Search

Absent a nationwide plan for universal public prekindergarten, states and districts have taken various approaches to increasing access to school-based educational opportunities for their youngest learners. Though some of these programs have focused on making public prekindergarten available to all families, others have targeted families most in need by introducing means-tested programs. Early efforts to expand access to public schooling for preschool-age children, such as the War on Poverty-era federal Head Start program, focused on reaching low-income children. More recently, states and cities have made significant investments in their own public preschool programs, often citing their utility in combatting racial, ethnic, and income-based test score differences that are apparent at school entry. A long-standing debate has weighed the relative merits of targeting limited resources to equalize early educational opportunities versus providing universal programs. But an underexplored distinction between the two approaches is how they might contribute to racial and ethnic school segregation among this country's youngest students. Key findings in this essay include: (1) Black and Hispanic prekindergarten students in means-tested programs experience consistently higher rates of racial isolation and lower rates of exposure to white students relative to their K-12 counterparts than those in open programs; (2) In states with open-enrollment prekindergarten, the proportion of Black prekindergarten students enrolled in racially isolated schools is nearly identical to that of Black first-graders (roughly 46 percent), but in states where prekindergarten is means tested, the proportion of Black preschoolers in racially isolated schools is 6 percentage points higher than that of Black first-graders; (3) Hispanic preschool students are 5 percentage points more likely than their first-grade counterparts to attend racially isolated schools in states with means-tested prekindergarten and are 3 percentage points less likely than first-grade students to do so in states where prekindergarten is open; and (4) In comparing states with diverse and robust prekindergarten programs, the data show that Black and Hispanic prekindergarten students in Texas and North Carolina, which have means-tested programs, have less exposure to white students than their counterparts in Georgia and Oklahoma, which have universal enrollment.

Few-shot audio classification is an emerging topic that attracts more and more attention from the research community. Most existing work ignores the specificity of the form of the audio spectrogram and focuses largely on the embedding space borrowed from image tasks, while in this work, we aim to take advantage of this special audio format and propose a new method by hallucinating high-frequency and low-frequency parts as structured concepts. Extensive experiments on ESC-50 and our curated balanced Kaggle18 dataset show the proposed method outperforms the baseline by a notable margin. The way that our method hallucinates high-frequency and low-frequency parts also enables its interpretability and opens up new potentials for the few-shot audio classification., Comment: Accepted at ICASSP 2023

The coupling of electrical batteries with variable renewable power generation can increase the production flexibility and revenue of power plant operators. This study focuses on developing an optimisation model to manage the operational revenue of a renewable power unit comprising a wind farm, solar photovoltaic (PV) power plant, and electrical battery. The power system integration was conducted by formulating a mixed-integer linear programme to schedule the day-ahead operation of the renewable power unit in two liberalised power markets: the Italian and Iberian day-ahead power markets. Several scenarios and case studies have been analysed to assess the value of storage for revenue maximisation. The proposed methodology results reveal that the average yearly net revenue of the hybrid PV-wind-storage power plant can increase by 4% compared to the standalone operation of the wind and solar PV power plants. Additionally, the results indicate that in the markets analysed the coupling of storage to wind power generates a higher revenue than coupling storage systems with solar power. The study demonstrated a correlation between the increase in revenue and the capacity installed in the battery and concluded that the use of hybrid VRE storage systems would be feasible in the Iberian Peninsula and Italy for battery installation costs ranging from €14,804/MW to €38,267/MW. © 2022 Elsevier Ltd

We propose an ensemble method to improve the generalization performance of trading strategies trained by deep reinforcement learning algorithms in a highly stochastic environment of intraday cryptocurrency portfolio trading. We adopt a model selection method that evaluates on multiple validation periods, and propose a novel mixture distribution policy to effectively ensemble the selected models. We provide a distributional view of the out-of-sample performance on granular test periods to demonstrate the robustness of the strategies in evolving market conditions, and retrain the models periodically to address non-stationarity of financial data. Our proposed ensemble method improves the out-of-sample performance compared with the benchmarks of a deep reinforcement learning strategy and a passive investment strategy.

Determining the status of epidermal growth factor receptor (EGFR) T790M mutation is crucial for guiding further treatment intervention in advanced non-small cell lung cancer (NSCLC) patients who develop acquired resistance to initial EGFR tyrosine kinase inhibitor (TKI) treatment. Circulating tumor cells (CTCs) which contain plentiful copies of well-preserved RNA offer an ideal source for noninvasive detection of T790M mutation in NSCLC. We developed a CTC-based digital assay which synergistically integrates NanoVelcro Chips for enriching NSCLC CTCs and reverse-transcription droplet digital PCR (RT-ddPCR) for quantifying T790M transcripts in the enriched CTCs. We collected 46 peripheral arterial and venous blood samples from 27 advanced NSCLC patients for testing this CTC-based digital assay. The results showed that the T790M mutational status observed by the CTC-based digital assay matched with those observed by tissue-based diagnostic methods. Furthermore, higher copy numbers of T790M transcripts were observed in peripheral arterial blood than those detected in the matched peripheral venous blood. In short, our results demonstrated the potential of the NanoVelcro CTC-digital assay for noninvasive detection of the T790M mutation in NSCLC, and suggested that peripheral arterial blood sampling may offer a more abundant CTC source than peripheral venous blood in advanced NSCLC patients.