Search

Annotating cross-document event coreference links is a time-consuming and cognitively demanding task that can compromise annotation quality and efficiency. To address this, we propose a model-in-the-loop annotation approach for event coreference resolution, where a machine learning model suggests likely corefering event pairs only. We evaluate the effectiveness of this approach by first simulating the annotation process and then, using a novel annotator-centric Recall-Annotation effort trade-off metric, we compare the results of various underlying models and datasets. We finally present a method for obtaining 97\% recall while substantially reducing the workload required by a fully manual annotation process. Code and data can be found at https://github.com/ahmeshaf/model_in_coref, Comment: The 17th Liguistics Annotation Workshop, 2023 (LAW-XVII) short paper. 10 pages, 6 figures, 1 table

This project demonstrates how medical corpus hypothesis generation, a knowledge discovery field of AI, can be used to derive new research angles for landscape and urban planners. The hypothesis generation approach herein consists of a combination of deep learning with topic modeling, a probabilistic approach to natural language analysis that scans aggregated research databases for words that can be grouped together based on their subject matter commonalities; the word groups accordingly form topics that can provide implicit connections between two general research terms. The hypothesis generation system AGATHA was used to identify likely conceptual relationships between emerging infectious diseases (EIDs) and deforestation, with the objective of providing landscape planners guidelines for productive research directions to help them formulate research hypotheses centered on deforestation and EIDs that will contribute to the broader health field that asserts causal roles of landscape-level issues. This research also serves as a partial proof-of-concept for the application of medical database hypothesis generation to medicine-adjacent hypothesis discovery.

We present two- and three-dimensional hydrodynamic simulations of $\sim$kpc-scale AGN jets with mean jet powers in the range $1-7\times10^{45}\,$erg~s$^{-1}$, in which the jet power varies (through variation of the Lorentz factor) according to a flicker or pink noise power spectrum. We find the morphology and dynamics of the jet-cocoon system depends on the amplitude of the variability with a clear correspondence between the shape of the cocoon and the historical activity. The jet advances quickly during high-power states, whereas quiescent periods instead produce passive periods of inflation resembling Sedov-Taylor blast waves. Periods of high activity preferentially produce hotspots and create stronger backflow as they maximise the pressure gradient between the jet head and cocoon. The variability can also lead to propagating internal shock structures along the jet. Our work suggests that variability and flickering in the jet power has important implications, which we discuss, for observations of radio galaxies, ultrahigh energy cosmic ray acceleration and jet power to luminosity correlations. We explore the link between morphology and fuelling, and suggest that chaotic cold accretion should introduce a relatively small scatter in radio luminosity ($\sim0.2$ dex) and modest imprints on morphology; sources such as Hercules A and Fornax A, which show evidence for more dramatic variability, may therefore require redder power spectra, or be triggered by mergers or other discrete events. We suggest ways to search for jet flickering observationally and propose that radio galaxies may be an important diagnostic of Myr timescale AGN fuelling, due to their `long-term memory'., Comment: 20 pages, 15 figures, 1 table. Accepted for publication in MNRAS. Animated version of figure 3 available at https://youtu.be/YhDHokWXuKk with other movies on the same youtube channel. Full supplementary material available at https://github.com/jhmatthews/flicker-supplement

This project aimed to design, simulate, and implement a two-axis inertially stabilised platform (ISP) for use in astronomical applications. It aimed to approximate the stabilisation of a Meade ETX-90 3.5" compound telescope at low-cost using a mechanical assembly designed to geometrically and inertially model the telescope. A set of system specifications was developed to guide design decisions and to provide an analysis framework against which the performance of the implemented system was compared. The electro-mechanical structure of the ISP was designed and manufactured, the associated electrical systems were specified and configured, an image processing script capable of detecting and locating the centre of the Moon in a camera field-of-view was written, a complete simulation model for the system was developed and used to design various classical controllers for the ISP control system. These controllers were implemented on a STM32F051 microcontroller and a user interface was written in LabVIEW to facilitate intuitive user control of the system and perform datalogging of the system runtime data., Comment: 2019 Southern African Universities Power Engineering Conference/Robotics and Mechatronics/Pattern Recognition Association of South Africa (SAUPEC/RobMech/PRASA)

We formalize and study a phenomenon called feature collapse that makes precise the intuitive idea that entities playing a similar role in a learning task receive similar representations. As feature collapse requires a notion of task, we leverage a simple but prototypical NLP task to study it. We start by showing experimentally that feature collapse goes hand in hand with generalization. We then prove that, in the large sample limit, distinct words that play identical roles in this NLP task receive identical local feature representations in a neural network. This analysis reveals the crucial role that normalization mechanisms, such as LayerNorm, play in feature collapse and in generalization.

Spin currents are key to spin torque devices, but determining the proper spin current is non-trivial. Here we derive a general quantum-mechanical formula for the intrinsic proper spin current showing that it is topological and can be finite in the gap. For topological insulators with an out of plane magnetization and the chemical potential in the surface state gap, the net spin torque is determined by the competition between the topological spin-Hall torque due to the bulk and the topological Edelstein effect due to the surface states. We also discuss spin-3/2 hole quantum wells.

Event Coreference Resolution (ECR) is the task of linking mentions of the same event either within or across documents. Most mention pairs are not coreferent, yet many that are coreferent can be identified through simple techniques such as lemma matching of the event triggers or the sentences in which they appear. Existing methods for training coreference systems sample from a largely skewed distribution, making it difficult for the algorithm to learn coreference beyond surface matching. Additionally, these methods are intractable because of the quadratic operations needed. To address these challenges, we break the problem of ECR into two parts: a) a heuristic to efficiently filter out a large number of non-coreferent pairs, and b) a training approach on a balanced set of coreferent and non-coreferent mention pairs. By following this approach, we show that we get comparable results to the state of the art on two popular ECR datasets while significantly reducing compute requirements. We also analyze the mention pairs that are "hard" to accurately classify as coreferent or non-coreferent. Code at https://github.com/ahmeshaf/lemma_ce_coref, Comment: Findings of the Association of Computational Linguistics, ACL 2023. 13 pages, 7 figures, 6 tables

Accessing the low-energy non-equilibrium dynamics of materials and their polaritons with simultaneous high spatial and temporal resolution has been a bold frontier of electron microscopy in recent years. One of the main challenges lies in the ability to retrieve extremely weak signals while simultaneously disentangling amplitude and phase information. Here, we present Free-Electron Ramsey Imaging (FERI), a microscopy approach based on light-induced electron modulation that enables coherent amplification of optical near-fields in electron imaging. We provide simultaneous time-, space-, and phase-resolved measurements of a micro-drum made from a hexagonal boron nitride membrane visualizing the sub-cycle dynamics of 2D polariton wavepackets therein. The phase-resolved measurements reveals vortex-anti-vortex singularities on the polariton wavefronts, together with an intriguing phenomenon of a traveling wave mimicking the amplitude profile of a standing wave. Our experiments show a 20-fold coherent amplification of the near-field signal compared to conventional electron near-field imaging, resolving peak field intensities in the order of ~W/cm2, corresponding to field amplitudes of a few kV/m. As a result, our work paves the way for spatio-temporal electron microscopy of biological specimens and quantum materials, exciting yet delicate samples that are currently difficult to investigate.

Solving optimization problems is challenging for existing digital computers and even for future quantum hardware. The practical importance of diverse problems, from healthcare to financial optimization, has driven the emergence of specialised hardware over the past decade. However, their support for problems with only binary variables severely restricts the scope of practical problems that can be efficiently embedded. We build analog iterative machine (AIM), the first instance of an opto-electronic solver that natively implements a wider class of quadratic unconstrained mixed optimization (QUMO) problems and supports all-to-all connectivity of both continuous and binary variables.Beyond synthetic 7-bit problems at small-scale, AIM solves the financial transaction settlement problem entirely in analog domain with higher accuracy than quantum hardware and at room temperature. With compute-in-memory operation and spatial-division multiplexed representation of variables, the design of AIM paves the path to chip-scale architecture with 100 times speed-up per unit-power over the latest GPUs for solving problems with 10,000 variables. The robustness of the AIM algorithm at such scale is further demonstrated by comparing it with commercial production solvers across multiple benchmarks, where for several problems we report new best solutions. By combining the superior QUMO abstraction, sophisticated gradient descent methods inspired by machine learning, and commodity hardware, AIM introduces a novel platform with a step change in expressiveness, performance, and scalability, for optimization in the post-Moores law era., Comment: Main sections plus supplementa material for a total of 41 pages. 7 figures

We discuss our recent constraints on the coupling of Very-Light Axion-Like Particles (of masses $<$$ 10^{-12} \ \mathrm{eV}$) to electromagnetism from $Chandra$ observations of the cluster-hosted Active Galactic Nuclei (AGN) H1821+643 and NGC1275. In both cases, the inferred high-quality AGN spectra excluded all photon-ALP couplings $g_\mathrm{a\gamma} > (6.3 - 8.0) \times 10^{-13} \ {\mathrm{GeV}}^{-1}$ at the $99.7\%$ level, respectively, based on the non-detection of spectral distortions attributed to photon-ALP inter-conversion along the cluster line-of-sight. Finally, we present the prospects of tightening current bounds on such ALPs by up to a factor of 10 with next-generation X-ray observatories such as $Athena$, $AXIS$ and $LEM$ given their improved spectral and spatial resolution and collecting area compared to current missions., Comment: 4 pages, 1 figure. Accepted for publication to Memorie della SAIt for the Proceedings of the European Astronomical Society 2022 (EAS 2022) Annual Meeting Symposium S3 "The Dark matter multi-messenger challenge"

We report the discovery of new lensed quasar candidates in the imaging data of the Hyper Suprime-Cam Subaru Strategic Program (HSC-SSP) DR4, covering $1\,310~{\rm deg}^2$ of the sky with seeing of $\approx0.6''$. In addition to two catalogs of MILLIQUAS and AllWISEAGN, which contain confirmed and candidate quasars, we preselect quasar sources using color cuts from the HSC ($grizy$) and unWISE ($W1+W2$) photometric data based on SDSS spectroscopic catalogs. We search for the presence of multiple point sources with similar color through the convolution of the Laplacian of the preselected quasar image cutouts with the Laplacian of the point spread function, resulting in a reduction of lens candidates from $1\,652\,329$ to $121\,511$ ($7.4\%$). After visual binary classification, we grade $6\,199$ ($0.4\%$) potential lenses on a scale of 0 to 3, with 3 indicating a lens and 0 indicating a non-lens. Finally we obtain 162 lens candidates with an average grade of $\geq2$, and among them, we successfully recover 18 known lenses. By fitting the light distribution and removing the known contaminants, we discover that 57 new systems contain at least two point sources and a galaxy in between, including 10 possible quadruply lensed quasars. This new sample of lens candidates exhibits a median separation of $1.26''$ and a magnitude limit of $i\approx22$. Spectroscopic or high-resolution imaging follow up on these newly discovered lensed quasar candidates will further allow their natures to be confirmed., Comment: 25 pages, 14 figures

In this work, we report on the growth of hexagonal boron nitride (hBN) crystals from an iron flux at atmospheric pressure and high temperature and demonstrate that (i) the entire sheet of hBN crystals can be detached from the metal in a single step using hydrochloric acid and that (ii) these hBN crystals allow the fabrication of high carrier mobility graphene devices. By combining spatially-resolved confocal Raman spectroscopy and electrical transport measurements, we confirm the excellent quality of these crystals for high-performance hBN-graphene-based van der Waals heterostructures. The full width at half maximum of the graphene Raman 2D peak is as low as 16 cm$^{-1}$, and the room temperature charge carrier mobilitiy is around 80000 cm$^2$/(Vs) at a carrier density 1$\times$10$^{12}$cm$^{-12}$. This is fully comparable with devices of similar dimensions fabricated using crystalline hBN synthesized by the high pressure and high temperature method. Finally, we show that for exfoliated high-quality hBN flakes with a thickness between 20 nm and 40 nm the line width of the hBN Raman peak, in contrast to the graphene 2D line width, is not useful for benchmarking hBN in high mobility graphene devices., Comment: 7 pages, 4 figures

The microlensing signal in the light curves of gravitationally lensed quasars can shed light on the dark matter (DM) composition in their lensing galaxies. Here, we investigate a sample of six lensed quasars from the most recent and best COSMOGRAIL observations: HE~1104$-$1805, HE~0435$-$1223, RX~J1131$-$1231, WFI~2033$-$4723, PG~1115$+$080, and J1206$+$4332, yielding a total of eight microlensing light curves, when combining independent image pairs and typically spanning ten years. We explore the microlensing signals to determine whether the standard assumptions on the stellar populations are sufficient to account for the amplitudes of the measured signals. We use the most detailed lens models to date from the H0LiCOW/TDCOSMO collaboration to generate simulated microlensing light curves. Finally, we propose a methodology based on the Kolmogorov-Smirnov test to verify whether the observed microlensing amplitudes in our data are compatible with the most standard scenario, whereby galaxies are composed of stars as compact bodies and smoothly distributed DM. Given our current sample, we show that the standard scenario cannot be rejected, in contrast with previous results by Hawkins (2020a), claiming that a population of stellar mass primordial black holes (PBHs) is necessary to explain the amplitude of the microlensing signals in lensed quasar light curves. We further estimate the number of microlensing light curves needed to distinguish between the standard scenario with stellar microlensing and a scenario that describes all the DM contained in galaxies in the form of compact objects such as PBHs, with a mean mass of $0.2M_{\odot}$. We find that about 900 microlensing curves from the Rubin Observatory will be sufficient to discriminate between the two extreme scenarios at a 95\% confidence level.