134 research outputs found
Objective metrics for human-subjects evaluation in explainable reinforcement learning
No full textExplanation is a fundamentally human process. Understanding the goal and audience of the explanation is vital, yet existing work on explainable reinforcement learning (XRL) routinely does not consult humans in their evaluations. Even when they do, they routinely resort to subjective metrics, such as confidence or understanding, that can only inform researchers of users' opinions, not their practical effectiveness for a given problem. This paper calls on researchers to use objective human metrics for explanation evaluations based on observable and actionable behaviour to build more reproducible, comparable, and epistemically grounded research. To this end, we curate, describe, and compare several objective evaluation methodologies for applying explanations to debugging agent behaviour and supporting human-agent teaming, illustrating our proposed methods using a novel grid-based environment. We discuss how subjective and objective metrics complement each other to provide holistic validation and how future work needs to utilise standardised benchmarks for testing to enable greater comparisons between research
Explainable Autonomous Vehicle Intelligence
No full text## This item has been replaced by the one which can be found at https://doi.org/10.7488/ds/7867 ## HEADD is a dataset of natural language explanations elicited from online participants via Prolific with corresponding annotations for each explanation similarly given by online (but different) participants. The data is about driving scenarios in which the behaviour and driving decisions of a single blue agent need to be explained, while the scenarios contain various other agents and environmental elements that influence the behaviour of the blue agent. The dataset contains 14 unique scenarios with qualitatively distinct and interesting driving scenarios including simulated video recordings, ASAM OpenDrive maps, and ASAM OpenScenario descriptions. In addition, HEADD includes 1308 explanations in natural language with 4 explanatory modes (descriptive, teleological, mechanistic, counterfactual) from 54 participants in each of the 14 scenarios, of which 947 non-descriptive explanations are annotated with 5 unique annotations regarding the causal content and trustworthiness of the explanations under the various circumstances in the scenarios.Please refer to the following files to get a full picture of how the data is structured:
(A) README.md: Contains a high-level overview of the dataset, its contents, a brief description of the collection process, and the exact questions asked to participants;
(B) scenarios_metadata.json: A semi-structured file in the JSON-format containing the description and metadata of the 14 scenarios;
(C) explanations_metadat.json: A semi-structured file in the JSON-format containing the description and metadata of the explanations elicited from participants for each of the 14 scenarios;
(D) annotations_metadata.json: A semi-structured file in the JSON-format containing the description and metadata of the annotations for each non-descriptive explanation
Balint Vaszonyi Collection
No full textBalint Vazsonyi (March 7, 1936-January 2003) was a Hungarian-American pianist, perhaps best known for playing all thirty-two chronological cycles of Beethovens sonatas. Vazsonyi also was an author who wrote extensively on political science in the Washington Times and other media sources. The collection consists of concert programs, reviews, correspondence, newspaper and magazine articles, recordings, lectures, scrapbooks, brochures, manuscripts, books, and videos related to Vazsonyis performing and teaching careers, the ensembles he was involved in, events in his life, his political involvement, and his relationships with several people and organizations, especially his teacher, Ernst von Dohnanyi, and his management, Kazuko Hillyer International, Inc
Balint seminars: The transatlantic experience through videoconference
No full textIntroduction: The Balint seminar is used in many family medicine residencies to improve and strengthen the patient-doctor relationship: to make better doctors. Given the lack of Balint leaders in developing countries, the family medicine department at the American University of Beirut (AUB) decided to collaborate with the Medical University of South Carolina (MUSC)-with more than 30 years of experience-to start Balint seminars through videoconferencing. Objective: Evaluate the feasibility and effectiveness of Balint seminars conducted through videoconference. Methods: A qualitative research using focus group and leaders' feedback to evaluate feasibility of delivery of Balint seminars through videoconference. A Polycom videoconference was set up between residents at AUB and two credentialed Balint leaders at MUSC. The videoconference was composed of two parts: (i) MUSC faculty facilitating Balint seminars; and (ii) MUSC and AUB faculty debriefing following each Balint session. Results: Twenty-six videoconferences were conducted from 15 February 2013 to 31 March 2014. Four themes emerged: technology and connectivity issues, administrative issues, coordination among different time zones and cultural/contextual issues. The videoconferencing with family medicine residents at AUB seemed quite natural and very familiar to the Balint leaders at MUSC. The seminars encouraged the residents to see things from the patients' perspective, inspiring new thoughts and ideas on how to deal with troubling patients. Conclusion: Videoconference Balint seminars offer a promising way to extend the activity to health care providers in other disciplines, states and countries. Moreover, this format has the potential to increase the number of trained Balint leaders. © The Author 2014. Published by Oxford University Press. All rights reserved
Balint groups: a doctor-student mutual investment company
No full textBalint Groups were traditionally established as reflective groups by psychoanalyst Michael and Enid Balint for general practitioners to reflect on the doctor-patient relationship. Balint described components of this relationship between doctor and patient including the collusion of anonymity, the doctor as a drug and the mutual investment company. This paper discusses 2 case examples from the perspective of a junior doctor facilitating medical student Balint groups and from the junior doctor participating in a peer group. Comparisons between the doctor and student emotional expression, empathic ability and apparent preconceived ideas of the “doctor role” are discussed, with reflection on potential origins and contributing factors to such internalised views and responses. The author explores potential professional benefits of medical student Balint groups facilitated by junior doctors in influencing empathic response and internalised personas, as discussed through the eyes of Balint’s components of the interpersonal doctor-patient relationship
Demystifying Balint culture and its impact: An autoethnographic analysis.
No full textBalint groups are a structured group process designed for physicians and other helping professionals to provide feedback to each other concerning troublesome patient encounters. The process of Balint groups has survived and even thrived for more than a half century since their inception, suggesting their inherent value. However, measuring the impact in order to communicate the nature of that value has been elusive. A recent literature review highlighted the equivocal nature of research results and suggested that well-designed qualitative research may be more effective in selecting target measures which could result in better evidence for this group process. This paper describes such an effort using a qualitative analysis of the author\u27s own reflections about significant moments and learnings while participating in Balint groups
¿Qué fue de los tradicionales grupos Balint? En defensa de la perspectiva Balint y una metodología adaptada en los espacios de coordinación con Atención Primaria
No full textbased on a two-year experience in coordination spaces, in close work with primary care teams, plus comments and ideas collected from several professionals -who already took part into Balint groups-, and a review on related bibliography, the following work attempts to analyze all determining factors that caused the Balint Groups to fall into desuse in Spain. This statement takes into consideration the severe changes happening nowadays, both socialwise and in public health specifically; and about the inner methodology used by these groups in the past as possible factors for such disappearance.
The author positions himself on adaptations, suggesting a review of the Balint method itself, yet defending it�s perspective as a basic ingredient on professional training, empathizing
it�s importance on all current meeting points between both Mental Health and Primary Care consultancy.a partir de la experiencia de dos años, en espacios de coordinación, junto a equipos de atención primaria, comentarios e ideas de diferentes profesionales que participaron en grupos balint y una revisión bibliográfica, el autor del presente trabajo trata de analizar los posibles determinantes en la desaparición de los grupos balint en España. Se reflexiona en torno a los graves cambios sanitarios actuales, sociales y a la propia metodología empleada en los grupos de aquel entonces, como posibles factores implicados.
El autor apuesta por adaptaciones, sugiriendo modificaciones de dicha metodología, y defiende la perspectiva balint como ingrediente básico en la formación de los profesionales
y esencia de los actuales espacios de encuentro o consultoría entre atención primaria y salud mental
Towards Trustworthy Autonomous Systems via Conversations and Explanations
No full textAutonomous systems fulfil an increasingly important role in our societies, however, AI-powered systems have seen less success over the years, as they are expected to tackle a range of social, legal, or technological challenges and modern neural network-based AI systems cannot yet provide guarantees to many of these challenges. Particularly important is that these systems are black box decision makers, eroding human oversight, contestation, and agency. To address this particular concern, my thesis focuses on integrating social explainable AI with cognitive methods and natural language processing to shed light on the internal processes of autonomous systems in a way accessible to lay users. I propose a causal explanation generation model for decision-making called CEMA based on counterfactual simulations in multi-agent systems. I also plan to integrate CEMA with a broader natural language processing pipeline to support targeted and personalised explanations that address people's cognitive biases. I hope that my research will have a positive impact on the public acceptance of autonomous agents by building towards more trustworthy AI
Doctors at times of national instability: What Balint seminars reveal
No full textBackground: It is not known in what ways is the doctor whose practice is secure in a clinic challenged to maintain a therapeutic doctor–patient relationship when confronting a flood of immigrants within a country that is politically volatile, internally fractionalized, and surrounded by sporadic military incursions? Methods: During Balint seminars, a family medicine resident presents a troubling case which all group members reflect upon from the perspective of the doctor, the patient, and their relationship. Balint leaders later debrief and review the work of the group. Lebanon has passed through many political, social, and religious conflicts and was affected by the onset of the Syrian Civil War in 2010. The Balint leaders had begun to see in resident case presentations reflections of war’s disruption of the doctor–patient relationship. Two Balint leaders reviewed a log of all the cases between 2013 and 2016. Results: In our observations, the discussion of the presented cases mirrored the cultural, social, religious, and political context of the country. First, the political situation was reflected in the dynamics of the group: agitation, conflicts, hopelessness, and a search for norms. Second, the residents subconsciously chose words in their discussion that reflect the country’s situation. Third, the presented case was stirred by a tragic war-related event. Conclusions: The social/political/religious context in which the physician is practicing distracts the doctor from fulfilling his/her professional role. Balint seminars are an example of direct, experiential learning that provide an excellent opportunity for the special training of primary care physicians who deal with refugees and citizens to self-reflect on war’s impact on them and their profession. © The Author(s) 2018
Engaging indigenous Maori and inward migrating Asian professionals into a Pakeha (White European)-dominated Balint community in New Zealand
No full text This inquiry began with two questions: How can the established predominately Pakeha/Caucasian (White European) Balint community in New Zealand more successfully engage both indigenous populations of both Maori and Pacifica origin into Balint work? And what is the existing Balint community doing to address the lack of Asian members of the Balint community in New Zealand, at a time when Asian health professionals are being recruited into the health sector at an increasingly high rate in comparison to White European entrants to the profession? These questions, and their preliminary answers presented here, invite the reader to reflect on both the challenges and opportunities in reaching out to groups different from our own. The author hopes readers may begin to see what can be done to allow new entrants to benefit from all that participation in Balint work offers while not losing sight of the uniqueness which each person can bring. It is hoped that sharing such questions and their subsequent explorations will help Balint leaders feel more confident in reaching out to a wider ethic and cultural mix within their local populations and encouraging them to enter the exciting world of the Balint group. </jats:p
- …
