Matching Items (7)
Filtering by
- All Subjects: Bias
- Creators: Liu, Huan
- Creators: Grisso, Thomas
- Resource Type: Text
Description
Catastrophe events occur rather infrequently, but upon their occurrence, can lead to colossal losses for insurance companies. Due to their size and volatility, catastrophe losses are often treated separately from other insurance losses. In fact, many property and casualty insurance companies feature a department or team which focuses solely on modeling catastrophes. Setting reserves for catastrophe losses is difficult due to their unpredictable and often long-tailed nature. Determining loss development factors (LDFs) to estimate the ultimate loss amounts for catastrophe events is one method for setting reserves. In an attempt to aid Company XYZ set more accurate reserves, the research conducted focuses on estimating LDFs for catastrophes which have already occurred and have been settled. Furthermore, the research describes the process used to build a linear model in R to estimate LDFs for Company XYZ's closed catastrophe claims from 2001 \u2014 2016. This linear model was used to predict a catastrophe's LDFs based on the age in weeks of the catastrophe during the first year. Back testing was also performed, as was the comparison between the estimated ultimate losses and actual losses. Future research consideration was proposed.
ContributorsSwoverland, Robert Bo (Author) / Milovanovic, Jelena (Thesis director) / Zicarelli, John (Committee member) / School of Mathematical and Statistical Sciences (Contributor) / Barrett, The Honors College (Contributor)
Created2018-05
Description
Exabytes of data are created online every day. This deluge of data is no more apparent than it is on social media. Naturally, finding ways to leverage this unprecedented source of human information is an active area of research. Social media platforms have become laboratories for conducting experiments about people at scales thought unimaginable only a few years ago.
Researchers and practitioners use social media to extract actionable patterns such as where aid should be distributed in a crisis. However, the validity of these patterns relies on having a representative dataset. As this dissertation shows, the data collected from social media is seldom representative of the activity of the site itself, and less so of human activity. This means that the results of many studies are limited by the quality of data they collect.
The finding that social media data is biased inspires the main challenge addressed by this thesis. I introduce three sets of methodologies to correct for bias. First, I design methods to deal with data collection bias. I offer a methodology which can find bias within a social media dataset. This methodology works by comparing the collected data with other sources to find bias in a stream. The dissertation also outlines a data collection strategy which minimizes the amount of bias that will appear in a given dataset. It introduces a crawling strategy which mitigates the amount of bias in the resulting dataset. Second, I introduce a methodology to identify bots and shills within a social media dataset. This directly addresses the concern that the users of a social media site are not representative. Applying these methodologies allows the population under study on a social media site to better match that of the real world. Finally, the dissertation discusses perceptual biases, explains how they affect analysis, and introduces computational approaches to mitigate them.
The results of the dissertation allow for the discovery and removal of different levels of bias within a social media dataset. This has important implications for social media mining, namely that the behavioral patterns and insights extracted from social media will be more representative of the populations under study.
Researchers and practitioners use social media to extract actionable patterns such as where aid should be distributed in a crisis. However, the validity of these patterns relies on having a representative dataset. As this dissertation shows, the data collected from social media is seldom representative of the activity of the site itself, and less so of human activity. This means that the results of many studies are limited by the quality of data they collect.
The finding that social media data is biased inspires the main challenge addressed by this thesis. I introduce three sets of methodologies to correct for bias. First, I design methods to deal with data collection bias. I offer a methodology which can find bias within a social media dataset. This methodology works by comparing the collected data with other sources to find bias in a stream. The dissertation also outlines a data collection strategy which minimizes the amount of bias that will appear in a given dataset. It introduces a crawling strategy which mitigates the amount of bias in the resulting dataset. Second, I introduce a methodology to identify bots and shills within a social media dataset. This directly addresses the concern that the users of a social media site are not representative. Applying these methodologies allows the population under study on a social media site to better match that of the real world. Finally, the dissertation discusses perceptual biases, explains how they affect analysis, and introduces computational approaches to mitigate them.
The results of the dissertation allow for the discovery and removal of different levels of bias within a social media dataset. This has important implications for social media mining, namely that the behavioral patterns and insights extracted from social media will be more representative of the populations under study.
ContributorsMorstatter, Fred (Author) / Liu, Huan (Thesis advisor) / Kambhampati, Subbarao (Committee member) / Maciejewski, Ross (Committee member) / Carley, Kathleen M. (Committee member) / Arizona State University (Publisher)
Created2017
Description
We integrate multiple domains of psychological science to identify, better understand, and manage the effects of subtle but powerful biases in forensic mental health assessment. This topic is ripe for discussion, as research evidence that challenges our objectivity and credibility garners increased attention both within and outside of psychology. We begin by defining bias and provide rich examples from the judgment and decision making literature as they might apply to forensic assessment tasks. The cognitive biases we review can help us explain common problems in interpretation and judgment that confront forensic examiners. This leads us to ask (and attempt to answer) how we might use what we know about bias in forensic clinicians’ judgment to reduce its negative effects.
ContributorsNeal, Tess M.S. (Author) / Grisso, Thomas (Author)
Created2014-05
Description
We conducted an international survey in which forensic examiners who were members of professional associations described their two most recent forensic evaluations (N=434 experts, 868 cases), focusing on the use of structured assessment tools to aid expert judgment. This study describes:
1. The relative frequency of various forensic referrals.
2. What tools are used globally.
3. Frequency and type of structured tools used.
4. Practitioners’ rationales for using/not using tools.
We provide general descriptive information for various referrals. We found most evaluations used tools (74.2%) and used several (on average 4). We noted the extreme variety in tools used (286 different tools). We discuss the implications of these findings and provide suggestions for improving the reliability and validity of forensic expert judgment methods. We conclude with a call for an assessment approach that seeks structured decision methods to advance greater efficiency in the use and integration of case-relevant information.
ContributorsNeal, Tess M.S. (Author) / Grisso, Thomas (Author)
Created2014-09-25
Description
Generative Adversarial Networks are designed, in theory, to replicate the distribution of the data they are trained on. With real-world limitations, such as finite network capacity and training set size, they inevitably suffer a yet unavoidable technical failure: mode collapse. GAN-generated data is not nearly as diverse as the real-world data the network is trained on; this work shows that this effect is especially drastic when the training data is highly non-uniform. Specifically, GANs learn to exacerbate the social biases which exist in the training set along sensitive axes such as gender and race. In an age where many datasets are curated from web and social media data (which are almost never balanced), this has dangerous implications for downstream tasks using GAN-generated synthetic data, such as data augmentation for classification. This thesis presents an empirical demonstration of this phenomenon and illustrates its real-world ramifications. It starts by showing that when asked to sample images from an illustrative dataset of engineering faculty headshots from 47 U.S. universities, unfortunately skewed toward white males, a DCGAN’s generator “imagines” faces with light skin colors and masculine features. In addition, this work verifies that the generated distribution diverges more from the real-world distribution when the training data is non-uniform than when it is uniform. This work also shows that a conditional variant of GAN is not immune to exacerbating sensitive social biases. Finally, this work contributes a preliminary case study on Snapchat’s explosively popular GAN-enabled “My Twin” selfie lens, which consistently lightens the skin tone for women of color in an attempt to make faces more feminine. The results and discussion of the study are meant to caution machine learning practitioners who may unsuspectingly increase the biases in their applications.
ContributorsJain, Niharika (Author) / Kambhampati, Subbarao (Thesis advisor) / Liu, Huan (Committee member) / Manikonda, Lydia (Committee member) / Arizona State University (Publisher)
Created2020
Description
Machine learning models can pick up biases and spurious correlations from training data and projects and amplify these biases during inference, thus posing significant challenges in real-world settings. One approach to mitigating this is a class of methods that can identify filter out bias-inducing samples from the training datasets to force models to avoid being exposed to biases. However, the filtering leads to a considerable wastage of resources as most of the dataset created is discarded as biased. This work deals with avoiding the wastage of resources by identifying and quantifying the biases. I further elaborate on the implications of dataset filtering on robustness (to adversarial attacks) and generalization (to out-of-distribution samples). The findings suggest that while dataset filtering does help to improve OOD(Out-Of-Distribution) generalization, it has a significant negative impact on robustness to adversarial attacks. It also shows that transforming bias-inducing samples into adversarial samples (instead of eliminating them from the dataset) can significantly boost robustness without sacrificing generalization.
ContributorsSachdeva, Bhavdeep Singh (Author) / Baral, Chitta (Thesis advisor) / Liu, Huan (Committee member) / Yang, Yezhou (Committee member) / Arizona State University (Publisher)
Created2021
Description
AARP estimates that 90% of seniors wish to remain in their homes during retirement. Seniors need assistance as they age, historically they have received assistance from either family members, nursing homes, or Continuing Care Retirement Communities. For seniors not wanting any of these options, there has been very few alternatives. Now, the emergence of the continuing care at home program is providing hope for a different method of elder care moving forward. CCaH programs offer services such as: skilled nursing care, care coordination, emergency response systems, aid with personal and health care, and transportation. Such services allow seniors to continue to live in their own home with assistance as their health deteriorates over time. Currently, only 30 CCaH programs exist. With the growth of the elderly population in the coming years, this model seems poised for growth.
ContributorsSturm, Brendan (Author) / Milovanovic, Jelena (Thesis director) / Hassett, Matthew (Committee member) / School of Mathematical and Statistical Sciences (Contributor) / Economics Program in CLAS (Contributor) / Barrett, The Honors College (Contributor)
Created2019-05