Search

Seeking answers to questions within long scientific research articles is a crucial area of study that aids readers in quickly addressing their inquiries. However, existing question-answering (QA) datasets based on scientific papers are limited in scale and focus solely on textual content. To address this limitation, we introduce SPIQA (Scientific Paper Image Question Answering), the first large-scale QA dataset specifically designed to interpret complex figures and tables within the context of scientific research articles across various domains of computer science. Leveraging the breadth of expertise and ability of multimodal large language models (MLLMs) to understand figures, we employ automatic and manual curation to create the dataset. We craft an information-seeking task involving multiple images that cover a wide variety of plots, charts, tables, schematic diagrams, and result visualizations. SPIQA comprises 270K questions divided into training, validation, and three different evaluation splits. Through extensive experiments with 12 prominent foundational models, we evaluate the ability of current multimodal systems to comprehend the nuanced aspects of research articles. Additionally, we propose a Chain-of-Thought (CoT) evaluation strategy with in-context retrieval that allows fine-grained, step-by-step assessment and improves model performance. We further explore the upper bounds of performance enhancement with additional textual information, highlighting its promising potential for future research and the dataset's impact on revolutionizing how we interact with scientific literature., Comment: preprint

Occlusal contacts are the locations at which the occluding surfaces of the maxilla and the mandible posterior teeth meet. Occlusal contact detection is a vital tool for restoring the loss of masticatory function and is a mandatory assessment in the field of dentistry, with particular importance in prosthodontics and restorative dentistry. The most common method for occlusal contact detection is articulating paper. However, this method can indicate significant medically false positive and medically false negative contact areas, leaving the identification of true occlusal indications to clinicians. To address this, we propose a multiclass Vision Transformer and Fully Convolutional Network ensemble semantic segmentation model with a combination hierarchical loss function, which we name as Hierarchical Fully Convolutional Branch Transformer (H-FCBFormer). We also propose a method of generating medically true positive semantic segmentation masks derived from expert annotated articulating paper masks and gold standard masks. The proposed model outperforms other machine learning methods evaluated at detecting medically true positive contacts and performs better than dentists in terms of accurately identifying object-wise occlusal contact areas while taking significantly less time to identify them. Code is available at https://github.com/Banksylel/H-FCBFormer., Comment: 15 pages, 5 figures, 2 tables, 5 equations, peer reviewed and accepted to Medical Imaging Understanding and Analysis (MIUA 2024)

In recent years, the rapid increase in scientific papers has overwhelmed traditional review mechanisms, resulting in varying quality of publications. Although existing methods have explored the capabilities of Large Language Models (LLMs) for automated scientific reviewing, their generated contents are often generic or partial. To address the issues above, we introduce an automated paper reviewing framework SEA. It comprises of three modules: Standardization, Evaluation, and Analysis, which are represented by models SEA-S, SEA-E, and SEA-A, respectively. Initially, SEA-S distills data standardization capabilities of GPT-4 for integrating multiple reviews for a paper. Then, SEA-E utilizes standardized data for fine-tuning, enabling it to generate constructive reviews. Finally, SEA-A introduces a new evaluation metric called mismatch score to assess the consistency between paper contents and reviews. Moreover, we design a self-correction strategy to enhance the consistency. Extensive experimental results on datasets collected from eight venues show that SEA can generate valuable insights for authors to improve their papers.

The paper referred to in the title is Allard/Almgren 1981 [AA81]. Several months ago Francesco Maggi emailed me saying that the inequalities 5.3(4),(5) of \cite{AA81} were wrong. In fact, as he pointed out, their incorrectness is immediately apparent if one takes $Z=0$ there. Maggi and his coauthor wanted to use these inequalities in their paper \cite{MN}. They were able to obtain a version of these inequalities which suffice for the carrying out the work in \cite{MN}. I started writing this paper in order to provide a version of these inequalities as needed in [AA81]. In thinking about this material I began to realize there were other problems with the paper. As a result I ended up {\em completely rewriting 5.1-5.4 on pages 243-248 of \cite{AA81}}; this rewrite is the contents of this paper. In addition to many annoying misprints many of the needed definitions and proofs in 5.1-5.4 are incomplete or absent. For example, it is not said where $z$ in 5.1(2) comes from; this omission completely surprised me since I remember doing a lot of work to come up with $z$ when \cite{AA81} was being written. Also, much of the necessary material about the reach of a submanifold as in \cite{FE2} is not provided in \cite{AA81}. The table of contents can serve as an index. In particular one sees there where the constants $\epsilon_1$ through $\epsilon_6$ are introduced. This material is extremely technical. One way to navigate this paper would be to start looking at Proposition 9.5 and work backwards.

Testing in production-like test environments is an essential part of quality assurance processes in many industries. Provisioning of such test environments, for information-intensive services, involves setting up databases that are rich-enough to enable simulating a wide variety of user scenarios. While production data is perhaps the gold-standard here, many organizations, particularly within the public sectors, are not allowed to use production data for testing purposes due to privacy concerns. The alternatives are to use anonymized data, or synthetically generated data. In this paper, we elaborate on these alternatives and compare them in an industrial context. Further we focus on synthetic data generation and investigate the use of recurrent neural networks for this purpose. In our preliminary experiments, we were able to generate representative and highly accurate data using a recurrent neural network. These results open new research questions that we discuss here, and plan to investigate in our future research., Comment: This paper was published in the proceedings of RAISE@ICSE in 2019

Human-computer interaction scholars are increasingly touching on topics related to politics or democracy. As these concepts are ambiguous, an examination of concepts' invoked meanings aids in the self-reflection of our research efforts. We conduct a thematic analysis of all papers with the word `politics' in abstract, title or keywords ($n$=378) and likewise 152 papers with the word `democracy.' We observe that these words are increasingly being used in human-computer interaction, both in absolute and relative terms. At the same time, we show that researchers invoke these words with diverse levels of analysis in mind: the early research focused on mezzo-level (i.e., small groups), but more recently the work has begun to include macro-level analysis (i.e., society and politics as played in the public sphere). After the increasing focus on the macro-level, we see a transition towards more normative and activist research, in some areas it replaces observational and empirical research. These differences indicate semantic differences, which -- in the worst case -- may limit scientific progress. We bring these differences visible to help in further exchanges of ideas and human-computer interaction community to explore how it orients itself to politics and democracy.

A draft of a paper by Mandelbaum, "The dynamic complementarity problem", was circulated in 1987, but has never been published. We give an exposition of two important results from that paper which are not readily accessible in the literature. The first is an example of a Skorokhod problem in two dimensions in the quadrant for which there is not uniqueness. The second is a proof of uniqueness for the Skorokhod problem in two dimensions in the quadrant in a critical case.

This work is motivated by two key trends. On one hand, large language models (LLMs) have shown remarkable versatility in various generative tasks such as writing, drawing, and question answering, significantly reducing the time required for many routine tasks. On the other hand, researchers, whose work is not only time-consuming but also highly expertise-demanding, face increasing challenges as they have to spend more time reading, writing, and reviewing papers. This raises the question: how can LLMs potentially assist researchers in alleviating their heavy workload? This study focuses on the topic of LLMs assist NLP Researchers, particularly examining the effectiveness of LLM in assisting paper (meta-)reviewing and its recognizability. To address this, we constructed the ReviewCritique dataset, which includes two types of information: (i) NLP papers (initial submissions rather than camera-ready) with both human-written and LLM-generated reviews, and (ii) each review comes with "deficiency" labels and corresponding explanations for individual segments, annotated by experts. Using ReviewCritique, this study explores two threads of research questions: (i) "LLMs as Reviewers", how do reviews generated by LLMs compare with those written by humans in terms of quality and distinguishability? (ii) "LLMs as Metareviewers", how effectively can LLMs identify potential issues, such as Deficient or unprofessional review segments, within individual paper reviews? To our knowledge, this is the first work to provide such a comprehensive analysis.

In recent years, many studies have been focusing on predicting the scientific impact of research papers. Most of these predictions are based on citations count or rely on features obtainable only from already published papers. In this study, we predict the likelihood for a research paper of winning an award only relying on information available at publication time. For each paper, we build the citation subgraph induced from its bibliography. We initially consider some features of this subgraph, such as the density and the global clustering coefficient, to make our prediction. Then, we mix this information with textual features, extracted from the abstract and the title, to obtain a more accurate final prediction. We made our experiments considering the ArnetMiner citation graph, while the ground truth on award-winning papers has been obtained from a collection of best paper awards from 32 computer science conferences. In our experiment, we obtained an encouraging F1 score of 0.694. Remarkably, The high recall and the low false negatives rate, show how the model performs very well at identifying papers that will not win an award. This behavior can help researchers in getting a first evaluation of their work at publication time. Lastly, we made some first experiments on interpretability. Our results highlight some interesting patterns both in topological and textual features.

Research-paper blog posts help scientists disseminate their work to a larger audience, but translating papers into this format requires substantial additional effort. Blog post creation is not simply transforming a long-form article into a short output, as studied in most prior work on human-AI summarization. In contrast, blog posts are typically full-length articles that require a combination of strategic planning grounded in the source document, well-organized drafting, and thoughtful revisions. Can tools powered by large language models (LLMs) assist scientists in writing research-paper blog posts? To investigate this question, we conducted a formative study (N=6) to understand the main challenges of writing such blog posts with an LLM: high interaction costs for 1) reviewing and utilizing the paper content and 2) recurrent sub-tasks of generating and modifying the long-form output. To address these challenges, we developed Papers-to-Posts, an LLM-powered tool that implements a new Plan-Draft-Revise workflow, which 1) leverages an LLM to generate bullet points from the full paper to help users find and select content to include (Plan) and 2) provides default yet customizable LLM instructions for generating and modifying text (Draft, Revise). Through a within-subjects lab study (N=20) and between-subjects deployment study (N=37 blog posts, 26 participants) in which participants wrote blog posts about their papers, we compared Papers-to-Posts to a strong baseline tool that provides an LLM-generated draft and access to free-form LLM prompting. Results show that Papers-to-Posts helped researchers to 1) write significantly more satisfying blog posts and make significantly more changes to their blog posts in a fixed amount of time without a significant change in cognitive load (lab) and 2) make more changes to their blog posts for a fixed number of writing actions (deployment)., Comment: 28 pages, 9 figures in main text (not appendix)

Recent advancements in Artificial Intelligence (AI), particularly the widespread adoption of Large Language Models (LLMs), have significantly enhanced text analysis capabilities. This technological evolution offers considerable promise for automating the review of scientific papers, a task traditionally managed through peer review by fellow researchers. Despite its critical role in maintaining research quality, the conventional peer-review process is often slow and subject to biases, potentially impeding the swift propagation of scientific knowledge. In this paper, we propose RelevAI-Reviewer, an automatic system that conceptualizes the task of survey paper review as a classification problem, aimed at assessing the relevance of a paper in relation to a specified prompt, analogous to a "call for papers". To address this, we introduce a novel dataset comprised of 25,164 instances. Each instance contains one prompt and four candidate papers, each varying in relevance to the prompt. The objective is to develop a machine learning (ML) model capable of determining the relevance of each paper and identifying the most pertinent one. We explore various baseline approaches, including traditional ML classifiers like Support Vector Machine (SVM) and advanced language models such as BERT. Preliminary findings indicate that the BERT-based end-to-end classifier surpasses other conventional ML methods in performance. We present this problem as a public challenge to foster engagement and interest in this area of research.

This document addresses the description of the corresponding "Urubots" Team for the 2024 Fira Air League, "Air Emergency Service (Indoor)." We introduce our team and an autonomous Unmanned Aerial Vehicle (UAV) that relies on computer vision for its flight control. This UAV has the capability to perform a wide variety of navigation tasks in indoor environments, without requiring the intervention of an external operator or any form of external processing, resulting in a significant decrease in workload and manual dependence. Additionally, our software has been designed to be compatible with the vehicle's structure and for its application to the competition circuit. In this paper, we detail additional aspects about the mechanical structure, software, and application to the FIRA competition., Comment: Team Description Paper for the FIRA RoboWorld Cup 2024

This document presents the design of an autonomous car developed by the UruBots team for the 2024 FIRA Autonomous Cars Race Challenge. The project involves creating an RC-car sized electric vehicle capable of navigating race tracks with in an autonomous manner. It integrates mechanical and electronic systems alongside artificial intelligence based algorithms for the navigation and real-time decision-making. The core of our project include the utilization of an AI-based algorithm to learn information from a camera and act in the robot to perform the navigation. We show that by creating a dataset with more than five thousand samples and a five-layered CNN we managed to achieve promissing performance we our proposed hardware setup. Overall, this paper aims to demonstrate the autonomous capabilities of our car, highlighting its readiness for the 2024 FIRA challenge, helping to contribute to the field of autonomous vehicle research., Comment: Team Description Paper for the FIRA RoboWorld Cup 2024

This paper proposes a mini autonomous car to be used by the team UruBots for the 2024 FIRA Autonomous Cars Race Challenge. The vehicle is proposed focusing on a low cost and light weight setup. Powered by a Raspberry PI4 and with a total weight of 1.15 Kilograms, we show that our vehicle manages to race a track of approximately 13 meters in 11 seconds at the best evaluation that was carried out, with an average speed of 1.2m/s in average. That performance was achieved after training a convolutional neural network with 1500 samples for a total amount of 60 epochs. Overall, we believe that our vehicle are suited to perform at the FIRA Autonomous Cars Race Challenge 2024, helping the development of the field of study and the category in the competition., Comment: Team Description Paper for the FIRA RoboWorld Cup 2024

An emerging area of research in situated and multimodal interactive conversations (SIMMC) includes interactions in scientific papers. Since scientific papers are primarily composed of text, equations, figures, and tables, SIMMC methods must be developed specifically for each component to support the depth of inquiry and interactions required by research scientists. This work introduces Conversational Papers (cPAPERS), a dataset of conversational question-answer pairs from reviews of academic papers grounded in these paper components and their associated references from scientific documents available on arXiv. We present a data collection strategy to collect these question-answer pairs from OpenReview and associate them with contextual information from LaTeX source files. Additionally, we present a series of baseline approaches utilizing Large Language Models (LLMs) in both zero-shot and fine-tuned configurations to address the cPAPERS dataset., Comment: 14 pages, 1 figure

The United States retains many aspects of a healthy open society, but there are indicators of trouble and deep divisions around the meaning and importance of democratic values. This debate has significant repercussions for universities and their academic communities. In the most-simple terms, there is a red and blue state divide over the role and importance of public institutions, including universities -- red representing largely rural states in which most voters vote Republican and blue being majority Democratic voters, often with one of the two parties having majorities in their respective state legislatures. Then there are so-called purple states in which both parties are vying for dominance, but they are fewer in number. This brief discusses this contemporary dynamic and its implication for higher education and science policy.

This brief discusses cases of neo-nationalist violations of academic freedom in Hungary and Poland. The most prominent case of neo-nationalist violation of academic freedom in Hungary is the fate of the Central European University (CEU). The circumstances of CEU's forced move out of Hungary came before the European Court of Justice regarding it a possible violation of EU law. The Court cited the General Agreement on Trade in Services (GATS) under one of the three pillars of the World Trade Organisation (WTO) 1994 agreement, free trade, and the determination that CEU was a form of international educational services that should not be denied to the people of Hungary. Poland has a similar hostile environment to academics and academic freedom, although with a glimmer of hope following recent elections. The brief also discusses how such open breaches of academic freedom as in Hungary or Poland, in which politicians directly try to exert influence on research institutions and professors, are fortunately rather rare in Germany. However, a confluence of factors perhaps obscures the differences between "academic freedom" and the "freedom of opinion." In Germany, academic freedom includes the search for topics, rigorous methodical investigation, and professional norms to express findings and competent opinions, whereas the free speech is outside of these professional norms. The brief concludes with a discussion of the role of universities and the future of democracy in the context of ensuring a space for free and open debate.

Starting this year, tens of thousands of Russian freshmen found themselves attending a new mandatory course -- "Foundations of Russian Statehood." Swiftly designed under the auspices of Putin's administration, this ideologically charged course aims to position Russia as a unique civilization-state, bolstering Putin's political narrative and providing justification for the full-scale invasion of Ukraine. Consider, for example, this excerpt from the course's instructional video: "The 'Russian world' extends beyond current Russian borders, transcending ethnicities, territories, religions, political systems, and ideological preferences." As this curriculum becomes standard in Russian universities, it contributes to the emerging trend of weaponizing Russian universities and turning them into instruments in Russia's war of attrition with Ukraine and its broader stand-off with the West. This report discusses this weaponization process and the impact it is having on Russian universities, faculty, students, and the academic communities they belong to. It is regrettably a story of back to the future, reminiscent of the Soviet era of repression and attempts at control and manipulation of academics.

China's higher-education system has been shaped in recent years by a trio of factors: the COVID-19 pandemic, the ambitions of Chinese leader Xi Jinping to make his country into an innovation superpower that is loyal to the Communist Party, and western alarm about those ambitions. But a fourth development, the slowing of China's formerly super-charged economy, could play a more prominent role going forward. This article examines these four factors.

Since 2020, the federal government has distributed almost $2 billion to American school districts to mitigate the effects of COVID-19 pandemic school closures. Released in three separate tranches through the Elementary and Secondary School Emergency Relief (ESSER) fund, the Trump and Biden administrations outlined parameters for spending this unprecedented amount of money and ensured that most of it went directly to local school districts, with a comparative sliver going to state education agencies. This report gives an overview of the most common ESSER expenditures to date and assesses the factors that might influence whether district ESSER investments will make a difference for students. It also provides recommendations for how stakeholders should evaluate the ESSER experiment once it officially ends.

Since establishing its first campus in 1868, the University of California (UC), California's land-grant university, developed into the nation's first multi-campus system in the United States, and is today widely recognized as the world's premier network of public research universities. This short essay provides an historical brief on the role that shared governance, and specifically the role of the Academic Senate, played in creating an academic culture of excellence and high achievement in pursuing its tripartite mission of teaching and learning, research and knowledge production, and public service. A key component in understanding the critical role of the Senate in UC's evolution from a single campus in Berkeley to now a ten-campus system is the university's unusual designation as a public trust in the state constitution that, beginning in 1879, protected the university at critical times from external political pressures and allowed the university to develop an internal academic culture guided by the Academic Senate. By the 1920s, the emergence of California's unique and innovative public system of higher education, with UC as the sole public provider of doctoral degrees and state funded research, also helps explain the ability of the UC system to maintain its mission and formulate what is termed a "One University" model. The Academic Senate has created coherency and shared values within UC, and a culture and expectation for faculty performance that is unique among universities around the world. This essay also offers a brief reflection on the Academic Senate's past influence, its current status, and prospective role. The overall intent is to provide context for the current academic community and higher education scholars regarding the past and future role of faculty in university governance and management, and what distinguishes UC in the pantheon of major research universities.

Following up on an individual-level randomized controlled trial of a Dana Center Math Pathways (DCMP) model, this study assessed longer-term impacts on students' math completion, academic progress, and academic attainment. The version of the DCMP that was assessed in this study diversified the developmental and college-level math course content that students take, separating it into distinct pathways that better aligned with their career interests. It also streamlined developmental math sequences into a one-semester developmental course for all students, regardless of placement level, and implemented evidence-based curricula and pedagogy to engage students in active problem solving that was pertinent to real-life situations. The study, which followed 1,411 students from four Texas community colleges and ten campuses, found that, in the five years after random assignment, program group students were consistently more likely to successfully complete their first college-level math courses than control group students. The study did not find impacts after five years on the number of overall college credits that students accrued or on the likelihood that students attained a credential or transferred to a four-year college.

Multiple measures assessment (MMA) has gained considerable momentum over the past decade as an alternative to traditional test-based procedures for placing incoming students into developmental or college-level coursework in math and English at broad-access colleges. Compared to standardized tests, which measure student performance at a single point in time, MMA (which often emphasizes high school GPA as a measure) provides a more holistic picture of students' academic preparation. Despite positive impacts on student outcomes that have been found by recent research on MMA, questions remain about whether the positive effects of MMA are sustained over time. This study--a follow-up to prior research using the same sample of students--employs a randomized controlled trial to investigate whether algorithmic MMA placement used at seven State University of New York (SUNY) community colleges led to better student outcomes, for up to four and a half years after randomization, than a system based on test scores alone. Nearly 13,000 incoming students who arrived at the seven colleges in fall 2016, spring 2017, and fall 2017 took placement tests and were randomly assigned to be placed using either the status quo method (business-as-usual group) or the alternative, algorithmic MMA method (program group). Using this sample, we estimate the overall treatment effects on placement into, enrollment in, and completion of college-level math and English as well as effects on other outcomes. We conduct similar analyses on race/ethnicity, Pell recipient status, and gender subgroups. We also descriptively examine the proportion of program group students who were bumped up (i.e., their placement changed from a developmental course placement to a college-level course placement) and bumped down (i.e., their placement changed from a college-level course placement to a developmental course placement) by the MMA algorithm, and we perform a cost-effectiveness analysis. We find that the MMA method used at the colleges improved access to and success in college-level courses and that lower cut scores in English rather than math are associated with larger and longer lasting impacts on completion of college-level coursework. While MMA improved outcomes among student subgroups, it had little to no impact on gaps in outcomes between subgroups. We also find that bumped-up students had substantially better outcomes in both math and English, while bumped-down students had substantially worse outcomes. Our results suggest that increased access to college-level courses is the driving factor in the positive outcomes experienced by program group students and that placement into standalone developmental courses can have detrimental effects on student outcomes. In the discussion of the study's results, we make recommendations for adopting MMA at colleges. Implemented together with other initiatives to support students, MMA can be a first step on the path to success for incoming students. [For the accompanying research brief, see ED632528.]

The growth of homeschooling from a somewhat fringe movement during the 1970s and 1980s to a more widespread and socially accepted approach in recent decades has provided a strong foundation of flexible learning models. When Florida's school choice expansion, House Bill 1, was introduced in January 2023, one of its goals was to allow more homeschoolers to participate in the education savings account (ESA) program. But many homeschoolers and homeschool advocacy groups balked at the proposal. Having worked hard to achieve independence, they were loath to be linked to government funding. Even if the program were optional, they feared that associated regulations would eventually extend to traditional homeschoolers. In the end, the bill passed with new language that satisfied traditional homeschoolers by creating a new option for parents to educate their children at home. As states continue to enact and expand education savings accounts, Colleen Hroncich argues that it is crucial that policymakers craft bills in a way that maximizes freedom and minimizes roadblocks.

We investigate the impact of attending a free school in England -- that is, a new start-up school that enjoys considerable autonomy while remaining in the state sector. We analyse the effects of two secondary free schools with different teaching philosophies: one follows a 'no excuse' paradigm, while the other one adopts a 'classical liberal', knowledge-rich approach. We establish causal effects exploiting admission lotteries and a distance-based regression discontinuity design. Both schools have a strong positive impact on student test scores on average. However, we also find heterogeneous effects: the 'no excuse' school mostly benefits boys, while the 'classical liberal' school mainly benefits White British and non-poor students. Both schools similarly reduce student absences and school mobility. Peer quality, teacher characteristics, and inspectorate ratings cannot fully explain the schools' effectiveness. Instead, a quantitative text analysis of the schools' 'vision and ethos' statements shows that the 'no excuse' and 'classical liberal' philosophies adopted by the two free schools clearly set them apart from the counterfactual schools where rejected applicants enrol, and likely explain their heterogeneous effects.

We advocate for a large scale imaging survey of nearby young moving groups and star-forming regions to directly detect exoplanets over an unexplored range of masses, ages and orbits. Discovered objects will be identified early enough in JWST's lifetime to leverage its unparalleled capabilities for long-term atmospheric characterisation, and will uniquely complement the known population of exoplanets and brown dwarfs. Furthermore, this survey will constrain the occurrence of the novel wide sub-Jovian exoplanet population, informing multiple theories of planetary formation and evolution. Observations with NIRCam F200W+F444W dual-band coronagraphy will readily provide sub-Jupiter mass sensitivities beyond ~0.4" (F444W) and can also be used to rule out some contaminating background sources (F200W). At this large scale, targets can be sequenced by spectral type to enable robust self-referencing for PSF subtraction. This eliminates the need for dedicated reference observations required by GO programs and dramatically increases the overall science observing efficiency. With an exposure of ~30 minutes per target, the sub-Jupiter regime can be explored across 250 targets for ~400 hours of exposure time including overheads. An additional, pre-allocated, ~100 hours of observing time would enable rapid multi-epoch vetting of the lowest mass detections (which are undetectable in F200W). The total time required for a survey such as this is not fixed, and could be scaled in conjunction with the minimum number of detected exoplanet companions., Comment: 5 pages, 2 figures. This white paper was submitted following a call from the "Working Group on Strategic Exoplanet Initiatives with HST and JWST" (https://sites.google.com/view/exoplanet-strategy-wg, final report in 10.48550/arXiv.2404.02932)

We advocate for further prioritisation of atmospheric characterisation observations of high mass transiting exoplanets and brown dwarfs. This population acts as a unique comparative sample to the directly imaged exoplanet and brown dwarf populations, of which a range of JWST characterisation observations are planned. In contrast, only two observations of transiting exoplanets in this mass regime were performed in Cycle 1, and none are planned for Cycle 2. Such observations will: improve our understanding of how irradiation influences high gravity atmospheres, provide insights towards planetary formation and evolution across this mass regime, and exploit JWST's unique potential to characterise exoplanets across the known population., Comment: 4 pages, 2 figures. This white paper was submitted following a call from the "Working Group on Strategic Exoplanet Initiatives with HST and JWST" (https://sites.google.com/view/exoplanet-strategy-wg, final report in 10.48550/arXiv.2404.02932)

We present a commentary on the famous 1964 paper of John Bell that rules out the entire class of underlying hidden variable theories for quantum mechanics that are local., Comment: Commentary, 4 pages, to appear in the November issue of Resonance

Equitable urban transportation applications require high-fidelity digital representations of the built environment: not just streets and sidewalks, but bike lanes, marked and unmarked crossings, curb ramps and cuts, obstructions, traffic signals, signage, street markings, potholes, and more. Direct inspections and manual annotations are prohibitively expensive at scale. Conventional machine learning methods require substantial annotated training data for adequate performance. In this paper, we consider vision language models as a mechanism for annotating diverse urban features from satellite images, reducing the dependence on human annotation to produce large training sets. While these models have achieved impressive results in describing common objects in images captured from a human perspective, their training sets are less likely to include strong signals for esoteric features in the built environment, and their performance in these settings is therefore unclear. We demonstrate proof-of-concept combining a state-of-the-art vision language model and variants of a prompting strategy that asks the model to consider segmented elements independently of the original image. Experiments on two urban features -- stop lines and raised tables -- show that while direct zero-shot prompting correctly annotates nearly zero images, the pre-segmentation strategies can annotate images with near 40% intersection-over-union accuracy. We describe how these results inform a new research agenda in automatic annotation of the built environment to improve equity, accessibility, and safety at broad scale and in diverse environments.

A crucial yet challenging task in galaxy evolution studies is the identification of distant merging galaxies, a task which suffers from a variety of issues ranging from telescope sensitivities and limitations to the inherently chaotic morphologies of young galaxies. In this paper, we use random forests and convolutional neural networks to identify high-redshift JWST CEERS galaxy mergers. We train these algorithms on simulated $3

High proved the following theorem. If the intersections of any two congruent copies of a plane convex body are centrally symmetric, then this body is a circle. In our paper we extend the theorem of High to the sphere and the hyperbolic plane, and partly to spaces of constant curvature. We also investigate the dual question about the convex hull of the unions, rather than the intersections. Let us have in $H^2$ proper closed convex subsets $K,L$ with interior points, such that the numbers of the connected components of the boundaries of $K$ and $L$ are finite. We exactly describe all pairs of such subsets $K,L$, whose any congruent copies have an intersection with axial symmetry; there are nine cases. (The cases of $S^2$ and ${\Bbb{R}}^2$ were described in Part I, i.e., \cite{5}.) Let us have in $S^d$, ${\Bbb{R}}^d$ or $H^d$ proper closed convex $C^2_+$ subsets $K,L$ with interior points, such that all sufficiently small intersections of their congruent copies are symmetric w.r.t.\ a particular hyperplane. Then the boundary components of both $K$ and $L$ are congruent, and each of them is a sphere, a parasphere or a hypersphere. Let us have a pair of convex bodies in $S^d$, ${\Bbb{R}}^d$ or $H^d$, which have at any boundary points supporting spheres (for $S^d$ of radius less than $\pi /2$). If the convex hull of the union of any congruent copies of these bodies is centrally symmetric, then our bodies are congruent balls (for $S^d$ of radius less than $\pi /2$). An analogous statement holds for symmetry w.r.t.\ a particular hyperplane. For $d=2$, suppose the existence of the above supporting circles (for $S^2$ of radius less than $\pi /2$), and, for $S^2$, smoothness of $K$ and $L$. If we suppose axial symmetry of all the above convex hulls, then our bodies are (incongruent) circles (for $S^2$ of radii less than $\pi /2$)., Comment: 48 pages. arXiv admin note: text overlap with arXiv:1601.04494

In a recent review of a paper by the Yakutsk Group, submitted to the Journal Physics of Atomic Nuclei and arXiv, the energy scales of the Yakutsk and Telescope Array (TA) experiments were examined. The authors developed a custom detector response simulator incorporating ionization, bremsstrahlung, pair production, and Compton scattering. Applying this simulator to both Yakutsk and TA surface detectors, they concluded that the TA energy scale might be incorrect due to a misdefined ``response unit.'' They referenced the TA's ``energy deposit formula'' from the literature, scaling it by two factors attributed to the thickness and density of the TA scintillator. Their simulations, using the QGSJET-II-04 hadronic interaction model, agreed with TA's calculations for vertical showers but not for inclined showers, suggesting an incorrect VEM unit of 2.05 MeV. However, this conclusion was found to be incorrect. The TA's energy deposit formula, derived from detailed Monte Carlo simulations using GEANT4, accurately represents the most probable energy deposit by a charged particle in the TA scintillator. The value of 2.05 MeV accounts for the scintillator's thickness and density and is validated by excellent agreement between TA's simulated and observed data. The Yakutsk Group's misinterpretation and incorrect application of the TA formula led to their erroneous conclusion., Comment: 2pages, comments on arXiv:2404.16948

This white paper summarizes the Workshop on Polarized Target Studies with Real Photons in Hall D at Jefferson Lab, that took place on 21 February 2024. The Workshop included about 45 participants both online and in person at Florida State University in Tallahassee. Contributions describe the experimental infrastructure available in Hall D and potential physics applications. The rate and detection capabilities of Hall D are outlined, as well as the properties of a circularly polarized photon beam and a polarized target. Possible physics measurements include light and strange quark baryon spectroscopy, the GDH sum rule, proton structure accessed through measurement of Generalized Parton Distributions and modification of nucleon structure within the nuclear medium., Comment: 25 pages, 8 figures

We introduce the sequence classification problem CIViC Evidence to the field of medical NLP. CIViC Evidence denotes the multi-label classification problem of assigning labels of clinical evidence to abstracts of scientific papers which have examined various combinations of genomic variants, cancer types, and treatment approaches. We approach CIViC Evidence using different language models: We fine-tune pretrained checkpoints of BERT and RoBERTa on the CIViC Evidence dataset and challenge their performance with models of the same architecture which have been pretrained on domain-specific text. In this context, we find that BiomedBERT and BioLinkBERT can outperform BERT on CIViC Evidence (+0.8% and +0.9% absolute improvement in class-support weighted F1 score). All transformer-based models show a clear performance edge when compared to a logistic regression trained on bigram tf-idf scores (+1.5 - 2.7% improved F1 score). We compare the aforementioned BERT-like models to OpenAI's GPT-4 in a few-shot setting (on a small subset of our original test dataset), demonstrating that, without additional prompt-engineering or fine-tuning, GPT-4 performs worse on CIViC Evidence than our six fine-tuned models (66.1% weighted F1 score compared to 71.8% for the best fine-tuned model). However, performance gets reasonably close to the benchmark of a logistic regression model trained on bigram tf-idf scores (67.7% weighted F1 score).

A response to commenter Ke Lan's comment on our paper published in Nature Communications (2023)14:5782 by J. Yan et al

We identify two major issues in the SoftDist paper (Xia et al.): (1) the failure to run all steps of different baselines on the same hardware environment, and (2) the use of inconsistent time measurements when comparing to other baselines. These issues lead to flawed conclusions. When all steps are executed in the same hardware environment, the primary claim made in SoftDist is no longer supported., Comment: comment on arXiv:2406.03503, 4 pages, 1 figure and 1 table