Search

Traditional reference-based metrics, such as BLEU and ROUGE, are less effective for assessing outputs from Large Language Models (LLMs) that produce highly creative or superior-quality text, or in situations where reference outputs are unavailable. While human evaluation remains an option, it is costly and difficult to scale. Recent work using LLMs as evaluators (LLM-as-a-judge) is promising, but trust and reliability remain a significant concern. Integrating human input is crucial to ensure criteria used to evaluate are aligned with the human's intent, and evaluations are robust and consistent. This paper presents a user study of a design exploration called EvaluLLM, that enables users to leverage LLMs as customizable judges, promoting human involvement to balance trust and cost-saving potential with caution. Through interviews with eight domain experts, we identified the need for assistance in developing effective evaluation criteria aligning the LLM-as-a-judge with practitioners' preferences and expectations. We offer findings and design recommendations to optimize human-assisted LLM-as-judge systems., Comment: 14 pages, 6 figures, Accepted for publication in ACL 2024 Workshop HuCLLM

We present near-infrared spectroscopy of ten weak emission-line quasars (WLQs) at redshifts of $z\sim2$, obtained with the Palomar 200-inch Hale Telescope. WLQs are an exceptional population of type 1 quasars that exhibit weak or no broad emission lines in the ultraviolet (e.g., the C IV $\lambda 1549$ line), and they display remarkable X-ray properties. We derive H$\beta$-based single-epoch virial black-hole masses (median value $\rm 1.7 \times 10^{9} M_{\odot}$) and Eddington ratios (median value $0.5)$ for our sources. We confirm the previous finding that WLQ H$\beta$ lines, as a major low-ionization line, are not significantly weak compared to typical quasars. The most prominent feature of the WLQ optical spectra is the universally weak/absent [O III] $\lambda 5007$ emission. They also display stronger optical Fe II emission than typical quasars. Our results favor the super-Eddington accretion scenario for WLQs, where the weak lines are a result of a soft ionizing continuum; the geometrically thick inner accretion disk and/or its associated outflow is responsible for obscuring the nuclear high-energy radiation and producing the soft ionizing continuum. We also report candidate extreme [O III] outflows (blueshifts of $\approx 500$ and $\rm 4900 km s^{-1}$) in one object., Comment: 19 pages, 8 figures, accepted for publication in ApJ

Cleavable metallic antiferromagnets may be of use for low-dissipation spintronic devices; however, few are currently known. Here we present orthorhombic MnRhBi3 as one such compound and present a thorough study of its physical properties. Exfoliation is demonstrated experimentally, and the cleavage energy and electronic structure are examined by density functional theory calculations. It is concluded that MnRhBi3 is a van der Waals layered material that cleaves easily between neighboring Bi layers, and that the Bi atoms have lone pairs extending into the van der Waals gaps. A series of four phase transitions are observed below room temperature, and neutron diffraction shows that at least two of the transitions involve the formation of antiferromagnetic order. Anomalous thermal expansion points to a crystallographic phase transition and/or strong magnetoelastic coupling. This work reveals a complex phase evolution in MnRhBi3 and establishes this cleavable antiferromagnetic metal as an interesting material for studying the interplay of structure, magnetism, and transport in the bulk and ultrathin limits as well as the role of lone pair electrons in interface chemistry and proximity effects in van der Waals heterostructures.

Detecting dynamic patterns of task-specific responses shared across heterogeneous datasets is an essential and challenging problem in many scientific applications in medical science and neuroscience. In our motivating example of rodent electrophysiological data, identifying the dynamical patterns in neuronal activity associated with ongoing cognitive demands and behavior is key to uncovering the neural mechanisms of memory. One of the greatest challenges in investigating a cross-subject biological process is that the systematic heterogeneity across individuals could significantly undermine the power of existing machine learning methods to identify the underlying biological dynamics. In addition, many technically challenging neurobiological experiments are conducted on only a handful of subjects where rich longitudinal data are available for each subject. The low sample sizes of such experiments could further reduce the power to detect common dynamic patterns among subjects. In this paper, we propose a novel heterogeneous data integration framework based on optimal transport to extract shared patterns in complex biological processes. The key advantages of the proposed method are that it can increase discriminating power in identifying common patterns by reducing heterogeneity unrelated to the signal by aligning the extracted latent spatiotemporal information across subjects. Our approach is effective even with a small number of subjects, and does not require auxiliary matching information for the alignment. In particular, our method can align longitudinal data across heterogeneous subjects in a common latent space to capture the dynamics of shared patterns while utilizing temporal dependency within subjects.

Future X-ray astrophysics missions will survey large areas of the sky with unparalleled sensitivity, enabled by lightweight, high-resolution optics. These optics inherently produce curved focal surfaces with radii as small as 2 m, requiring a large area detector system that closely conforms to the curved focal surface. We have embarked on a project using a curved charge-coupled device (CCD) detector technology developed at MIT Lincoln Laboratory to provide large-format, curved detectors for such missions, improving performance and simplifying design. We present the current status of this work, which aims to curve back-illuminated, large-format (5 cm x 4 cm) CCDs to 2.5-m radius and confirm X-ray performance. We detail the design of fixtures and the curving process, and present intial results on curving bare silicon samples and monitor devices and characterizing the surface geometric accuracy. The tests meet our accuracy requirement of <5 $\mu$m RMS surface non-conformance for samples of similar thickness to the functional detectors. We finally show X-ray performance measurements of planar CCDs that will serve as a baseline to evaluate the curved detectors. The detectors exhibit low noise, good charge-transfer efficiency, and excellent, uniform spectroscopic performance, including in the important soft X-ray band., Comment: 18 pages, 13 figures, submitted to the Proceedings of SPIE, Astronomical Telescopes + Instrumentation 2024

At optical/ultraviolet energies, blazars display an underlying thermal (unpolarized) contribution from the accretion disc, torus and line emitting regions, diluting the polarized emission from the jet-component. Optical polarimetry can be used to disentangle the thermal and non-thermal components, and place constraints on the particle populations and acceleration mechanisms responsible for the non-thermal emission. We present the results of a linear optical spectropolarimetric observing campaign of 18 blazars (6 BLLs and 12 FSRQs) undertaken with the Southern African Large Telescope between 2016 and 2022. This was done to observe these systems during flaring states, as well as long term monitoring of PKS1510-089, AP Lib and PKS 1034-293. The observations traced the frequency dependence of the degree and angle of polarization, as well as changes in the spectral line strengths. We investigated possible correlations between the polarization and other observed characteristics for the sources. While an indication of correlation was found between the frequency dependence and the average level of polarization for some sources, a correlation was not found for the population as a whole. These results highlight that continuous observations and in-depth modelling of polarization and its frequency dependence is required to obtain a more holistic view of TeV blazars., Comment: 17 pages, 21 figures, 3 tables, accepted for publication in MNRAS

Determining whether two STRIPS planning instances are isomorphic is the simplest form of comparison between planning instances. It is also a particular case of the problem concerned with finding an isomorphism between a planning instance $P$ and a sub-instance of another instance $P_0$ . One application of such a mapping is to efficiently produce a compiled form containing all solutions to P from a compiled form containing all solutions to $P_0$. We also introduce the notion of embedding from an instance $P$ to another instance $P_0$, which allows us to deduce that $P_0$ has no solution-plan if $P$ is unsolvable. In this paper, we study the complexity of these problems. We show that the first is GI-complete, and can thus be solved, in theory, in quasi-polynomial time. While we prove the remaining problems to be NP-complete, we propose an algorithm to build an isomorphism, when possible. We report extensive experimental trials on benchmark problems which demonstrate conclusively that applying constraint propagation in preprocessing can greatly improve the efficiency of a SAT solver.

In this project, we investigate the accuracy of forecasting intraday and daily trading volume of the exchange-traded fund SPY. The ability to forecast volume over varying time intervals with high accuracy is a critical element to many trading strategies. After performing exploratory data analysis on intraday and daily SPY data we identify three methods for our analysis: ARIMA and ARIMAX models, with or without seasonality, as well as a Frequency Domain Process Representation. To evaluate predictive power of our models, we use mean squared error, mean absolute percentage error, and volume weighted average price (VWAP) tracking error. All models for both intraday and daily data output strong VWAP predictions in comparison to the VWAP estimates produced by naive baseline methodologies. In both cases volume is most accurately forecasted using ARIMA models with exogenous variables in the form of technical indicators, with intraday incorporating a seasonal component and daily not.

Integrating language models into robotic exploration frameworks improves performance in unmapped environments by providing the ability to reason over semantic groundings, contextual cues, and temporal states. The proposed method employs large language models (GPT-3.5 and Claude Haiku) to reason over these cues and express that reasoning in terms of natural language, which can be used to inform future states. We are motivated by the context of search-and-rescue applications where efficient exploration is critical. We find that by leveraging natural language, semantics, and tracking temporal states, the proposed method greatly reduces exploration path distance and further exposes the need for environment-dependent heuristics. Moreover, the method is highly robust to a variety of environments and noisy vision detections, as shown with a 100% success rate in a series of comprehensive experiments across three different environments conducted in a custom simulation pipeline operating in Unreal Engine., Comment: 9 pages (22 with references and appendix), 12 figures (including 6 in the appendix), 1 table

We present a detailed benchmarking of different treatments of the QCD evolution of unpolarized PDFs at approximate $N^3LO$ ($aN^3LO$) order in the QCD coupling. Namely, the implementations in the public $aN^3LO$ releases of the MSHT and NNPDF global PDF fitters, as well as that of the theoretical FHMRUVV collaboration are compared. This follows the same procedure as in previous benchmarking exercises at lower order, that is by considering the impact of this evolution on a set of simple toy PDFs.

Atomic vapour cells are an indispensable tool for quantum technologies (QT), but potential improvements are limited by the capacities of conventional manufacturing methods. Using an additive manufacturing (AM) technique - vat polymerisation by digital light processing - we demonstrate, for the first time, a 3D-printed glass vapour cell. The exploitation of AM capacities allows intricate internal architectures, overprinting of 2D optoelectronical materials to create integrated sensors and surface functionalisation, while also showing the ability to tailor the optical properties of the AM glass by in-situ growth of gold nanoparticles. The produced cells achieve ultra-high vacuum of $2 \times 10^{-9}$ mbar and enable Doppler-free spectroscopy; we demonstrate laser frequency stabilisation as a QT application. These results highlight the transformative role that AM can play for QT in enabling compact, optimised and integrated multi-material components and devices.

Language models (LMs) have been shown to behave unexpectedly post-deployment. For example, new jailbreaks continually arise, allowing model misuse, despite extensive red-teaming and adversarial training from developers. Given most model queries are unproblematic and frequent retraining results in unstable user experience, methods for mitigation of worst-case behavior should be targeted. One such method is classifying inputs as potentially problematic, then selectively applying steering vectors on these problematic inputs, i.e. adding particular vectors to model hidden states. However, steering vectors can also negatively affect model performance, which will be an issue on cases where the classifier was incorrect. We present KL-then-steer (KTS), a technique that decreases the side effects of steering while retaining its benefits, by first training a model to minimize Kullback-Leibler (KL) divergence between a steered and unsteered model on benign inputs, then steering the model that has undergone this training. Our best method prevents 44% of jailbreak attacks compared to the original Llama-2-chat-7B model while maintaining helpfulness (as measured by MT-Bench) on benign requests almost on par with the original LM. To demonstrate the generality and transferability of our method beyond jailbreaks, we show that our KTS model can be steered to reduce bias towards user-suggested answers on TruthfulQA. Code is available: https://github.com/AsaCooperStickland/kl-then-steer.

A significant challenge in measurements of neutrino oscillations is reconstructing the incoming neutrino energies. While modern fully-active tracking calorimeters such as liquid argon time projection chambers in principle allow the measurement of all final state particles above some detection threshold, undetected neutrons remain a considerable source of missing energy with little to no data constraining their production rates and kinematics. We present the first demonstration of tagging neutrino-induced neutrons in liquid argon time projection chambers using secondary protons emitted from neutron-argon interactions in the MicroBooNE detector. We describe the method developed to identify neutrino-induced neutrons and demonstrate its performance using neutrons produced in muon-neutrino charged current interactions. The method is validated using a small subset of MicroBooNE's total dataset. The selection yields a sample with $60\%$ of selected tracks corresponding to neutron-induced secondary protons.

JWST has revealed a population of compact and extremely red galaxies at $z>4$, which likely host active galactic nuclei (AGN). We present a sample of 434 ``little red dots'' (LRDs), selected from the 0.54 deg$^2$ COSMOS-Web survey. We fit galaxy and AGN SED models to derive redshifts and physical properties; the sample spans $z\sim5$-$9$ after removing brown dwarf contaminants. We consider two extreme physical scenarios: either LRDs are all AGN, and their continuum emission is dominated by the accretion disk, or they are all compact star-forming galaxies, and their continuum is dominated by stars. If LRDs are AGN-dominated, our sample exhibits bolometric luminosities $\sim10^{45-47}$ erg\,s$^{-1}$, spanning the gap between JWST AGN in the literature and bright, rare quasars. We derive a bolometric luminosity function (LF) $\sim100$ times the (UV-selected) quasar LF, implying a non-evolving black hole accretion density of $\sim10^{-4}$ M$_\odot$ yr$^{-1}$ Mpc$^{-3}$ from $z\sim2$-$9$. By contrast, if LRDs are dominated by star formation, we derive stellar masses $\sim10^{8.5-10}\,M_\odot$. MIRI/F770W is key to deriving accurate stellar masses; without it, we derive a mass function inconsistent with $\Lambda$CDM. The median stellar mass profile is broadly consistent with the maximal stellar mass surface densities seen in the nearby universe, though the most massive $\sim50$\% of objects exceed this limit, requiring substantial AGN contribution to the continuum. Nevertheless, stacking all available X-ray, mid-IR, far-IR/sub-mm, and radio data yields non-detections. Whether dominated by dusty AGN, compact star-formation, or both, the high masses/luminosities and remarkable abundance of LRDs implies a dominant mode of early galaxy/SMBH growth., Comment: 27 pages, 13 figures. Submitted to ApJ, comments welcome! Data access at https://github.com/hollisakins/akins24_cw

We present a deep learning-based method for estimating the neutrino energy of charged-current neutrino-argon interactions. We employ a recurrent neural network (RNN) architecture for neutrino energy estimation in the MicroBooNE experiment, utilizing liquid argon time projection chamber (LArTPC) detector technology. Traditional energy estimation approaches in LArTPCs, which largely rely on reconstructing and summing visible energies, often experience sizable biases and resolution smearing because of the complex nature of neutrino interactions and the detector response. The estimation of neutrino energy can be improved after considering the kinematics information of reconstructed final-state particles. Utilizing kinematic information of reconstructed particles, the deep learning-based approach shows improved resolution and reduced bias for the muon neutrino Monte Carlo simulation sample compared to the traditional approach. In order to address the common concern about the effectiveness of this method on experimental data, the RNN-based energy estimator is further examined and validated with dedicated data-simulation consistency tests using MicroBooNE data. We also assess its potential impact on a neutrino oscillation study after accounting for all statistical and systematic uncertainties and show that it enhances physics sensitivity. This method has good potential to improve the performance of other physics analyses.

Tails used as inertial appendages induce body rotations of animals and robots, a phenomenon that is governed largely by the ratio of the body and tail moments of inertia. However, vertebrate tails have more degrees of freedom (e.g., number of joints, rotational axes) than most current theoretical models and robotic tails. To understand how morphology affects inertial appendage function, we developed an optimization-based approach that finds the maximally effective tail trajectory and measures error from a target trajectory. For tails of equal total length and mass, increasing the number of equal-length joints increased the complexity of maximally effective tail motions. When we optimized the relative lengths of tail bones while keeping the total tail length, mass, and number of joints the same, this optimization-based approach found that the lengths match the pattern found in the tail bones of mammals specialized for inertial maneuvering. In both experiments, adding joints enhanced the performance of the inertial appendage, but with diminishing returns, largely due to the total control effort constraint. This optimization-based simulation can compare the maximum performance of diverse inertial appendages that dynamically vary in moment of inertia in 3D space, predict inertial capabilities from skeletal data, and inform the design of robotic inertial appendages.

The diffusionless martensitic phase transition from a hexagonal close-packed (hcp) arrangement to the face-centered close-packed (fcc) and subsequently the body-centered cubic (bcc) lattice is discussed for a Lennard-Jones solid. The associated lattice vectors to construct the underlying bi-lattice for a Burgers-Bain-type of transformation require a minimum of four parameters $(a,\alpha,\beta,\gamma=c/a)$ describing, beside the change in the base lattice parameters $a$ and $c$, the shear force acting on the hexagonal base plane through the parameter $\alpha$, and the sliding force of the middle layer in the original AB hexagonal packing arrangement through the parameter $\beta$. By optimizing the lattice parameters $a$ and $\beta$ for a $(n,m)$-Lennard-Jones potential, we obtain a simple two-dimensional picture for the complete Burgers-Bain-type hcp$\leftrightarrow$fcc$\leftrightarrow$bcc phase transition. From the generalized lattice vectors we were able to construct the corresponding lattice sums in terms of inverse power potentials applying fast converging Bessel function expansions using a Terras decomposition of the Epstein zeta function combined with a Van der Hoff-Benson expansion for the lattice sums. This allows the cohesive energy to be determined to computer precision for a Lennard-Jones solid. For six different combinations of $(n,m)$-Lennard-Jones potentials the energy $(\alpha,\gamma)$ hypersurface was then mapped out and studied in more detail. We show that for a Lennard-Jones model the minimum energy path is found to be a two-step hcp$\rightarrow$fcc$\rightarrow$bcc transition process. The lowest transition state in each case can be regarded as an upper limit to a hypothetical true minimum energy path out of the many possibilities in a hcp$\leftrightarrow$fcc$\leftrightarrow$bcc phase transition.

To develop rigorous knowledge about ML models -- and the systems in which they are embedded -- we need reliable measurements. But reliable measurement is fundamentally challenging, and touches on issues of reproducibility, scalability, uncertainty quantification, epistemology, and more. This dissertation addresses criteria needed to take reliability seriously: both criteria for designing meaningful metrics, and for methodologies that ensure that we can dependably and efficiently measure these metrics at scale and in practice. In doing so, this dissertation articulates a research vision for a new field of scholarship at the intersection of machine learning, law, and policy. Within this frame, we cover topics that fit under three different themes: (1) quantifying and mitigating sources of arbitrariness in ML, (2) taming randomness in uncertainty estimation and optimization algorithms, in order to achieve scalability without sacrificing reliability, and (3) providing methods for evaluating generative-AI systems, with specific focuses on quantifying memorization in language models and training latent diffusion models on open-licensed data. By making contributions in these three themes, this dissertation serves as an empirical proof by example that research on reliable measurement for machine learning is intimately and inescapably bound up with research in law and policy. These different disciplines pose similar research questions about reliable measurement in machine learning. They are, in fact, two complementary sides of the same research vision, which, broadly construed, aims to construct machine-learning systems that cohere with broader societal values., Comment: Ph.D. Dissertation

The Dark Energy Spectroscopic Instrument Milky Way Survey (DESI MWS) will explore the assembly history of the Milky Way by characterising remnants of ancient dwarf galaxy accretion events and improving constraints on the distribution of dark matter in the outer halo. We present mock catalogues that reproduce the selection criteria of MWS and the format of the final MWS data set. These catalogues can be used to test methods for quantifying the properties of stellar halo substructure and reconstructing the Milky Way's accretion history with the MWS data, including the effects of halo-to-halo variance. The mock catalogues are based on a phase-space kernel expansion technique applied to star particles in the Auriga suite of six high-resolution $\Lambda$CDM magneto-hydrodynamic zoom-in simulations. They include photometric properties (and associated errors) used in DESI target selection and the outputs of the MWS spectral analysis pipeline (radial velocity, metallicity, surface gravity, and temperature). They also include information from the underlying simulation, such as the total gravitational potential and information on the progenitors of accreted halo stars. We discuss how the subset of halo stars observable by MWS in these simulations corresponds to their true content and properties. These mock Milky Ways have rich accretion histories, resulting in a large number of substructures that span the whole stellar halo out to large distances and have substantial overlap in the space of orbital energy and angular momentum., Comment: Accepted for publication in MNRAS, 31 pages, 27 figues, 7 tables. The mock catalogues are available at https://data.desi.lbl.gov/public/papers/mws/auridesi/v1

Self-supervised learning (SSL) representations from massively multilingual models offer a promising solution for low-resource language speech tasks. Despite advancements, language adaptation in TTS systems remains an open problem. This paper explores the language adaptation capability of ZMM-TTS, a recent SSL-based multilingual TTS system proposed in our previous work. We conducted experiments on 12 languages using limited data with various fine-tuning configurations. We demonstrate that the similarity in phonetics between the pre-training and target languages, as well as the language category, affects the target language's adaptation performance. Additionally, we find that the fine-tuning dataset size and number of speakers influence adaptability. Surprisingly, we also observed that using paired data for fine-tuning is not always optimal compared to audio-only data. Beyond speech intelligibility, our analysis covers speaker similarity, language identification, and predicted MOS., Comment: Accepted to Interspeech 2024

This paper proposes a speech synthesis system that allows users to specify and control the acoustic characteristics of a speaker by means of prompts describing the speaker's traits of synthesized speech. Unlike previous approaches, our method utilizes listener impressions to construct prompts, which are easier to collect and align more naturally with everyday descriptions of speaker traits. We adopt the Low-rank Adaptation (LoRA) technique to swiftly tailor a pre-trained language model to our needs, facilitating the extraction of speaker-related traits from the prompt text. Besides, different from other prompt-driven text-to-speech (TTS) systems, we separate the prompt-to-speaker module from the multi-speaker TTS system, enhancing system flexibility and compatibility with various pre-trained multi-speaker TTS systems. Moreover, for the prompt-to-speaker characteristic module, we also compared the discriminative method and flow-matching based generative method and we found that combining both methods can help the system simultaneously capture speaker-related information from prompts better and generate speech with higher fidelity., Comment: Accepted for presentation at Interspeech 2024 (with more analysis in the final Appendix part)

The ability to automatically and robustly self-verify periodicity present in time-series astronomical data is becoming more important as data sets rapidly increase in size. The age of large astronomical surveys has rendered manual inspection of time-series data less practical. Previous efforts in generating a false alarm probability to verify the periodicity of stars have been aimed towards the analysis of a constructed periodogram. However, these methods feature correlations with features that do not pertain to periodicity, such as light curve shape, slow trends and stochastic variability. The common assumption that photometric errors are Gaussian and well determined is also a limitation of analytic methods. We present a novel machine learning based technique which directly analyses the phase folded light curve for its false alarm probability. We show that the results of this method are largely insensitive to the shape of the light curve, and we establish minimum values for the number of data points and the amplitude to noise ratio.

Pretraining for partial differential equation (PDE) modeling has recently shown promise in scaling neural operators across datasets to improve generalizability and performance. Despite these advances, our understanding of how pretraining affects neural operators is still limited; studies generally propose tailored architectures and datasets that make it challenging to compare or examine different pretraining frameworks. To address this, we compare various pretraining methods without optimizing architecture choices to characterize pretraining dynamics on different models and datasets as well as to understand its scaling and generalization behavior. We find that pretraining is highly dependent on model and dataset choices, but in general transfer learning or physics-based pretraining strategies work best. In addition, pretraining performance can be further improved by using data augmentations. Lastly, pretraining can be additionally beneficial when fine-tuning in scarce data regimes or when generalizing to downstream data similar to the pretraining distribution. Through providing insights into pretraining neural operators for physics prediction, we hope to motivate future work in developing and evaluating pretraining methods for PDEs., Comment: 29 pages, 5 figures

Effective teaching relies on knowing what students know-or think they know. Revealing student thinking is challenging. Often used because of their ease of grading, even the best multiple choice (MC) tests, those using research based distractors (wrong answers) are intrinsically limited in the insights they provide due to two factors. When distractors do not reflect student beliefs they can be ignored, increasing the likelihood that the correct answer will be chosen by chance. Moreover, making the correct choice does not guarantee that the student understands why it is correct. To address these limitations, we recommend asking students to explain why they chose their answer, and why "wrong" choices are wrong. Using a discipline-trained artificial intelligence-based bot it is possible to analyze their explanations, identifying the concepts and scientific principles that maybe missing or misapplied. The bot also makes suggestions for how instructors can use these data to better guide student thinking. In a small "proof of concept" study, we tested this approach using questions from the Biology Concepts Instrument (BCI). The result was rapid, informative, and provided actionable feedback on student thinking. It appears that the use of AI addresses the weaknesses of conventional MC test. It seems likely that incorporating AI-analyzed formative assessments will lead to improved overall learning outcomes.

This paper defines Spoof Diarization as a novel task in the Partial Spoof (PS) scenario. It aims to determine what spoofed when, which includes not only locating spoof regions but also clustering them according to different spoofing methods. As a pioneering study in spoof diarization, we focus on defining the task, establishing evaluation metrics, and proposing a benchmark model, namely the Countermeasure-Condition Clustering (3C) model. Utilizing this model, we first explore how to effectively train countermeasures to support spoof diarization using three labeling schemes. We then utilize spoof localization predictions to enhance the diarization performance. This first study reveals the high complexity of the task, even in restricted scenarios where only a single speaker per audio file and an oracle number of spoofing methods are considered. Our code is available at https://github.com/nii-yamagishilab/PartialSpoof., Comment: Accepted to Interspeech 2024

Motivated by the recent detection of ultra-long period radio transients, we investigate new models of coherent radio emission via low-altitude electron-positron pair production in neutron stars beyond rotationally-powered curvature radiation deathlines. We find that plastic motion (akin to 'continental drift') and qualitatively similar thermoelectric action by temperature gradients in the crusts of slowly rotating, highly magnetized neutron stars could impart mild local magnetospheric twists. Regardless of which mechanism drives twists, we find that particle acceleration initiates pair cascades across charge-starved gaps above a mild critical twist. Cascades are initiated via resonant inverse-Compton scattered photons or curvature radiation, and may produce broadband coherent radio emission. We compute the pair luminosity (maximum allowed radio luminosity) for these two channels, and derive deathlines and 'active zones' in $P-\dot{P}$ space from a variety of considerations. We find these twist-initiated pair cascades only occur for magnetar-like field strengths $B \gtrsim 10^{14}$ G and long periods: $P_{\rm RICS} \gtrsim 120 \; (T/10^{6.5} {\rm K})^{-5} \, {\rm sec}$ and $P_{\rm curv} \gtrsim 150 \; ({\rm v_{\rm pl}}/10^{3} {\, \rm cm \, yr^{-1}})^{-7/6} \, {\rm sec}$. Using a simplified geometric model, we find that plastic motion or thermoelectrically driven twists might naturally reproduce the observed luminosities, timescales, and timing signatures. We further derive 'active zones' in which rotationally-powered pair creation occurs via resonantly scattered photons, beyond standard curvature deathlines for pulsars. All cascades are generically accompanied by simultaneous (non-)thermal X-ray/UV counterparts which might be detectable with current instrumentation., Comment: Replaced with accepted version

The azimuthal correlation angle, $\Delta\phi$, between the scattered lepton and the leading jet in deep inelastic $e^{\pm}p$ scattering at HERA has been studied using data collected with the ZEUS detector at a centre-of-mass energy of $\sqrt{s} = 318 \;\mathrm{GeV}$, corresponding to an integrated luminosity of $326 \;\mathrm{pb}^{-1}$. A measurement of jet cross sections in the laboratory frame was made in a fiducial region corresponding to photon virtuality $10 \;\mathrm{GeV}^2 < Q^2 < 350 \;\mathrm{GeV}^2$, inelasticity $0.04 < y < 0.7$, outgoing lepton energy $E_e > 10 \;\mathrm{GeV}$, lepton polar angle $140^\circ < \theta_e < 180^\circ$, jet transverse momentum $2.5 \;\mathrm{GeV} < p_\mathrm{T,jet} < 30 \;\mathrm{GeV}$, and jet pseudorapidity $-1.5 < \eta_\mathrm{jet} < 1.8$. Jets were reconstructed using the $k_\mathrm{T}$ algorithm with the radius parameter $R = 1$. The leading jet in an event is defined as the jet that carries the highest $p_\mathrm{T,jet}$. Differential cross sections, $d\sigma/d\Delta\phi$, were measured as a function of the azimuthal correlation angle in various ranges of leading-jet transverse momentum, photon virtuality and jet multiplicity. Perturbative calculations at $\mathcal{O}(\alpha_{s}^2)$ accuracy successfully describe the data within the fiducial region, although a lower level of agreement is observed near $\Delta\phi \rightarrow \pi$ for events with high jet multiplicity, due to limitations of the perturbative approach in describing soft phenomena in QCD. The data are equally well described by Monte Carlo predictions that supplement leading-order matrix elements with parton showering.

This report presents a short summary of the activities of the "Standard Model" working group for the "Physics at TeV Colliders" workshop (Les Houches, France, 12-30 June, 2023)., Comment: Proceedings of the Standard Model Working Group of the 2023 Les Houches Workshop, Physics at TeV Colliders, Les Houches 12-30 June 2023. 48 pages

Tabular data are omnipresent in various sectors of industries. Neural networks for tabular data such as TabNet have been proposed to make predictions while leveraging the attention mechanism for interpretability. However, the inferred attention masks are often dense, making it challenging to come up with rationales about the predictive signal. To remedy this, we propose InterpreTabNet, a variant of the TabNet model that models the attention mechanism as a latent variable sampled from a Gumbel-Softmax distribution. This enables us to regularize the model to learn distinct concepts in the attention masks via a KL Divergence regularizer. It prevents overlapping feature selection by promoting sparsity which maximizes the model's efficacy and improves interpretability to determine the important features when predicting the outcome. To assist in the interpretation of feature interdependencies from our model, we employ a large language model (GPT-4) and use prompt engineering to map from the learned feature mask onto natural language text describing the learned signal. Through comprehensive experiments on real-world datasets, we demonstrate that InterpreTabNet outperforms previous methods for interpreting tabular data while attaining competitive accuracy., Comment: ICML 2024 Spotlight