The program consists of the following core activities

  • the requirements for the Ph.D. in Machine Learning;
  • the four core course requirements for the CNBC certificate;
  • exposure to experimental techniques in the form of a lab rotation;
  • a roughly semester-long project to satisfy the PNC first-year research requirement and the first of the MLD speaking skills requirements;
  • a year-long project that would satisfy both the PNC second-year research requirement and the MLD Data Analysis Project requirements;
  • participation in CNBC activities as a CNBC student; and
  • a Ph.D. thesis on a neuroscientific topic; if there is a single advisor, that person should be both a CNBC faculty member and affiliated with MLD; otherwise, the student may two co-advisors who, between them, have CNBC and MLD affiliations.

Additional satellite activities through the CNBC will also foster students’ professional and scientific development.

Current PNC students looking to join the joint PNC/Machine Learning program may petition to do so by sending an email to the PNC co-directors explaining how they intend to satisfy the degree requirements. See the Graduate Student Handbook for details.

Course requirements

Students must complete the four-course requirement of the CNBC certificate program:

They must gain graduate level training through coursework in the following three areas: (i) cell and molecular neuroscience/neurophysiology, (ii) systems neuroscience, and (iii) cognitive neuroscience. Recommended courses fulfilling this requirement include

  • (i) 03-762 Advanced Cellular Neuroscience (CMU) or NROSCI 2100/2101 Cellular and Molecular Neurobiology (Pitt)
  • (ii) 03-763 Systems Neuroscience (CMU) or NROSCI 2102 Systems Neuroscience (Pitt), and
  • (iii) 85-765 Cognitive Neuroscience.

To complete the computational requirement, students must take:

  • 36-759 Statistical Models of the Brain (CMU) / Math 3375 Computational Neuroscience (Pitt)
  • Math 3370 Mathematical Neuroscience / CMU course number TBD.

Note that this is not exactly the same as the standard CNBC computational requirement.

To meet the course requirements in MLD they successfully complete the 5 ML Core courses, with an average GPA of 3.5 or better. These include:

  • 10-715 Advanced Introduction to Machine Learning
  • 10-702 Statistical Machine Learning
  • 10-705 Intermediate Statistics

Plus any two of the following:

  • 10-703 Deep Reinforcement Learning or 10-707 Topics in Deep Learning
  • 10-708 Probabilistic Graphical Models
  • 10-725 Convex Optimization
  • 15-750 Algorithms or 15-853 Algorithms in the Real World
  • 15-780 Graduate Artificial Intelligence
  • 15-826 Multimedia Databases and Data Mining
  • 36-707 Regression Analysis
  • 36-752 Advanced Probability

Any substitutions or exemptions from coursework must be recommended by the student’s advisor and approved by the program co-directors and the co-directors of graduate studies in MLD.

Program Milestones

First year research requirement: By the end of the first calendar year in the program, all students are required to complete a computational project. This project will be evaluated by a committee consisting of at least three faculty, of whom at least two are PNC training faculty. The project requires the student to identify a biological problem, understand the data collection process, articulate the goals of building a model or performing a particular kind of analysis and implement this computational approach. In some cases this project may be a precursor to the student’s eventual thesis project. This project cannot substantially overlap with a project completed for a class, although it may be on the same topic as a class project, provided that it represents a substantial extension of that work.

Students should begin formally discussing this research project no later than the end of the spring term. Initial steps should include forming this committee and organizing a meeting to discuss/outline the project with your committee. The committee must consist of at least two PNC training faculty and one MLD core or affiliate faculty. The makeup of this committee should be approved by the program co‐directors. At this first meeting the committee should approve the project proposal or indicate steps necessary to identify a new project. Then, before the start of the fall term, students must schedule a committee meeting where they present/defend their results. This meeting should occur before Oct 15. The initial part of this meeting involves a 30 minute presentation by the student, which is open to the public. This will be followed by a meeting with the committee and the student, during which the committee will ask detailed questions about the work. Based on this meeting, the committee will evaluate the student’s work and will decide whether a student passes, fails or needs to revise the project, subject to re‐evaluation. Questions about the content of the presentation should be raised by the student with committee members well before the evaluation meeting. Students who wish to enter the joint program from MLD after their first year may be able to waive this requirement with the permission of the PNC training faculty.

Second year research requirement: All students will be required to complete a deeper computational project. The student’s work on the project should demonstrate that the student has 1) the ability to analyze and interpret experimental data in a particular area 2) the ability to develop and implement a computational approach incorporating the relevant level of biological detail and 3) the ability to organize, interpret and present the results of the computational work. This project should be a body of work suitable for publication. It is expected that the research will be written up as a paper to be submitted to a journal in the relevant field. In the second year, students are expected to work on research about 1/3 of their time during the academic year and full time during the summer. In most cases this project will be on an area related to the student’s eventual thesis project, and in most cases it should be completed by the end of the student’s second calendar year in the program. In addition, the results of the project will be presented publicly in the form of a seminar. This seminar must be advertised to both the PNC and ML communities at least one week prior to the event. (To advertise, send the talk announcement including the date, time, place, title, abstract, and faculty committee to the ML and PNC graduate program coordinators.) This project, which counts as the Data Analysis Project in MLD, will be evaluated by a committee consisting of at least three faculty, two of whom are PNC training faculty and one of whom is ML faculty appropriate to the topic.

Ph.D. Thesis proposal: Required coursework should be completed by the end of the third year. During the fourth year a Ph.D. candidate should present a thesis proposal first to his or her thesis committee and then to the CNBC and MLD community.

The student will have two joint advisors, one from MLD and the other a CNBC faculty member from outside of MLD. A thesis committee will be formed and should be composed of at least four members, one of whom is an external member (typically from outside CMU and Pitt); two must be PNC training faculty; two must be MLD faculty; and at least one CMU or Pitt member must be from a discipline outside of statistics and computer science. The thesis committee is subject to approval by the PNC training faculty and the MLD faculty.

The thesis proposal should include: a succinct summary of the proposed research problem; the significance of the proposed research; a review of relevant literature relating to the problem; a review of the candidate’s work leading up to the thesis, including preliminary results; a clear statement of remaining research; and a tentative schedule for completing the work. It should also conform to the stylistic requirements for thesis proposals in MLD. The thesis committee must offer its preliminary approval of the proposal. The student then arranges to present the proposal publicly, so that CNBC and MLD faculty and other community members can attend. Formal approval is conferred by the MLD faculty and the PNC training faculty.

Ph.D. Thesis Defense: Normally, the dissertation is completed during the student’s fifth year. The final defense is a public presentation, in accordance with the College and University requirements for the Ph.D. It is the candidate’s responsibility to ensure that the Departmental, College and University guidelines are followed for publicity of the defense and availability of the thesis document at least two weeks prior to the defense. Note that the defense must be held at least 21 days before the date the degree is awarded.