural network, you must present/develop concept clearly in both words & math).
A discussion/conclusion where you thoughtfully present what you learned, what additional steps were needed to understand the paper beyond the fundamentals covered in this class, and, crucially what was wrong with the paper and what you would suggest is needed for improvement (or, what should be done to extend the paper for future work -- what would be critical to examine next).

Deliverable: You will hand in the compiled (PDF) report to a MyCourses dropbox. The report will be graded on formatting, style (coherence, clarity, completeness), content (at minimum does it adhere to the above criterion/bullet points).
NOTE: label your slides file via convention (replace "lastnameX" with a teammate's surname): lastname1_lastname2_lastname3_report.pdf

Presentation (Due May 1, Talks start at 4:15pm; Slides due 4:10pm on MyCourses)

You will also be required to present to the class on your topic, approximately 10 minutes. You may use Powerpoint-style slides or not (you could user Beamer, for example), as you see best to present the particular topic. Also note that your classmates should be well aware of the basic material, so you should review any course content very briefly before proceeding to the advances discussed in the papers. All members of the group must contribute to the presentation, not necessarily in equal measure. There should also be slides/content that address the same rough points described above for the report.

Deliverable: You will hand in any slides that you use to a MyCourses dropbox. The presentation will be graded on content (i.e. appropriate level for your fellow students), materials as appropriate, and presentation style (coherence, clarity, ability to answer questions). NOTE: label your slides file via convention:
lastname1_lastname2_lastname3_talk.[file_suffix]

Peer/Team Evaluation (Due May 1): you will also be required to submit a bit of text outlining your team contributions and evaluating your team performance. In this bit of text, you must answer 2 questions: 1) what did you specifically focus on / contribute to the project (slide/talk-wise, paper-wise, etc.), and 2) how would you describe and evaluate your team performance qualitatively (in a few sentences)? This is where you can list any possible issues or concerns and/or strengths/positives. Due on MyCourses May 1, 11:59pm. Note that each member must write and submit their own private team evaluation document.

Grading