Imitation learning algorithms provide a simple and straightforward approach for training control policies via supervised learning. By maximizing the likelihood of good actions provided by an expert demonstrator, supervised imitation learning can produce effective policies without the algorithmic complexities and optimization challenges of reinforcement learning, at the cost of requiring an expert demonstrator to provide the demonstrations. In this paper, we ask: can we take insights from imitation learning to design algorithms that can effectively acquire optimal policies from scratch without any expert demonstrations? The key observation that makes this possible is that, in the multi-task setting, trajectories that are generated by a suboptimal policy can still serve as optimal examples for other tasks. In particular, when tasks correspond to different goals, every trajectory is a successful demonstration for the goal state that it actually reaches. We propose a simple algorithm for learning goal-reaching behaviors without any demonstrations, complicated user-provided reward functions, or complex reinforcement learning methods. Our method simply maximizes the likelihood of actions the agent actually took in its own previous rollouts, conditioned on the goal being the state that it actually reached. Although related variants of this approach have been proposed previously in imitation learning with demonstrations, we show how this approach can effectively learn goal-reaching policies from scratch. We present a theoretical result linking self-supervised imitation learning and reinforcement learning, and empirical results showing that it performs competitively with more complex reinforcement learning methods on a range of challenging goal reaching problems, while yielding advantages in terms of stability and use of offline data.
Background: Rheumatic heart disease (RHD) is still a common form of heart disease among children and young adults, especially in developing countries like India. Between 1940 and 1983, the prevalence rate of RHD varied from 1.8 to 11 per 1000 (national average 6 per 1000), while between 1984 and 1995 the rate varied from 1 to 5.4 per 1000 [1]. The study was carried out to assess the accuracy of a medical student's clinical evaluation of valvular heart disease and compare it with that of an echocardiographic evaluation and to determine the sensitivity, specificity and predictive values of clinical examination as compared to echocardiography for the various lesions in RHD patients. Method: 50 children between the ages of 5-16 years, attending the out patient department or admitted in a large teaching hospital, satisfying the criteria of RHD, were included in the study. Each patient underwent detailed clinical evaluation and relevant investigations including echocardiography. Results: Mitral valve was involved most often both by echocardiography and clinically. Isolated aortic valve involvement was rare. The most common lesion was mitral regurgitation (MR) both by auscultation and by echo. Mixed lesions were seen more often than pure lesions. Mitral stenosis (MS) had the highest sensitivity while tricuspid regurgitation (TR) had the highest specificity. MR had the highest positive predictive value and MS the highest negative predictive value. Sensitivity and specificity of aortic regurgitation (AR) was very low when compared to earlier studies. There was a statistically significant difference between echo diagnosis and clinical diagnosis (p < 0.05). Conclusion: It is recommended that echocardiography be done routinely for the diagnosis of cardiac lesions in patients of RHD as clinical examination alone can miss various lesions, especially when the lesions are mild or when multiple lesions are present.
BackgroundRates of care abandonment for retinoblastoma (RB) demonstrate significant geographical variation; however, other variables that place a patient at risk of abandoning care remain unclear. This study aims to identify the risk factors for care abandonment across a multinational set of patients.MethodsA prospective, observational study of 692 patients from 11 RB centres in 10 countries was conducted from 1 January 2019 to 31 December 2019. Multivariate logistic regression was used to identify risk factors associated with higher rates of care abandonment.ResultsLogistic regression showed a higher risk of abandoning care based on country (high-risk countries include Bangladesh (OR=18.1), Pakistan (OR=45.5) and Peru (OR=9.23), p<0.001), female sex (OR=2.39, p=0.013) and advanced clinical stage (OR=4.22, p<0.001). Enucleation as primary treatment was not associated with a higher risk of care abandonment (OR=0.59, p=0.206).ConclusionCountry, advanced disease and female sex were all associated with higher rates of abandonment. In this analysis, enucleation as the primary treatment was not associated with abandonment. Further research investigating cultural barriers can enable the building of targeted retention strategies unique to each country.
This is a case report of Rosai-Dorfman syndrome in a 36-year-old Caucasian male, involving the lacrimal gland, cervical lymph nodes, nasal and sinusal mucosa. It was successfully treated with appropriate immunosuppression. He had initially presented to the ENT surgeon with nasal and sinusal mucosal thickening and bleeding. Cervical lymph node biopsy produced a histological diagnosis compatible with Rosai-Dorfman disease. Later he developed an acute red proptotic eye. He had severe proptosis due to an enlarged lacrimal gland. He refused surgical excision of the tumour, which is suggested if there is an ocular adnexal involvement. Conservative treatment with systemic steroid resulted in the resolution of lacrimal gland swelling, nasal sinusal mucosal thickening and cervical lymphadenopathy. Previous studies have shown that patients with Rosai-Dorfman syndrome are often black males1 and require surgery.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
hi@scite.ai
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
Copyright © 2024 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.