49755 research outputs found
Sort by
补贴战与现代产业政策
No full textTraditional economic nationalism, techno-nationalism and national security concerns have converged into a modern industrial policy, which is being pursued by multiple actors, including the US, China, India, Republic of Korea, and the EU. Driven by new subsidy wars, these policies pose a tremendous challenge for global economic relations
东盟在制定FDI规则方面不断扩大的作用
No full textThis Perspective discusses the fiscal and economic impacts of the global minimum tax; currently being implemented by many countries around the world. The global minimum tax is expected to raise tax revenues, reduce profit-shifting, and allow jurisdictions to strike a better balance between supporting investment and mobilising domestic revenues
Essays in Behavioral and Experimental Economics
No full textIn this dissertation, I revisit two classic phenomena in behavioral economics: choice overload and the gambler's fallacy. Despite their claims to fame, our understanding of these two phenomena are insufficient, especially in comparison to the attention that they have received. All chapters in this dissertation bring two methodologies to remedy this situation, namely microeconomic theory and laboratory experiments. Competing theoretical explanations for each of the two phenomena are formally presented and analyzed, and an experimental design is formulated in order to produce data that is capable of differentiating between such competing explanations.
Chapter 1 presents and experimentally tests a collection of search theoretic explanations for `choice overload', the phenomena by which a default alternative is selected more often in larger choice sets. A standard search model, with constant search costs and a known distribution of item quality, cannot give rise to choice overload. If one instead assumes that either (i) the Decision Maker (DM) must learn the quality distribution (ii) search costs are increasing or (iii) the DM decides the search strategy in advance, then choice overload can occur. Unlike existing models, our approach does not require ad hoc psychological costs (decision avoidance), or for the DM to assume the choice set was selected by a profit maximizing firm (contextual inference). Data from our laboratory experiments are consistent with choice overload caused by search with learning and increasing costs, and cannot be explained by decision avoidance or contextual inference. A classic explanation for the gambler's fallacy is that subjects believe that sequences of tosses from a fair coin should be representative of the randomness of the uniform distribution, and so should not have observable patterns.
In chapter 2, I introduce an information-theoretic formalization of this representativeness heuristic in terms of complexity and contrast it to the existing recency-weighted reversal model by Rabin and Vayanos. In order to test between these explanations, I collect rich choice and belief data from subjects predicting the next item from fully randomized sequences of binary outcomes, allowing me to take the analysis to the level of individual sequences. The basic results confirm the existence of the gambler’s fallacy in the aggregate. However, there is also significant heterogeneity among subjects. I identify four types, depending on whether they report correct beliefs or incorrect beliefs that go in the gambler's fallacy direction, its opposite or a mix of both. Taking this heterogeneity into account, both models perform well when looking at an aggregate level, but a closer look at individual sequences reveals violations of the representativeness model which lead to a superior performance of the recency-weighted reversal model. The main component of this superior performance comes from the recency bias that subjects exhibit, which is not accounted for in the representativeness model.
Chapter 3 continues the work in chapter 2 by extending the framework for predictions about the next two outcomes instead of just one. The models that were presented in chapter 2 are extended to this new scenario and a novel experiment is presented, in which multidimensional belief and choice data is collected. The results from this experiment further reinforce those from chapter 2. Particularly striking is the finding that subjects suffering from the classic definition of gambler's fallacy believe, in the case of alternating sequences, the continuation of the alternating pattern to be the most likely outcome
Effects of Fully Immersive Virtual Reality on Nurses’ Knowledge, Decision-Making, Self-Efficacy, and Engagement
No full textNewly graduated nurses often face challenges in managing medical crises, including emergencies such as diabetes-related ketoacidosis (DKA). Addressing these challenges during onboarding is essential to prepare them for providing effective and timely care in acute settings. V bAs modern learners, many from Millennial and Gen Z cohorts benefit from dynamic, technology-driven educational methods. Fully immersive virtual reality (FIVR) offers an innovative approach to creating engaging learning environments, yet its use in nurse onboarding remains limited.
This study examined the impact of FIVR on knowledge, decision-making, self-efficacy, and engagement compared to traditional onboarding methods. A quasi-experimental pretest/posttest design was used with newly graduated nurses (N = 55) from two hospitals in a health care system. Participants were assigned, over alternating months, to an intervention group using FIVR or a comparison group using the traditional onboarding lecture format. The study employed three instruments: the RN Knowledge and Decision-Making Processes in the Management of DKA modified with permission for this study into the TKDM Short Form, the Learning Self-Efficacy Scale for Clinical Skills, and the Web-based Learning Tools Evaluation Scale.
Knowledge and decision-making were assessed during the pretest and posttest, while self-efficacy and engagement were measured posttest only. The study included validating the TKDM Short Form and obtaining psychometric properties. Findings from a mixed ANOVA showed that the knowledge scores of both groups increased significantly from the pretest to the posttest, but the degree of change was similar. No significant changes were observed in decision-making scores; however, the modified instrument failed to demonstrate adequate reliability. While self-efficacy scores were also similar between groups, engagement was significantly higher in the FIVR group, indicating that this method fosters greater learner involvement during onboarding.
These findings suggest that while FIVR enhances engagement, it does not significantly improve knowledge, decision-making, or self-efficacy compared to traditional methods in onboarding for DKA management. The low reliability of the TKDM Short Form was a major limitation of the study. The findings underscore the need for further research with larger, more diverse samples and varied clinical scenarios with validated instruments to determine whether FIVR can effectively improve knowledge and decision-making in nursing practice. FIVR represents a promising and engaging tool for onboarding newly graduated nurses, particularly in fostering active participation. However, its effectiveness in advancing clinical knowledge and decision-making remains inconclusive
In Wonder: Writing as Discernment Between the Call, the Self(s), and (Generative Artificial) Technology
No full textThis dissertation is a call—and a pause. At a moment when generative artificial intelligence is being rapidly integrated into writing classrooms, I ask not simply what we should do in response, but how we might discern—how we might pause, wonder, and reflect on what this technology is asking of us, and what we are becoming in relationship with it. Grounded in Ignatian spirituality and shaped by the epistemological and poetic practices of writing, this work proposes that discernment is a necessary and ongoing process for navigating human-technology relationships.
Through a recursive structure that moves between wonder, theory, practice, and becoming, I explore how writing can serve as a meditative, ethical, and spiritual practice—one that helps students and educators alike to respond to the call of our moment. Drawing from Heidegger, I trace how writing offers a way of attending Dasein, a call of Being-there in the world. Each chapter dwells in relational tensions between the call, the self(s), and technology, culminating in a proposed framework for discernment that invites teachers and students to reflect on their own engagements with generative AI through attention, intention, and relationality.
Alongside this framework, I offer implementable writing practices and exercises that can be adapted in both secondary and post-secondary educational settings. These are not prescriptive methods, but invitations—ways of helping students and educators pause, ask questions, and dwell in the uncertainty of meaningful learning
Predictive Privacy: A Framework for Quantifying Harm
No full textData-driven, omnipresent apps that track all aspects of our day-by-day lives send our information to corporations, third-party data brokers, and, often, the government. It feels like this much invasion of our privacy should be illegal, but given the lack of general privacy laws in the US, it is not. That said, the feeling of discomfort that comes along with being constantly watched does not go away; in fact, when it comes to consumer data usage and the lack of protections provided by privacy laws, the word “creepy” is, as Tene and Polonetsky have noted, the best one that can be used to describe the overall feeling of unease. This is because there is no uniform understanding of privacy harm, which has consequences. For example, the lack of tangible and concrete harm can cause courts and regulatory agencies to do nothing. A company is more often than not held to its privacy policy the way it would be to a legal contract, as established by Solove et al., due to failures to establish promissory estoppel and/or damages. In one case, Spokeo v. Robins, the Supreme Court held that Robins’ complaint of harm lacked “concreteness.” In TransUnion LLC v. Ramirez, the Court only awarded damages to those who were able to prove that harmful information about them had been propagated, rather than to those who had only experienced “an injury in law,” as discussed by Husi and Robbennolt. But all privacy harm is important. For example, the Federal Trade Commission Act gives the FTC jurisdiction only over behavior that is “likely to cause substantial injury to consumers.” But what is “injury,” let alone “substantial injury”? Citron has noted that “[f]or most courts, privacy and data security harms are too speculative and hypothetical.” Solove and Hartzog, quoting FTC documents, note that “[m]onetary, health, and safety risks are common injuries considered ‘substantial,’ but trivial, speculative, emotional, and ‘other more subjective types of harm’ are usually not considered substantial for unfairness purposes.” In addition to the issue of defining privacy harm in general, today we must confront the rise of machine learning, which can identify—or purport to identify—information about an individual that has not been directly observed or collected. For example, in Sterling v. Borough of Minersville, when a young man committed suicide under the threat of being outed as gay, it was held that people have the right to privacy when it comes to sexuality. But with today’s predictive technology, researchers at MIT have shown that sexual identity can be deduced from Facebook “friend” patterns. Any theory of harm must account for the social and psychological impacts that can occur as well. To address these issues, we have devised a scheme called Predictive Privacy, an experimental technique to create a standardized system for quantifying various degrees of loss and harm from disclosure of private information. The goal is to provide regulators and courts an objective basis to ascertain if there is, in fact, actual injury. By providing an objective measure of privacy injury, our approach provides regulators and the courts with a concrete basis for adjudicating claims, with the ultimate result of more informed and effective privacy protection policy. We built a database using existing synthetic data from the Pew Research Center in combination with differential privacy techniques to generate several sensitive columns of synthetic data per person with an accurate statistical distribution similar to that seen today in the American population. We then use a machine learning algorithm to cluster entries in our database. We used workers recruited via Prolific.com to score the harm from disclosure of information in randomly selected members of each cluster. We asked the scorers to rate the harm from various scenarios in two different experiments. Part I involved participants scoring harm for data that was considered 100% accurate, and Part II had participants scoring data that varied in accuracy between 75% and 50% confidence. The goal was to see the impact of data accuracy on harm perception and if perceptions changed in different scenarios. The dataset has various synthetic people with different categories of sensitive attributes, so with these questions, we hope to gain relatively accurate insight into how people view harm and risk for different groups. Finally, we use the worker responses to train a supervised ML algorithm to predict the harm score per person in the different scenarios aforementioned. The net result will be to make informational privacy harm as concrete as, say, monetary harm
Optimizing Distributed Transactions via Modern AI, Storage and Networking Technologies
No full textDistributed ACID transactions, once declared as hopelessly unscalable and unnecessary, are back by popular developer demand. Unfortunately, designing on-disk database systems that support distributed transactions remains quite challenging. Designers of such systems typically face a difficult choice. They can either use an expensive commit protocol like two-phase commit (2PC) to guarantee atomicity, and suffer from slow distributed transactions, or forgo 2PC, which leads to weaker semantics, limitations to the programming model, or constrained scalability, making the system less general. Therefore, there is a trade-off between speed and generality in distributed transactions database systems.
This thesis posits that it is time to revisit that trade-off. We argue that modern developments in AI, storage, and networking unlock the potential of database system designs that offer fast and general distributed transactions. We tackle the problem from different angles. First, we leverage low-latency networking, storage hardware and system software to directly reduce the cost of the commit protocol. Second, when such a low latency stack is unavailable, we design algorithms to mask latency and maintain high throughput in the face of slow 2PC. Finally, we explore applying Reinforcement Learning to the sharding problem so that we reduce the number of distributed transactions the system has to execute without sacrificing other important objectives
Representation Learning for CRISPR-Cas13d Efficacy and Single-Cell RNA Sequencing Data
No full textThis thesis develops multiple novel methods spanning computational genomics and machine learning. CRISPR-Cas13d is a programmable RNA-targeting system that can knockdown specific RNA transcripts. Our method, TIGER (Wessels et al., 2023), is the first tool to model CRISPR-Cas13d efficacy as a function of both the target RNA and the guide RNA (gRNA) sequences; doing so enables biologists using our model to design the most effective gRNA for a target transcript, check a gRNA library for any unintended off-target effects, and engineer mismatches between a gRNA and its target to titrate CRISPR-Cas13d efficacy.
We leverage TIGER to study CRISPR-Cas13d binding affinity at junction splice sites (Megan D Schertzer et al., 2023) finding CRISPR-Cas13d can uniquely target 89% of human isoforms with high efficacy. Thereafter, we develop two methods (Stirn & Knowles, 2020; Stirn et al., 2023) for generating well-calibrated heteroscedastic variance estimates, which we integrate into TIGER to study sequence-based heteroscedasticity in CRISPR-Cas13d. The goal of single-cell RNA sequencing (scRNA-seq) integration is to isolate technical variation from the biological signals of interest. Several popular integration methods use a semi-supervised variational autoencoding (VAE) framework (Kingma et al., 2014) with partially observed cell-type labels to learn a low-dimensional representation disentangled from technical effects.
To better handle partially observed labels in the amortized variational setting, we develop a new distribution on the simplex (Stirn et al., 2019) that mimics the Dirichlet distribution but has analytic reparameterization gradients and thus low gradient variance. Additionally, we develop a novel method for learning structured latent embeddings for the VAE (Kingma & Welling, 2014) that outperforms existing clustering methods on benchmark datasets and state-of-the-art scRNA-seq integration methods when combined with scVI (Lopez et al., 2018), a popular VAE-based scRNA-seq integration method.
Our final chapter includes theoretical and empirical results on how to improve the VampPrior’s (Tomczak & Welling, 2018) marginal likelihood by decoupling the prior and posterior variances. We further increase the VampPrior’s flexibility by replacing its uniform mixture with a Dirichlet process mixture. In tandem, these changes both boost the VampPrior’s modeling performance and reduce cluster utilization
Pollinator Protection in a Master Planned Community
Get PDFBaseline is a new master planned mixed-use community near Denver, Colorado, developed by McWhinney. Slated for 9,000 homes, a town center, schools, and commercial industrial uses, Baseline will also be the new home of Butterfly Pavilion, the first Association of Zoos and Aquariums-accredited non-profit invertebrate zoo in the world. Pollinator protection in the face of climate change is central to Baseline’s plan, which features pollinator-friendly public landscapes. The community will implement Butterfly Pavilion’s innovative “pollinator district” certification strategy that requires land use-specific environmental management, community engagement, and governance practices to support pollinator populations and biodiversity. Longitudinal assessments of pollinator population size and diversity will be important to understanding the impacts of Baseline’s targeted biodiversity protection strategy
Movilidad urbana y mitigación del cambio climático: acciones locales en la Zona Metropolitana de Monterrey antes y después de COVID-19
Get PDFLa llegada de la pandemia de COVID-19 condujo a gobiernos y comunidades a pensar en modos de vida en torno al distanciamiento social. La movilidad urbana, como medio para acceder a bienes y servicios modificó sus dinámicas tajantemente para adaptarse a la necesidad de quedarse en casa. Tal escenario obligó al fortalecimiento de los esquemas de conectividad para el trabajo a distancia en actividades no esenciales, a mejorar las infraestructuras de movilidad activa y a reducir aforos en el transporte público, transformando la vida en el espacio público. Este caso de estudio analiza las acciones locales demovilidad cotidiana que se llevaron a cabo antes y durante la pandemia en la Zona Metropolitana de Monterrey (ZMM), la segunda zona metropolitana más grande de México. Concretamente se describen acciones y proyectos de movilidad realizados desde antes y después de 2020, así como la manera en que estos afectaron el logro de los objetivos de sostenibilidad planteados previamente. Además, se esbozan ideas clave para el logro de la accesibilidad vía la planificación vis-a-vis las soluciones y gobernanza urbana imperantes de la movilidad.
Palabras clave: sistemas urbanos de transporte, movilidad urbana, movilidad activa, COVID-19, sostenibilidad urban