Doctoral Program Rules
Academic Year 1996-7
College of Computing
Georgia Institute of Technology

The doctoral program in the College of Computing prepares exceptionally qualified individuals for research careers. Graduates receive the degree of Doctor of Philosophy for performance of original research resulting in a significant contribution to the discipline's body of knowledge.

The program has several phases, and normally requires four-six years. Although there is no set number of hours of course work required for the attainment of the degree, 80 credit hours of course work is typical for this degree. (These hours include work performed for a Master's degree if the student chooses to petition for such a degree upon meeting its requirements.) This document summarizes the College of Computing's Ph.D. program and should be supplemented by the general Institute rules regarding Ph.D. dissertations (available through the Georgia Tech catalog).

All degree requirements must be completed within five years from the end of the quarter in which a student passes their Qualifying Examination. When both the dissertation and examination are satisfactory and the requirements of residence and the minor field have been fulfilled, the candidate will be certified as qualified to receive the degree of Doctor of Philosophy. Students must submit their petition for graduation during the quarter preceding the one in which they wish to receive their degree.

Goals of the Ph.D. Program

Quality and excellence
To train graduate students who are broadly knowledgeable in computer science and deeply knowledgeable and creative in a specialized area
To evaluate students primarily on the basis of their research capabilities and progress, but with strong secondary emphasis on their full professional development
To encourage research early in students' graduate careers
To help students focus and acquire depth of knowledge in their specialty area(s)
To guide students to develop creative talents early in their graduate careers

Program Overview

The program that achieves these goals has four parts. These are presented along with guidelines on advisement and milestones.

Orientation and Introduction to Research
Breadth
Achieving Ph.D. Candidacy (Qualifying exam and thesis proposal)
Defense of Research

The program is meant to be substantial, but flexible enough to support many degrees through the College of Computing and through interdisciplinary research. All will require a core of knowledge in computer science, a set of specialized courses, a depth of knowledge in some area, creativity, and progress on the part of the students.

1. Orientation and Introduction to Research

An orientation course, CS 7100, integrates students into the Ph.D. program. The course has several parts. First, students are taught good strategies for getting through the program, what is expected of them as students, etc. Second, students are introduced to the facilities. Third, students are introduced to the College's areas of specialization. Each area is introduced as a general research area, with a presentation and discussion of interesting open issues. Then, work in the College of Computing is presented. Students work on small exploratory group projects in a variety of areas as part of the course. Students are required to take the orientation course the first time it is given during their tenure in the Ph.D. program.

2. Breadth

The aim here is for students to acquire breadth in computer science. Breadth is defined as the knowledge that a Master's graduate is expected to possess. The breadth requirement has four parts:

Students demonstrate that they have core knowledge important to all of computer science. This knowledge is at the level that is expected of a solid graduate of the College's undergraduate program.
Students take a set of courses to achieve breadth across many different areas of computer science.
Students achieve more depth in some breadth areas by taking additional courses.
Students attend departmental colloquia in a variety of areas.

Core areas

Students must demonstrate competence in the five areas of computer science identified below. These areas are fundamental to the core of computer science, and all students should have a good working knowledge of each. The required knowledge in each core area will be approximately equivalent to the course material of an upper-level undergraduate course in that area.

Algorithms and Data Structures
Computer Organization
Formal Languages and Automata Theory
Operating Systems
Programming Languages

Students have two ways of exhibiting competence in these core areas: (1) by passing a written examination or (2) by achieving an appropriate grade in a pertinent course. Exact grades and courses required can be found on the Breadth requirement confirmation form.

Knowledge of the core material will be tested by a 4-hour comprehensive written examination, the Core Exam, that tests competence in all five areas. The examination is given early in the fall quarter. It is expected to be at approximately the same level as the advanced GRE test in computer science. All reading appropriate in preparing for GRE advanced examination in computer science is appropriate for this examination as well.

The Core Examination is designed to be diagnostic. For each area passed, the core requirement in that area is satisfied. For each area not passed, the student will be directed to take an appropriate course, either at the graduate or undergraduate level. Students are expected to make an acceptable grade in the core area course. The Core Requirement confirmation form lists appropriate courses and acceptable grades. Because the exam is diagnostic, students may elect not to take the entire exam or certain parts, and simply take the appropriate courses. The core requirement should be fulfilled by the end of a student's sixth quarter in the Ph.D. program. (Students can repeat the exam or areas of the exam, if necessary.) The one exception to this rule is students entering the program without an extensive computer science background (perhaps from another major). In this case, students will be given extra time to take appropriate undergraduate courses and meet the core requirements.

Breadth Areas and Course Requirements

In addition to competence in core areas, students must learn about a number of broad areas of computing. A total of nine graduate courses are to be taken to meet the breadth requirement. Students take courses in six of the following eleven areas, and take at least two courses in three of those areas. Students are expected to secure a grade of B or better in each course, and must maintain a grade-point average appropriate to a Ph.D. student, that is, 3.5 or better. This Breadth requirement should be completed by a student's ninth quarter in residence.

AI and Cognitive Science
Computer Architecture
Database Systems
Graphics and Visualization
Human-Computer Interaction
Intelligent Systems and Robotics
Networking and Communications
Programming Languages and Compilers
Software Methodology and Engineering
Systems (including Operating Systems, Distributed and Parallel Systems)
Theoretical Computer Science

The choice of the breadth areas and substitution of a graduate course by a seminar in the chosen area should be done in consultation with the academic advisor with prior approval from the graduate program committee or area advisor. With the permission of the graduate committee, students aiming toward interdisciplinary research can substitute a breadth area outside of the College of Computing for one of the above.

Students with strong graduate background in computer science transferring from other institutions may seek exemption from some of the breadth requirements. Exemption should not be considered by students to be automatic, however, since one purpose of the breadth requirements is to acquaint students with research in the College of Computing. Students should contact the appropriate area advisor with supporting materials from the prior course in order to seek transfer credit toward breadth requirements. The signature of the area advisor and graduate coordinator must be secured to realize transfer credit.

Note that courses used to fulfill the core requirements can also be used to fulfill the breadth requirement.

Minor Field of Study

In addition to an adequate knowledge of the major field of intended research, the student must demonstrate mastery of some other, smaller body of knowledge--the minor field-- outside the College of Computing. The purpose of the minor is to encourage a wider interest on the part of the student and to provide a broader basis for the evaluation of the student's capabilities. A minimum of fifteen hours of course credit of this type, approved by the Advisory Committee, must be completed on a letter grade basis while enrolled at Georgia Tech.

The minor will normally consist of at least fifteen quarter hours of work in related courses, chosen by the student in consultation with his/her faculty advisor. These courses should be at the 6000 level or above, but certain 4000 level courses may also be used with the advisor's approval. A maximum of 6 hours of directed study can be used. Students must maintain a B average in these courses. Graduate courses taken at other institutions may be included in the minor. Once the student has satisfactorily completed the minor, a confirmation letter, accompanied by course grades, must be sent to the graduate office for final approval and recording.

Although the student need not complete the minor as a prerequisite for admission to candidacy, the chosen field must be submitted, approved, and the program of study must be completed before clearance for the Ph.D. degree.

Participation in Colloquia

Students are encouraged to attend colloquia in a variety of areas throughout their tenure at Georgia Tech. The college has many such series, distinguished lectures, and informal lunch talks. Students should attend lectures spanning a diverse group of topics to learn about the many disciplines of computing.

Guided Research Projects

Faculty and students must work closely together if students are to be prepared for later research. Students should work on guided research projects from their second quarter onward. The purpose of these projects is twofold: to provide a learning experience for the students so that they are ready to work independently on research by their third year and to teach students about independent research so that their creativity can be tested at qualifying exam time. Students should sign up for at least three credit hours of CS8503 each quarter after their first quarter until passing the qualifying exam. From their first summer onward, students are encouraged to sign up for four to twelve hours of CS8503 per quarter. It would be appropriate for a student to sign up for only CS8503 some quarters. Students may work with one or a series of faculty members on guided research projects. Early on, it is advisable to work with a number of different faculty to learn about different areas of computing.

3. Achieving Ph.D. Candidacy

Students achieve candidacy in the Ph.D. program after proving their creative talents, showing that they have adequate depth in the area of their choice, and passing their dissertation research proposal.

Students must achieve depth beyond the narrow area they will focus on in their Ph.D. work. Depth will be tested by a qualifying exam administered any time prior to the end of the ninth quarter of the students' tenure in the Ph.D. program (in residence). Students are encouraged to take their qualifying exam earlier, however.

Students will be responsible for taking the courses and reading the material that is necessary in order to achieve depth. Getting involved in a research project also serves to promote depth in an area, since students grow to understand the problems that must be addressed.

In order to receive guidance in achieving appropriate depth, students should consult with faculty in the areas they are considering specializing. The faculty will advise students about the depth they should be achieving, how broad their knowledge should be, and how to achieve that depth and breadth.

Qualifying Exam

The qualifying exam (otherwise called the Area Exam) evaluates whether a student is capable of Ph.D. work. The exam tests the student's depth of knowledge in necessary areas, and the student's creative talents. Passing the exam requires not just capability, but also quality, in both areas. The objectives of the Area Examination are to test:

the deep understanding of the area in which the student intends to do Ph.D. research (both breadth and depth of knowledge),
the creativity of the student and
the student's capability for deep analytical/independent thought.

The exam is meant to cover one or two areas relevant to the research the student will pursue for his/her dissertation. Each student will specify one primary area and possibly one secondary area. This exam should be used as a milestone to determine whether a student has the potential to complete a Ph.D. The guidelines for the examination are listed below.

Breadth

The breadth of the exam is determined on an area basis. Current areas for the exam (as well as for the Breadth Requirement) include AI & Cognitive Science, Computer Architecture, Database Systems, Graphics and Visualization, Human-Computer Interaction, Intelligent Systems & Robotics, Networking & Communications, Programming Languages & Compilers, Software Methodology and Engineering, Systems and Theoretical Computer Science. Faculty within each area decide on questions that will test each students' knowledge within that area. Since the primary areas have subdivisions (or sub-areas), the faculty in each area will decide which sub-areas should be included on the exam. If an area has five defined sub-areas, it might be that each student is required to know three of them or maybe all five. This is up to the discretion of the area faculty.

Timing

The exam usually will be scheduled twice a year in each area. Typically, this will be in the late Fall and late Spring quarters. Students will take the exam in a given area as a group. This exam must be taken no later than the ninth quarter in the Ph.D. program. One quarter before the exam, an advisor will be chosen and an exam committee will be put together and approved by the Graduate Committee.

Committees

A number of different committees are involved in the exam process. Their roles are explained below. First, there is an implicit committee comprised of the area faculty who are responsible for making up and grading exam questions. Second, there is the student's exam committee, made up of the student's advisor plus at least three faculty representing the area(s) in which the student is taking the exam. The exam committee members are a subset of the area(s) faculty and are chosen by the advisor and the student with Graduate Committee approval. For some areas, the exam committee may be the entire area faculty. The exam committee will administer the individual oral part of the exam and evaluate the student's performance on it as well as the student's overall performance (i.e., including the written portion, prior research, etc.). The exam committee makes a decision on whether to pass the student or not.

Format

The content of the exam is determined by the area faculty, which includes the exam committee. In some cases, it may be tailored to the student. For example, when a student has two areas relevant to their research, questions from both areas will be included on the exam. However, each student is required to show a deep understanding of his/her primary area. The depth of knowledge required for the secondary area is of a lesser degree. All exams will contain a written as well as an oral component. In general there will be questions testing the student's creative problem solving ability and analytic ability.

All students in a particular (primary) area are given the same general set of questions. The area faculty may allow each student to choose a subset of those questions to answer. In fact, it is recommended that students answer roughly two-thirds of the questions on the exam. The question(s) that test the student's creativity and research potential may vary between students. This is again up to the discretion of the area faculty. If a student has a secondary area, some questions in that area may be substituted for questions from the primary area.

All exams consist of two parts:

testing breadth within the chosen filed(s) of specialty
testing creativity, analytical ability and research potential.

The exam format follows:

Part (1) is tested through a written exam and part (2) is tested through a written exam and an oral exam. Part (1) will typically take between four and six hours. The written portion of part (2) will typically take three days. The oral portion of part (2) will usually last about two hours.

Written questions testing area breadth (part 1) are made up by the area faculty. The questions are also graded by the area faculty. Student identities are kept anonymous at this stage.

At least one additional question must test the student's creativity and research potential (part 2). The answer must be written as a report and presented in the oral exam. This question will be posed by the student's advisor and the exam committee. The written response and oral presentation will be evaluated by the exam committee. Students taking the exam at the same time will not necessarily answer the same question for part (2). Again, this is a decision made by each area.

Pass/Fail

After the exam, the members of the exam committee, except for the student's advisor, will discuss the student's performance and recommend a pass or fail. The advisor may submit written documentation regarding the student, which will be used in the decision process. The decision to pass/fail the student depends on the performance in the exam as well as the capability of the student (e.g., as demonstrated by prior research accomplishments) to complete a Ph.D. The decision reflects the opinion of the majority of the committee. An appointed chair of the committee will be responsible for drafting a letter summarizing the committee's decision and for sending the letter to the graduate committee for official recording.

A conditional pass on the exam will not be allowed; students either pass or fail. Under rare circumstances, the exam committee may recommend that the student retake the exam. Such a recommendation must be communicated to the graduate committee where a decision will be made based on justification presented. A student who fails the exam is encouraged to meet the requirements for the Master's degree.

Changing Areas

A student who changes his/her area after taking the area exam will be required to show depth in the new area by retaking the exam in this area.

Proposing a Ph.D. Topic

Ph.D. Proposal will take place at least one quarter after passing the Qualifying Exam. The proposal will be administered by the student's Advisory Committee (the advisor plus at least two other faculty members), chosen jointly by the student and the advisor. A majority of members of the Advisory Committee must hold their primary appointments on the College of Computing faculty. In general, a student should maintain continuity of the Advisory Committee from exam time until thesis defense.

Students will write a proposal document in which they will both propose their research work and set a schedule for completion. The proposal will be given to each committee member at least three weeks prior to an oral presentation.

Candidacy

A student who has passed the qualifying exam and proposed their dissertation research has achieved Ph.D. Candidacy. After achieving candidacy, students are entitled to sign up for CS9000 (Doctoral Dissertation) on a pass/fail basis. To satisfy Georgia Tech requirements, each doctoral student must also spend at least three full-time quarters in residence, and must complete fifteen hours of course credit (exclusive of their minor) on a letter grade basis.

4. Defense of Research

Students will defend their research one to two quarters before they expect to complete writing their dissertation. At this time, it is expected that the research concept is well-formed, that any implementation is done, and only minor work is left to do. The dissertation defense committee, according to Institute rules, must consist of at least five people, with at least one from outside of the College of Computing. The defense of research is open to all faculty, however. Thus, students have two sets of people to address: the computer science community as a whole and their own technical community. The first group requires a presentation of the big picture along with a concrete statement of the student's contribution. The second requires the technical detail necessary to prove those contributions. When appropriate, students will demonstrate their work.

At the end of the defense of research, the advisory committee indicates requirements on the content and presentation of research in the thesis. Students will know what is expected in their thesis and whether their research has passed muster. The outcome of a defense of research, to be determined by the Advisory Committee, can be a clear pass (almost always with suggestions), a conditional pass with a set of issues that must be dealt with appropriately by thesis completion time, or a clear fail. While all students are expected to pass their defense of research the first time, students who fail may be given an additional chance. Because Ph.D. work requires timeliness, passing an oral defense remains valid for two quarters. If a student does not finish writing by then, another defense may be necessary.

If the defense of research is open to the public, then it may serve as the formal Institute Ph.D. dissertation defense.

Milestones and Scheduling

Year 1

Students enroll in CS 7100, the Ph.D. Introduction course
Students take the diagnostic Core Exam to place out of taking Core Courses if desired
Students choose an Interim Academic Advisor at end of their first quarter
Students begin taking directed study research courses with individual faculty in second quarter
Students should begin taking courses to fulfill Breadth requirement

Year 2

Students should finish taking courses to satisfy the Core Requirement
Students should focus in on a particular research area, taking advanced course in that area, and begin to prepare for Qualifying (Area) exam
Students should select a research advisor
Students should begin writing tech reports and research papers about their work

Year 2-3

Students should form Qualifying Exam committee and take exam
Students should complete courses for Breadth Requirement
Students should begin taking courses to complete outside Minor requirement
Students may potentially form dissertation committee and do Ph.D. thesis proposal

Year 4+

Students should form disseration committee and do Ph.D. thesis proposal
Students should continue and complete courses for outside Minor
Students should write conference and journal papers about their thesis research
Students should defend their dissertation research
Students should complete writing of PhD dissertation and graduate!

Advisement

There are several kinds of advisement students need: area advisement, academic advisement, guided research advisement, and thesis advisement. The following overview should help all students get well- integrated into the Ph.D. program.

Early Guided Advisement (Interim Academic Advisor)

During their first and second years, students will take courses and work on guided-research with the faculty. It will be helpful if each student has a particular person at this time to help them choose courses, seek out research projects, achieve breadth and depth, and generally progress well through the program. Consequently, at the end of their first quarter in the program, students must select an Interim Academic Advisor to serve this role. The Interim Academic Advisor is not a research advisor in the thesis sense. Rather, it is a person who will assist the student, monitor their progress, and be able to speak in more detail about the student's progress during student reviews. As a student transitions into working on a particular research theme and topic, he or she should then choose a Research Advisor in place of the Interim Academic Advisor.

Area Advisement

Area Advisors provide advice about a research area and the projects and courses therein. Areas are defined as the identified set for the breadth requirement presented earlier. Each year, one faculty member from each area in the College will serve as Area Advisor. The Area Advisor has three responsibilities:

to advise students of course options and requirements should the student decide to focus in that area
to help students decide on an advisor for their first and continuing CS8503 courses, and to guide them in setting up a schedule that will prepare them for research in their area
to help explain to students the general requirements of the program (who to see about what, and what forms need filling out, etc.). In the event the area advisor cannot answer a student's question, the graduate coordinator should be consulted.

The area advisor is available to provide advice to all students in their research area. Any student who is deciding between several areas would be wise to consult advisors in each of the focal areas.

Research Advisor

At least one quarter before the qualifying exam, students must choose, with the consent of that individual, a Research Advisor. The Research Advisor will oversee the student's qualifying exam and, later, the student's Ph.D. research. Although a research advisor does not have to be chosen until a quarter before the exam, students should be in contact with faculty in their chosen area well before that to ensure that they have the appropriate depth and breadth of knowledge for research in that area.

Advisory Committees

Students choose an advisory committee in preparation for their qualifying examination. Usually the members of this committee will carry forward to the student's dissertation committee. In some cases, the advisory committee will have to be updated from time to time. This is to be done jointly by the student and the advisor.

Changing Advisors

Students entering the Ph.D. program may or may not know the area in which they want to specialize. They are encouraged to work with a variety of faculty during their first two years in the Ph.D. program so that they are prepared to choose an area and advisor in their second year. Some students will be able to choose an area early on but will find it difficult to choose an advisor. Thus, the advisor chosen for the qualifying exam may not be the advisor who oversees the student's Ph.D. research.

Because getting a Ph.D. is a long and hard process, students should work on research that is interesting to them. Invariably, some students will select to change areas and/or advisors during their tenure. In general, it will be easier to switch to a new faculty member in the student's chosen area than to switch to a faculty member in another area. Nonetheless, changing advisor and/or area is acceptable and will occur in some cases. In all cases, advisors must agree to advise the student. The student will be required to show proof of competence in the new area. Also, in all cases, students will be required to show that they are progressing through the Ph.D. program at an appropriate rate. A student who changes areas or advisors once probably won't suffer, but changing areas or advisors more than that after the second year hinders progress in the program and is therefore discouraged.

Evaluation Procedures

Every 6 months, in the late fall and spring, the faculty meet to review graduate student progress. All faculty members discuss their students and, for each, provide concrete evidence of the student's progress. Students receive written feedback from the graduate committee after these meetings. If a student's progress and/or performance is judged to be unsatisfactory by the faculty, the student may be put on probation or be asked to leave the Ph.D. program.

Optional Programs of Study for Ph.D. Student

Ph.D. students in the College of Computing have several options concerning their program of study. Most students pursue a Ph.D. in Computer Science. An optional certificate program in Cognitive Science allows students to receive additional interdisciplinary training while still fulfilling all of the requirements of the College's Ph.D. program. An interdisciplinary program in Algorithms, Combinatorics, and Optimization (ACO) provides yet another alternative. Finally, the College also participates in the interdisciplinary Bioengineering Ph.D. (and Masters) Program.

Students desiring a Cognitive Science certificate enter the Ph.D. program in one of the participating departments (Computing, Psychology, or Industrial and Systems Engineering). They complete all of the requirements of that department's Ph.D. program. In addition, students must also satisfy the course requirements set by the Cognitive Science Certificate Program Committee in order to receive the certificate in addition to the Ph.D. degree.

The ACO program works differently. While students enroll in one of the participating departments (Math, Computing, or ISyE), this enrollment is in the ACO Ph.D. program, not in the unit's standard Ph.D. program. The student fulfills only the requirements of the ACO program, which include courses drawn from all of the participating departments, with a concentration in the home department. Both programs are defined and managed by an interdisciplinary committee.

Similarly, Bioengineering students are also assigned to a home unit (the College of Computing or other unit - see the Bioengineering Program information). Students then complete their coursework from a core of bioengineering courses from various departments, a concentration of courses from the student's home unit, plus mathematics, biology and elective courses. Students work with an College faculty advisor participating in the Bioengineering Program.

Doctoral Program Rules Academic Year 1996-7 College of Computing Georgia Institute of Technology