The qualification exam verifies that you have the foundational knowledge and research skills to conduct independent research. It consists of two parts: completing the required theory, systems, and AI/ML courses with grades of A or B; and an oral examination before a faculty committee. Full-time students with MS transfer credits should aim to qualify by the end of their first year or beginning of their second. Those without should aim to qualify by the end of their second year. All students must qualify by the end of the semester after completing their 36 hours of required coursework.
Each semester, the program selects two Computer Science faculty from the PhD program faculty to serve as fixed members of the qualifying committee for that semester. For each student qualifying, their examination committee consists of those two faculty (for consistency across students), a third faculty member selected for their familiarity with the student's research area, and the student's adviser as a non-voting member. This is not your Dissertation Committee.
At the start of the semester in which you intend to qualify, you and your adviser notify the program. The program director, the fixed committee members, and your adviser then consult to identify the third committee member for your examination.
You and your adviser select a significant paper in your specific area of research and submit it to the committee for approval. A significant paper is one that made a clear advance in the field and has been cited and built upon by other researchers, with contributions relevant to Computer Science. It should be a work that will play a notable role in your own research question.
Once your paper is approved, you are scheduled for your oral qualifying examination at the CSCI 8101 Doctoral Seminar. At least one full week before your scheduled examination, you will submit a 6-page writeup to the committee. This writeup reviews: (1) the background literature that led up to the selected paper, (2) the computational methods used, (3) the methodology and significant contributions, and (4) briefly addresses open questions.
This writeup is not a Dissertation Proposal. It is an examination of your ability to deeply understand existing work and its significance. It should draw mostly from background surveying and reproducing results that you have been engaged in since day one. Its purpose is to give you structured preparation and to let the committee come in with familiarity.
At the oral exam, you give a 20-30 minute presention of what you wrote up, followed by examination questions from the voting committee members only. The committee will examine your understanding of the research you presented, with emphasis on the foundational material covered in the prescribed course areas of theory, systems, and AI/ML. The voting members determine if you pass using the rubric below.
Qualifying exams are scheduled once per semester, with a maximum of three attempts to pass. The third attempt is essentially an appeal by you and your adviser. If you do not pass on the third attempt, you cannot continue in the program. At that point, your credits and work can be applied to the MS in Computer Science if you do not already hold that degree.
Foundational CS Concepts
I am satisfied that the student can apply relevant foundational computer science concepts from theory, systems, and ML/AI to the selected work and sub-field.
| Score | Description |
|---|---|
| (5) Excellent | In addition to "Very Good" criteria, demonstrates the ability to draw on more advanced concepts, and articulate connections and implications not explicitly addressed in the selected work. |
| (4) Very Good | Consistently uses appropriate terminology and well-understood concepts from theory, systems, or ML/AI. Identifies key implications that follow from foundational principles without prompting. |
| (3) Satisfactory | Uses mostly appropriate terminology. Invokes standard concepts from foundational coursework rather than resorting to ad hoc explanation. Aware of key implications. |
| (2) Marginal | Unfamiliar with some standard terminology or concepts expected from foundational coursework. Frequently resorts to ad hoc or informal explanation where established concepts apply. |
| (1) Unacceptable | Severe lack of basic CS knowledge. Cannot connect the work to foundational concepts in theory, systems, or ML/AI. |
Systems and Models
I am satisfied that the student understands the key computational systems and models in the selected work — how they work and why they are designed that way.
| Score | Description |
|---|---|
| (5) Excellent | In addition to "Very Good" criteria, demonstrates strong generalization of the details and their implications beyond the specific usage in the selected work. |
| (4) Very Good | Clearly explains the details of how the systems and models work and why they were designed that way for this work. Correctly applies standard concepts and principles in their explanations. |
| (3) Satisfactory | Can explain the core, well-understood details of how the systems and models in the selected work function. Can articulate some of the design choices that differentiate them for this work. |
| (2) Marginal | Displays "black box" thinking. Struggles to explain the inner workings of the systems and models beyond surface-level description. |
| (1) Unacceptable | Cannot answer fundamental questions about how the systems and models work. |
Literature Review
I am satisfied that the student has familiarity and understanding of the related computational work that shapes the selected work and defines the sub-field.
| Score | Description |
|---|---|
| (5) Excellent | In addition to "Very Good," articulates overarching principles that unify results across the sub-field and tie into foundational concepts and principles in Computer Science. |
| (4) Very Good | Thorough coverage of related work. Clearly articulates how each work is significant, how it built on prior work, and how it advanced the sub-field. |
| (3) Satisfactory | Familiar with the most impactful related works. Can explain the connection to the selected paper and articulate the advances made by each prior work. |
| (2) Marginal | Lacks thoroughness in the literature review. Missed significant results that shaped the selected work or fails to draw connections and explain significance. |
| (1) Unacceptable | No understanding of the sub-field, unaware of prior work. Views the selected work in isolation. |
Methodology and Evaluation
I am satisfied that the student understands the methodology and evaluation in the selected paper, including the justification of specific choices and awareness of alternatives.
| Score | Description |
|---|---|
| (5) Excellent | In addition to "Very Good" criteria, identifies subtle assumptions or limitations beyond what the authors acknowledge. Situates the methodology against alternatives from the related work to expose non-obvious trade-offs. |
| (4) Very Good | Understands the methodology and its trade-offs. Readily explains why specific approaches were chosen over alternatives and accurately articulates what was proved or demonstrated. |
| (3) Satisfactory | Understands what was done and what was proved or demonstrated, including general limitations. Can discuss alternatives, though comparative analysis may be limited. |
| (2) Marginal | Unprepared to discuss alternatives — cannot explain why this approach and not another. Superficial understanding of the evaluation or why the authors made the choices they did. |
| (1) Unacceptable | Misunderstands the paper's core results or methodology. Cannot explain the paper's significance or justify the choices made. |