Motivated by practical needs such as large-scale learning, we study the impact of adaptivity constraints to linear contextual bandits, a central problem in online active learning. We consider two popular limited adaptivity models in literature: batch learning and rare policy switches. We show that, when the context vectors are adversarially chosen in d-dimensional linear contextual bandits, the learner needs O(d log d log T ) policy switches to achieve the minimax-optimal regret, and this is optimal up to poly(log d, log log T ) factors; for stochastic context vectors, even in the more restricted batch learning model, only O(log log T ) batches are needed to achieve the optimal regret. Together with the known results in literature, our results present a complete picture about the adaptivity constraints in linear contextual bandits. Along the way, we propose the distributional optimal design, a natural extension of the optimal experiment design, and provide a both statistically and computationally efficient learning algorithm for the problem, which may be of independent interest.Author names are listed in alphabetical order.
This paper studies model-based bandit and reinforcement learning (RL) with nonlinear function approximations. We propose to study convergence to approximate local maxima because we show that global convergence is statistically intractable even for one-layer neural net bandit with a deterministic reward. For both nonlinear bandit and RL, the paper presents a model-based algorithm, Virtual Ascent with Online Model Learner (ViOL), which provably converges to a local maximum with sample complexity that only depends on the sequential Rademacher complexity of the model class. Our results imply novel global or local regret bounds on several concrete settings such as linear bandit with finite or sparse model class, and two-layer neural net bandit. A key algorithmic insight is that optimism may lead to over-exploration even for two-layer neural net model class. On the other hand, for convergence to local maxima, it suffices to maximize the virtual return if the model can also reasonably predict the size of the gradient and Hessian of the real return.
For a specific small-scale region with abundant resources, its copious resources tend to dictate the basic direction of its development, and may subsequently give rise to an industrial structure centered on the advantageous resources. This can give rise to an economic structure that lacks diversity, causing the economic development in the entire local region to fall into the dilemma of the resource curse. The present study conducts a case study from the perspective of small-scale regions, incorporating various types of resource-dependent cities in China, including Qingyang, Jinchang, and Baiyin, to interpret and analyze the resource curse effect by calculating a resource curse coefficient. Moreover, based on the regression model, the present study further discusses the empirical relations associated with the resource curse phenomenon. The results show that, regardless of whether a resource-dependent city is in the early, intermediate or late stage of its resource development, economic development is always plagued by the resource curse effect to a certain degree. Resource development cannot promote economic development, rather, it inhibits economic growth to some extent, resulting in an array of effects that are unfavorable to economic development, rendering the development unsustainable. For different types of resource-dependent cities, resource curse effect exhibits distinct characteristics. The resource curse effect is strongest for a resource-dependent city during an economic recession, is less severe during a development period, and is weakest during maturation. Resource development not only has a direct adverse impact on economic growth, but also often affects economic growth in multiple ways and on various levels through the Dutch disease effect, the crowding out effect, and the institution weakening effect. Until now, most results show that there is no obvious resource curse effect at the national and provincial level. The verification results of small-scale regions show that the resource curse effect at the city level still exists. In addition, the resource curse effect differs across different types of resource-dependent cities.
Yeast are among the most frequent pathogens in humans. The dominant yeast causing human infections belong to the genus Candida and Candida albicans is the most frequently isolated species. However, several non-C. albicans species are becoming increasingly common in patients worldwide. The relationships between yeast in humans and the natural environments remain poorly understood. Furthermore, it is often difficult to identify or exclude the origins of disease-causing yeast from specific environmental reservoirs. In this study, we compared the yeast isolates from tree hollows and from clinics in Hamilton, Ontario, Canada. Our surveys and analyses showed significant differences in yeast species composition, in their temporal dynamics, and in yeast genotypes between isolates from tree hollows and hospitals. Our results are inconsistent with the hypothesis that yeast from trees constitute a significant source of pathogenic yeast in humans in this region. Similarly, the yeast in humans and clinics do not appear to contribute to yeast in tree hollows.
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.