Search

Obtaining the solution of constrained optimization problems as a function of parameters is very important in a multitude of applications, such as control and planning. Solving such parametric optimization problems in real time can present significant challenges, particularly when it is necessary to obtain highly accurate solutions or batches of solutions. To solve these challenges, we propose a learning-based iterative solver for constrained optimization which can obtain very fast and accurate solutions by customizing the solver to a specific parametric optimization problem. For a given set of parameters of the constrained optimization problem, we propose a first step with a neural network predictor that outputs primal-dual solutions of a reasonable degree of accuracy. This primal-dual solution is then improved to a very high degree of accuracy in a second step by a learned iterative solver in the form of a neural network. A novel loss function based on the Karush-Kuhn-Tucker conditions of optimality is introduced, enabling fully self-supervised training of both neural networks without the necessity of prior sampling of optimizer solutions. The evaluation of a variety of quadratic and nonlinear parametric test problems demonstrates that the predictor alone is already competitive with recent self-supervised schemes for approximating optimal solutions. The second step of our proposed learning-based iterative constrained optimizer achieves solutions with orders of magnitude better accuracy than other learning-based approaches, while being faster to evaluate than state-of-the-art solvers and natively allowing for GPU parallelization., Comment: 13 pages, 1 figure

As current trends across the country indicate widespread declining student enrollment and attendance, EdChoice is interested in how states respond. It enlisted Hanover Research to summarize publicly available information for each of the 50 U.S. states about the role enrollment plays in state funding formulas, as well as any state-level policies that may reduce the impact of declining enrollment on education funding. This report presents an analysis of the policy scan and summarizes the key findings of the policy scan including trends in education funding formulas, funding, and enrollment protections across the United States. Key findings covered in this report include: the type of education funding formulas in each state; the metrics these formulas use to measure enrollment; as well as the declining enrollment; hold harmless, and temporary provisions for enrollment and funding protections identified in the initial policy scan. This policy scan analysis is part of a series of studies that will enable EdChoice to examine the broader landscape of the impact of declining enrollment on school funding and related policies across the United States. The full version of the policy scan can be found in "How States Protect Funding for K-12 Public Schools. State Profiles Index."

Public education funding is a complex matter. It is widely understood that dollars for public schools come from federal, state, and local sources, but the revenue stream from each source is calculated in complex formulas and criteria that can make it hard for anyone to understand the system fully. Calculating aid based on some measure of the number of students that a school district serves is the primary modus operandi by state education departments. Some states determine aid based on average daily attendance, and others use enrollment counts. Each approach has trade-offs. The goal of this article is to explain various school funding protections (also known as "hold harmless" provisions) that states take--which give additional funds to school districts that otherwise would not have received them--in order to provide a better understanding of the various types of funding protections and the trade-offs they present to policymakers and citizens. The first part of the article examines declining enrollment provisions and their trade-offs, while the second part does the same for funding guarantees. A follow-up discussion with suggestions for using either policy more effectively is provided.

As current trends across the country indicate widespread declining student enrollment and attendance, EdChoice is interested in conducting an evaluation of state-level school funding and enrollment policies that protect public schools against reduced funding due to declining enrollment. To support this effort, EdChoice enlisted Hanover to conduct a preliminary policy scan to explore current funding practices at each state-level department of education across the United States. This report summarizes publicly available information for each of the 50 states regarding the role of enrollment in state funding formulas and state-level policies that may reduce the impact of declining enrollment on education funding. Using publicly available sources related to state-level departments of education and government entities, Hanover sought to address the following research questions for each state: (1) What is the current structure or formula used for school-level funding? (2) What type of enrollment count and count date does the formula use for funding purposes? (3) What mechanisms in the formula are relevant to enrollment or enrollment projections? (4) What specific state-level funding and/or hold harmless policies exist to protect school-level funding against declining enrollment? (5) Which statutes and legislation define these enrollment protection policies? (6) How have these policies changed over time? and (7) Do any of these protections extend to public charter schools? Hanover reviewed secondary sources for each state to identify relevant policies for this analysis, prioritizing state legislation and sources from the state-level department of education. It used information from other secondary sources as necessary. This report contains two sections: (1) Overview of State Policies--summarizes state funding formulas and relevant enrollment protections for each of the 50 states; and (2) State Profiles--provides a detailed profile for each state, with an overview of its funding formula and a detailed summary of any funding protections that provide protection against funding reductions due to declining enrollment.

The goal of "The 123s" is to present the increasingly large body of private school choice research in a clear and easy-to-read format and cite the relevant studies so that anyone who is interested in the individual results can easily find them and read in more detail. This report is divided into 11 sections. The first section summarizes the number of studies and how many come to which conclusion. The following sections present the eight outcomes covered in this publication, including school safety and climate--a new subject of study. They are followed by a list of reviews that other researchers conducted about the eight outcomes covered. The last section discusses the strengths and limitations of research on school choice. Finally, tables in the Appendix present the various programs, organized by type. [For the 2022 report, see ED625418.]

This report is about the educational borders that have sprung up across America. Some are school district boundaries. They can form an invisible barrier between students and the schools that they might want to attend. Other borders, often aligned with the municipal boundaries of cities, counties, and states, restrict who can teach where, how much teachers are paid, and what their retirement looks like. All have profoundly shaped the way that American schools look. In this report, the authors explain how school district boundaries, which limit where children can attend school, can have negative impacts on students and homeowners, especially the most disadvantaged. They also explain how boundaries, in the form of pension rules, negatively affect teachers. They propose an alternative: a system with fewer boundaries--one that allows students to access a wide array of schools, regardless of where they live, and one that does not punish teachers for moving between public, charter, or private sectors, or between states (or even within states, in some instances)--will improve outcomes for students, increase equity among homeowners, and promote retirement security for teachers. The authors argue that as states continue to move toward providing more educational opportunities for families, teachers will need to fill some of the demands for a diverse set of options. The way that educational systems compensate teachers, however, is outdated and ill-equipped to facilitate a dynamic teacher labor market to meet families' needs.

The Michigan Legislature passed legislation in October 2021 to create the Student Opportunity Scholarship Program, which would give thousands of families across the state more spending power to help meet their children's educational needs. This report unpacks this proposed education savings account program and analyzes its potential fiscal impacts.

Retirement benefits are integral to teacher compensation; yet, most pension plans are designed to reward teachers who remain in a system for a full career while levying substantial penalties on those who do not. Even teachers who devote their lives to the profession but do not stay in a single system for their entire careers are not served well by many of these plans. States must facilitate a dynamic educational workforce while placing teachers on a path to a secure retirement. Unfortunately, many teachers' retirement plans today fall short of these objectives. This report discusses final salary defined benefit (DB) plans, pension borders, issues with how benefits accrue, and pension plan reforms and why conservatives should champion for them.

School choice critics argue that choice programs drain resources from public schools and therefore harm students who remain in them. Because policymakers are tasked with balancing their states' budgets and ensuring that their public schools meet educational provisions in their states' constitutions, they are concerned with the fiscal effects of these programs. More than two dozen studies have examined educational choice programs' effects on students enrolling in nearby public schools. Researchers have conducted a handful of systematic reviews of competitive effects research and, more recently, a meta-analysis of this body of research. In each of these reviews, researchers conclude that students who remain in district schools after exposure to educational choice programs tend to experience modest educational benefits. But the question remains whether educational choice programs lead to higher costs for taxpayers or fewer resources for students who remain in public schools. This report summarizes the fiscal effects of educational choice programs across the United States from an analysis of 40 private educational choice programs in 19 states plus D.C. The programs in the analysis include three education savings accounts programs, 19 school voucher programs, and 18 tax-credit scholarship programs. The report also provides context by presenting basic facts about the size and scope of each program, in terms of participation and funding, relative to each state's public school system. It presents the facts on taxpayer funding disparities between students using the choice programs and their peers in public schools.

While a substantial body of research exists to demonstrate the benefits that choice policies such as education savings accounts (ESAs) have on various student, family, community, and societal outcomes, the fiscal impact of these policies is also an important part of the debate. This report examines solely the potential fiscal effects of educational choice programs on New York State taxpayers and district schools. This report presents fiscal models for two ESA amounts: one worth $6,500 (90% of the state's average marginal per-pupil cost); and the other worth $9,900 (100% of the statewide average amount of total per-pupil state aid). For both scenarios, the models estimate the net fiscal effect of the program, which is the difference between the cost of the ESA program and the savings from "switchers"--that is, students who would have remained in a public school system if not for the program. As this report demonstrates, ESAs will provide fiscal benefits for the school system, meaning that students who remain in public schools generally end up with more resources on a per-pupil basis. The Empire State can be confident in expanding educational opportunities without adverse effect to its bottom line.

Education savings accounts (ESAs) are education funding mechanisms that allow for families to receive a deposit of public funds to a government-authorized savings account. Using student-level longitudinal data, this paper examines how families participating in the Florida Gardiner Scholarship Program use education savings account funds. Results indicate that families use an increasing proportion of ESA funds when students remain in the program. The longer students remain in the program, the share of ESA funds devoted to private school tuition decreases while expenditure shares increase for curriculum, instruction, tutoring, and specialized services. Students in rural areas not only use a greater portion of their ESA funds than families in urban and suburban areas, but they also spend smaller portions of their funds on tuition and appear to customize more.

From an analysis of 40 private educational choice programs in 19 states plus D.C., this report summarizes the facts and the evidence on the fiscal effects of educational choice programs across the United States. The programs in the analysis include 3 education savings accounts programs, 19 school voucher programs, and 18 tax-credit scholarship programs. This study estimates the combined net fiscal effects of each educational choice program on state and local taxpayers through FY 2018--in both the short run and the long run. The report also provides context by presenting basic facts about the size and scope of each program, in terms of participation and funding, relative to each state's public school system. It presents the facts on taxpayer funding disparities between students using the choice programs and their peers in public schools. The information contained in this report provides information to help understand whether educational choice programs have positive, negative, or neutral fiscal effects on state and local taxpayers. The results from this fiscal analysis are not surprising given that educational choice programs are funded at a significantly lower public expense than public school systems. Overall, students participating in educational choice programs comprise 2.3 percent of publicly funded K-12 students but represent just 1.0 percent of total public spending. These basic facts provide important background for evaluating claims that private educational choice programs will harm the resource levels for students who remain in district schools.

This paper discusses the potential fiscal effects of education savings accounts for K-12 in New Jersey on the state and local taxpayers. Education savings accounts are a way to expand educational opportunity for all families, particularly those in need. For parents who voluntarily elect to customize their child's education, a specified amount is annually deposited to an education savings account (ESA) at a state-selected agency. Families can then draw on this account for a variety of approved instructional expenses, including tuition and fees for independent school placements, textbooks, tutoring, online classes, transportation, and special services. This brief examines the potential fiscal effects of ESAs in New Jersey and proceeds as follows: the first section summarizes research on non-fiscal outcomes of educational choice programs that already exist in other states, followed by a discussion about how ESAs could improve New Jersey's finances. The next section explains the methodology for estimating the fiscal effects of educational choice programs. The paper then presents the results of a fiscal analysis of two hypothetical ESA programs and concludes.

Unfair stereotypical biases (e.g., gender, racial, or religious biases) encoded in modern pretrained language models (PLMs) have negative ethical implications for widespread adoption of state-of-the-art language technology. To remedy for this, a wide range of debiasing techniques have recently been introduced to remove such stereotypical biases from PLMs. Existing debiasing methods, however, directly modify all of the PLMs parameters, which -- besides being computationally expensive -- comes with the inherent risk of (catastrophic) forgetting of useful language knowledge acquired in pretraining. In this work, we propose a more sustainable modular debiasing approach based on dedicated debiasing adapters, dubbed ADELE. Concretely, we (1) inject adapter modules into the original PLM layers and (2) update only the adapters (i.e., we keep the original PLM parameters frozen) via language modeling training on a counterfactually augmented corpus. We showcase ADELE, in gender debiasing of BERT: our extensive evaluation, encompassing three intrinsic and two extrinsic bias measures, renders ADELE, very effective in bias mitigation. We further show that -- due to its modular nature -- ADELE, coupled with task adapters, retains fairness even after large-scale downstream training. Finally, by means of multilingual BERT, we successfully transfer ADELE, to six target languages., Comment: Accepted for EMNLP-Findings 2021

Critics of private school choice programs argue that they drain resources from public schools and increase taxpayer costs. Supporters of these programs, on the other hand, argue that these programs save taxpayers money because the average amount of scholarships is less than per pupil spending in public schools. Given these fiscal concerns, when legislators introduce a private school choice bill in a state, policymakers want to know about the potential fiscal effects of these programs on their state and local public schools' budgets. This paper identifies 27 distinct estimates of switcher rates from nine lottery-based studies of six private school choice programs in the United States that report information about which types of schools students enroll in after they apply to a choice program and do not win a lottery.

This paper reports results from a fiscal analysis of Pennsylvania SB 299, a proposal to expand the state's two tax-credit scholarship programs. Pennsylvania currently has two programs in operation, the Educational Improvement Tax Credit (EITC) Program and the Opportunity Scholarship Tax Credit (OSTC) Program. This paper is organized as follows. First, it provides context to help policy makers and stakeholders evaluate the fiscal costs and benefits of the state's two tax-credit scholarship programs and their expansion under SB 299. It then explains the data and methods used to conduct a fiscal analysis of the programs and SB 299. Next it reports results from the analysis. Finally, it provides concluding thoughts.

Are private schools safer than public schools? Do they use the same security procedures? Do they suspend or expel students at a higher rate? Do private schools that participate in school choice programs report the same safety issues as private schools that do not participate in school choice programs? How does charter school safety compare to traditional public school safety? This report seeks to rigorously address these questions--and more--by collecting data from 618 school leaders across Indiana. We used their responses to empirically examine the relationship between private schooling and the presence of safety-related practices such as building access, security cameras, wearing of badges or picture IDs and more.

For the past 30 years, America's K-12 education system has experienced an era of expanding educational choice. Although students who participate in private school choice programs and are enrolled in charter schools comprise a small portion of the K-12 population relative to students in district schools, choice is becoming a larger part of the K-12 public education landscape. It will be important for policymakers to begin thinking about how their states can foster a robust public education system where school choice is a feature, not an afterthought. To do this, they should give thought to designing funding systems that fund students directly. In this paper, the authors offer a different lens for policymakers to examine how states fund public K-12 education and propose general guidelines for fostering a robust system of public education. The goal is to explain how states might better organize their funding systems to create a coherent system that allows students to choose among conventional district public schools, public charter schools, private schools, and educational services outside of school walls--universal educational choice. Such a system of universal choice would maximize the benefits of education funding to students, educators, and taxpayers.