Open Access. Powered by Scholars. Published by Universities.®
Social and Behavioral Sciences Commons™
Open Access. Powered by Scholars. Published by Universities.®
- Discipline
Articles 1 - 3 of 3
Full-Text Articles in Social and Behavioral Sciences
Towards Interpretable Machine Reading Comprehension With Mixed Effects Regression And Exploratory Prompt Analysis, Luca Del Signore
Towards Interpretable Machine Reading Comprehension With Mixed Effects Regression And Exploratory Prompt Analysis, Luca Del Signore
Dissertations, Theses, and Capstone Projects
We investigate the properties of natural language prompts that determine their difficulty in machine reading comprehension tasks. While much work has been done benchmarking language model performance at the task level, there is considerably less literature focused on how individual task items can contribute to interpretable evaluations of natural language understanding. Such work is essential to deepening our understanding of language models and ensuring their responsible use as a key tool in human machine communication. We perform an in depth mixed effects analysis on the behavior of three major generative language models, comparing their performance on a large reading comprehension …
A 3d Characteristics Database Of Land Engraved Areas With Known Subclass, Entni Lin
A 3d Characteristics Database Of Land Engraved Areas With Known Subclass, Entni Lin
Student Theses
Subclass characteristics on bullets may mislead firearm examiners when they rely on traditional 2D images. In order to provide indelible examples for training and help avoid identification errors, 3D topography surface maps and statistical methods of pattern recognition are applied to toolmarks on bullets containing known subclass characteristics. This research was conducted by collecting 3D topography surface map data from land engraved areas of bullets fired through known barrels. This data was processed and used to train the statistical algorithms to predict their origin. The results from the algorithm are compared with the “right answers” (i.e. correct IDs) of the …
Bayesian Approach To Toolmark Analysis, Antonio W. Del Valle
Bayesian Approach To Toolmark Analysis, Antonio W. Del Valle
Student Theses
Statistical analysis of toolmarks using frequentist methods can be problematic for assorted reasons. Thus, in order to analyze toolmarks whilst avoiding these issues, a Bayesian approach is taken. Specifically for this thesis we discuss the computation of a specific Likelihood Ratio for toolmark comparisons. This Bayesian based approach involves using data already at hand in conjunction with a probability model in order to establish an estimate for its “value”, i.e. the “weight of evidence”. Making the calculations to obtain a Likelihood Ratio is very cumbersome and time consuming. Also many commercial software packages hide the process and underlying assumptions that …