skip to main content

Search for: All records

Creators/Authors contains: "Brown, Daniel"

Note: When clicking on a Digital Object Identifier (DOI) number, you will be taken to an external site maintained by the publisher. Some full text articles may not yet be available without a charge during the embargo (administrative interval).
What is a DOI Number?

Some links on this page may take you to non-federal websites. Their policies may differ from this site.

  1. Abstract Background The COVID-19 pandemic presented challenges that disproportionately impacted women. Household roles typically performed by women (such as resource acquisition and caretaking) became more difficult due to financial strain, fear of infection, and limited childcare options among other concerns. This research draws from an on-going study of hot flashes and brown adipose tissue to examine the health-related effects of the COVID-19 pandemic among 162 women aged 45–55 living in western Massachusetts. Methods We compared women who participated in the study pre- and early pandemic with women who participated mid-pandemic and later-pandemic (when vaccines became widely available). We collected self-reportedmore »symptom frequencies (e.g., aches/stiffness in joints, irritability), and assessments of stress, depression, and physical activity through questionnaires as well as measures of adiposity (BMI and percent body fat). Additionally, we asked open-ended questions about how the pandemic influenced women’s health and experience of menopause. Comparisons across pre-/early, mid-, and later pandemic categories were carried out using ANOVA and Chi-square analyses as appropriate. The Levene test for homogeneity of variances was examined prior to each ANOVA. Open-ended questions were analyzed for yes/no responses and general themes. Results Contrary to our hypothesis that women would suffer negative health-related consequences during the COVID-19 pandemic, we found no significant differences in women’s health-related measures or physical activity across the pandemic. However, our analysis of open-ended responses revealed a bi-modal distribution of answers that sheds light on our unexpected findings. While some women reported higher levels of stress and anxiety and lower levels of physical activity, other women reported benefitting from the remote life that the pandemic imposed and described having more time to spend on physical activity or in quality time with their families. Conclusions In this cross-sectional comparison of women during the pre-/early, mid-, and later-pandemic, we found no significant differences across means in multiple health-related variables. However, open-ended questions revealed that while some women suffered health-related effects during the pandemic, others experienced conditions that improved their health and well-being. The differential results of this study highlight a need for more nuanced and intersectional research on risk, vulnerabilities, and coping among mid-life women.« less
    Free, publicly-accessible full text available December 1, 2023
  2. Free, publicly-accessible full text available January 1, 2023
  3. Shared autonomy enables robots to infer user intent and assist in accomplishing it. But when the user wants to do a new task that the robot does not know about, shared autonomy will hinder their performance by attempting to assist them with something that is not their intent. Our key idea is that the robot can detect when its repertoire of intents is insufficient to explain the user’s input, and give them back control. This then enables the robot to observe unhindered task execution, learn the new intent behind it, and add it to this repertoire. We demonstrate with bothmore »a case study and a user study that our proposed method maintains good performance when the human’s intent is in the robot’s repertoire, outperforms prior shared autonomy approaches when it isn’t, and successfully learns new skills, enabling efficient lifelong learning for confidence-based shared autonomy.« less
  4. This presentation compares methods of estimating brown adipose tissue (BAT). As part of an ongoing study of BAT activity in relation to hot flashes, we asked women aged 45-55 to place their hand in cool (17oC) water. We took a thermal image of each woman (Flir camera) before and after the cooling of her hand. To estimate BAT activity, we compared the change in temperature in the supraclavicular area with a control area. Initially, we used a point on the mid-sternum as the control. Because we were concerned that there may be BAT tissue along the sternum, we also triedmore »a control region on the mid-right arm. We used two equations to estimate BAT activity. The first computed the difference in maximum supraclavicular temperature (SCT) minus the difference in the control temperature [(PostMaxSupraclavicular – PreMaxSupraclavicular) - (PostControlMean - PreControlMean)]. Mean BAT estimated from the maximum SCT and arm temperature was higher (0.80, s.d. 0.51, range 0 to 2.10) than from the maximum SCT and sternal temperature (0.63, s.d. 0.45, range 0 to 1.70). There was no relationship between biceps skinfold and arm temperature, or between other anthropometric measures (summed skinfolds, BMI, percent body fat) and estimates of BAT. The sample size is, to date, too small to draw conclusions (n=36), but as the reported severity of hot flashes increased (“none,” “a little,” “somewhat,” “a lot”) the mean BAT estimated with the sternal control also increased (0.49, 0.65, 0.68, 0.74). This was not true when the arm was used as the control. Support: NSF #BCS-1848330« less
  5. As humans interact with autonomous agents to perform increasingly complicated, potentially risky tasks, it is important to be able to efficiently evaluate an agent’s performance and correctness. In this paper we formalize and theoretically analyze the problem of efficient value alignment verification: how to efficiently test whether the behavior of another agent is aligned with a human’s values. The goal is to construct a kind of “driver’s test” that a human can give to any agent which will verify value alignment via a minimal number of queries. We study alignment verification problems with both idealized humans that have an explicitmore »reward function as well as problems where they have implicit values. We analyze verification of exact value alignment for rational agents and propose and analyze heuristic and approximate value alignment verification tests in a wide range of gridworlds and a continuous autonomous driving domain. Finally, we prove that there exist sufficient conditions such that we can verify exact and approximate alignment across an infinite set of test environments via a constant- query-complexity alignment test.« less
  6. Ruiter, Nicole V. ; Byram, Brett C. (Ed.)
  7. The difficulty in specifying rewards for many real world problems has led to an increased focus on learning rewards from human feedback, such as demonstrations. However, there are often many different reward functions that explain the human feedback, leaving agents with uncertainty over what the true reward function is. While most policy optimization approaches handle this uncertainty by optimizing for expected performance, many applications demand risk-averse behavior. We derive a novel policy gradient-style robust optimization approach, PG-BROIL, that optimizes a soft-robust objective that balances expected performance and risk. To the best of our knowledge, PG-BROIL is the first policy optimizationmore »algorithm robust to a distribution of reward hypotheses which can scale to continuous MDPs. Results suggest that PG-BROIL can produce a family of behaviors ranging from risk-neutral to risk-averse and outperforms state-of-the-art imitation learning algorithms when learning from ambiguous demonstrations by hedging against uncertainty, rather than seeking to uniquely identify the demonstrator’s reward function.« less