School of Electrical Engineering and Computer Science
http://hdl.handle.net/1957/7302
2014-12-19T09:04:06Z
2014-12-19T09:04:06Z
On edge disjoint Hamiltonian cycles in torus and Gaussian networks
Alazemi, Fawaz M.
http://hdl.handle.net/1957/54654
2014-12-17T16:52:48Z
2014-12-04T00:00:00Z
On edge disjoint Hamiltonian cycles in torus and Gaussian networks
Alazemi, Fawaz M.
Many algorithms in parallel systems can be easily solved if we can generate a Hamiltonian cycle on the underly network. Finding Hamiltonian cycle is a well known NP-complete problem. For specific instances of regular graphs, such as Torus and Gaussian network, one can easily find Hamiltonian cycles. In this thesis, we present a recurrence function that can generate 2[superscript r] ≥ 1 independent Gray codes from Z[supserscript n][subscript k] where 2[superscript r] ≤ n < 2[superscript r+1]. Such independent Gray codes corresponds to edge disjoint Hamiltonian cycles on the Torus graph T[supserscript n][subscript k] and multidimensional Gaussian network Gα[superscript ⌊n/2⌋], for 1 ≤ 2[superscript r] ≤ n < 2[superscript r+1].
Graduation date: 2015; Access restricted to the OSU Community, at author's request, from Dec. 16, 2014 - Dec. 16, 2015
2014-12-04T00:00:00Z
Dynamic biasing for ring amplification
Farahbakhshian, Farshad
http://hdl.handle.net/1957/54635
2014-12-15T17:16:44Z
2014-12-10T00:00:00Z
Dynamic biasing for ring amplification
Farahbakhshian, Farshad
New amplifier architectures are presented using non-traditional methods of biasing. Time-based dynamic biasing and signal-based dynamic biasing are discussed in the context of new architectures. This includes a new form of ring amplification with a dynamic deadzone, allowing for a structure whose coarse path does not consume static power.
Graduation Date: 2015; Access restricted to the OSU Community, at author's request, from Dec. 12, 2014 - June 12, 2015
2014-12-10T00:00:00Z
Personalizing machine learning systems with explanatory debugging
Kulesza, Todd
http://hdl.handle.net/1957/54622
2014-12-10T19:14:03Z
2014-12-01T00:00:00Z
Personalizing machine learning systems with explanatory debugging
Kulesza, Todd
How can end users efficiently influence the predictions that machine learning systems make on their behalf? Traditional systems rely on users to provide examples of how they want the learning system to behave, but this is not always practical for the user, nor efficient for the learning system. This dissertation explores a different personalization approach: a two-way cycle of explanations, in which the learning system explains the reasons for its predictions to the end user, who can then explain any necessary corrections back to the system. In formative work, we study the feasibility of explaining a machine learning system's reasoning to end users and whether this might help users explain corrections back to the learning system. We then conduct a detailed study of how learning systems should explain their reasoning to end users. We use the results of this formative work to inform Explanatory Debugging, our explanation-centric approach for personalizing machine learning systems, and present an example of how this novel approach can be instantiated in a text classification system. Finally, we evaluate the effectiveness of Explanatory Debugging versus a traditional learning system, finding that explanations of the learning system's reasoning improved study participants' understanding by over 50% (compared with participants who used the traditional system) and participants' corrections to this reasoning were up to twice as efficient as providing examples to the learning system.
Graduation date: 2015
2014-12-01T00:00:00Z
Improved Least Median of Squares Localization for Non-Line-of-Sight Mitigation
Qiao, Tianzhu
Liu, Huaping
http://hdl.handle.net/1957/54534
2014-12-03T18:56:35Z
2014-08-01T00:00:00Z
Improved Least Median of Squares Localization for Non-Line-of-Sight Mitigation
Qiao, Tianzhu; Liu, Huaping
Non-line-of-sight (NLOS) propagation is one of the
major problems that cause errors in localization systems. There
exist many algorithms to reduce NLOS caused errors. Most of
these algorithms require a priori information about the NLOS
links, which limits their application. Least median of squares
(LMedS) method is an exception and is thus attractive for its
simplicity. LMedS selects only a subset of the anchors with the
smallest residue for location calculation, but it tends to discard
more range measurements than necessary. Its performance can be
significantly improved, especially when the percentage of NLOS
links is not high. In this paper, we present an improved LMedS
method, which maximizes the subset of reliable anchors to be
used. We show via simulation that compared with the existing
LMedS method, the proposed algorithm has a significantly higher
localization accuracy and is more stable.
This is an author's peer-reviewed final manuscript, as accepted by the publisher. The published article is copyrighted by IEEE-Institute of Electrical and Electronics Engineers and can be found at: http://ieeexplore.ieee.org/xpl/RecentIssue.jsp?punumber=4234; ©2014 IEEE. Personal use of this material is permitted. Permission from IEEE must be obtained for all other users, including reprinting/republishing this material for advertising or promotional purposes, creating new collective works for resale or redistribution to servers or lists, or reuse of any copyrighted components of this work in other works.
2014-08-01T00:00:00Z