User:Jadrian Miles/Thesis proposal feedback: Difference between revisions
Jump to navigation
Jump to search
New page: Feedback includes /dhl notes from during the talk, /jfh email after the talk, /dhl meeting notes after Spike's email, and conversations with other committee and faculty members... |
No edit summary |
||
| Line 1: | Line 1: | ||
Feedback | Feedback on my October 2010 thesis proposal included [[/dhl notes]] from during the talk, [[/jfh email]] after the talk, [[/dhl meeting notes]] after Spike's email, and conversations with other committee and faculty members that were not recorded. Feedback on the style and organization of my talk was synthesized into the HOWTO page [[Give a talk]]. Feedback on the content of my ideas and their individual presentation is listed on this page. The committee has asked that I formally respond to these comments. | ||
# The term "unambiguous" in the thesis statement is poorly defined. The statement that the solution to this system could not be computed using other models may be unprovable. | |||
# More direct comparison to a greater variety of related work is needed, including global macrostructure models (such as spin-glass models), signal regularization schemes, and microstructure models. | |||
# The order in which curve clusters are selected for candidate merges is unclear, as are the consequences of different ordering choices. | |||
# The "black-box math" used to describe the curve-clustering algorithm's energy function is insufficiently specific. What is the principled reason for an algorithm to choose a middle ground between 300,000 singleton clusters and one whole-brain cluster? | |||
# The nature of the optimization algorithm for image-based bundle refinement is not specified. How does it relate to established techniques? What is the relationship between image differences and the space of candidate bundle refinements that they suggest? How does it avoid overfitting? | |||
# How will the model accomodate white matter fascicles that project outside of the brain? | |||
# How will the macrostructure modeling results be validated? | |||
# Would chained applications of the macrostructure fitting process converge quickly? If not, why not? If so, what is the nature of the fixed-point solution? | |||
# The stated choice of a Rician distribution for axon diameters seems inappropriate. | |||
# A concrete "toy example" is needed for proper evaluation of this proposal. This could take the form of a computational or physical phantom. | |||
Revision as of 02:48, 20 October 2010
Feedback on my October 2010 thesis proposal included /dhl notes from during the talk, /jfh email after the talk, /dhl meeting notes after Spike's email, and conversations with other committee and faculty members that were not recorded. Feedback on the style and organization of my talk was synthesized into the HOWTO page Give a talk. Feedback on the content of my ideas and their individual presentation is listed on this page. The committee has asked that I formally respond to these comments.
- The term "unambiguous" in the thesis statement is poorly defined. The statement that the solution to this system could not be computed using other models may be unprovable.
- More direct comparison to a greater variety of related work is needed, including global macrostructure models (such as spin-glass models), signal regularization schemes, and microstructure models.
- The order in which curve clusters are selected for candidate merges is unclear, as are the consequences of different ordering choices.
- The "black-box math" used to describe the curve-clustering algorithm's energy function is insufficiently specific. What is the principled reason for an algorithm to choose a middle ground between 300,000 singleton clusters and one whole-brain cluster?
- The nature of the optimization algorithm for image-based bundle refinement is not specified. How does it relate to established techniques? What is the relationship between image differences and the space of candidate bundle refinements that they suggest? How does it avoid overfitting?
- How will the model accomodate white matter fascicles that project outside of the brain?
- How will the macrostructure modeling results be validated?
- Would chained applications of the macrostructure fitting process converge quickly? If not, why not? If so, what is the nature of the fixed-point solution?
- The stated choice of a Rician distribution for axon diameters seems inappropriate.
- A concrete "toy example" is needed for proper evaluation of this proposal. This could take the form of a computational or physical phantom.