Search

Proton therapy can achieve a highly targeted treatment by utilising the advantageous dosimetric characteristics of the Bragg Peak. Protons traversing through a material will deposit their maximum energy at the Bragg Peak through ionisation and other interactions, transferring minimal excess dose to surrounding tissue and organs. This rate of energy loss is also quantified by the linear energy transfer (LET), which is indicative of radiation quality and radiobiological effects. However it is a challenging physical quantity to measure, as characterisation of radiation fields and the impact of LET on treatment requires advanced tools and technology. The MiniPIX-Timepix is a miniaturised, hybrid semiconductor pixel detector capable of high resolution spectrometric tracking, enabling wide-range detection of the deposited energy, position and direction of single particles. Experimental measurements were performed at a clinical facility, the Clatterbridge Cancer Centre which houses a 60 MeV ocular proton therapy beamline. A realistic end-to-end model of the facility was developed in the Monte Carlo code TOPAS (TOol for PArticle Simulation) and was used to simulate the experimental conditions. The detector was held at 45$^{\circ}$ and 60$^{\circ}$ perpendicular to the beam, and placed downstream of various thickness Polymethyl methacrylate (PMMA) blocks to acquire data along the dose deposition depth. Empirical cluster data providing track length and the energy deposition distributions were used to obtain the LET spectra. The determined values for the LET in silicon and dose averaged LET across the BP show general agreement with simulated results, supporting the applicability of the TOPAS CCC model. This work explores the capability of the MiniPIX detector to measure physical quantities to resolve the LET, and discusses experimental considerations and further possibilities., Comment: Prepared for submission to JINST, 20 pages, 10 figures, 2 tables

Proteins are composed of linear chains of amino acids that fold into complex three-dimensional structures. A general feature of folded proteins is that they are compact with a radius of gyration $R_g(N) \sim N^{\nu}$ that obeys power-law scaling with the number of amino acids $N$ in each protein and $\nu \sim 1/3$. In this study, we investigate the {\it internal} scaling of the radius of gyration $R_g(n)$ versus the chemical separation $n$ between amino acids for all subchains of length $n$. We show that for globular proteins $R_g(n) \sim n^{\nu_{1,2}}$ with a larger exponent $\nu_1 > 1/3$ for small $n$ and a smaller exponent $\nu_{2} < 1/3$ for large $n$, such that $R_g(N) \propto N^\nu$. To describe this scaling behavior for $R_g(n)$, we carry out folding simulations for a series of coarse-grained models for proteins beginning with the freely-jointed and freely-rotating chain models composed of spherical monomers and varying degrees of stereochemical constraints. We show that a minimal model, which coarse-grains amino acids into a single spherical backbone atom and one variable-sized spherical side-chain atom, and enforces bend- and dihedral-angle constraints, can recapitulate $R_g(n)$ for x-ray crystal structures of globular proteins. In addition, this model predicts the correct average packing fraction and size of the hydrophobic core, which are two key physical features that can be used to distinguish between computational `decoys' and correctly folded proteins in protein design applications., Comment: 8 pages, 6 figures

Though exponential distributions are ubiquitous in statistical physics and related computational models, directly sampling them from device behavior is rarely done. The superparamagnetic tunnel junction (SMTJ), a key device in probabilistic computing, is known to naturally exhibit exponentially distributed temporal switching dynamics. To sample an exponential distribution with an SMTJ, we need to measure it in the time domain, which is challenging with traditional techniques that focus on sampling the instantaneous state of the device. In this work, we leverage a temporal encoding scheme, where information is encoded in the time at which the device switches between its resistance states. We then develop a circuit element known as a probabilistic delay cell that applies an electrical current step to an SMTJ and a temporal measurement circuit that measures the timing of the first switching event. Repeated experiments confirm that these times are exponentially distributed. Temporal processing methods then allow us to digitally compute with these exponentially distributed probabilistic delay cells. We describe how to use these circuits in a Metropolis-Hastings stepper and in a weighted random sampler, both of which are computationally intensive applications that benefit from the efficient generation of exponentially distributed random numbers.

Extreme events over large spatial domains may exhibit highly heterogeneous tail dependence characteristics, yet most existing spatial extremes models yield only one dependence class over the entire spatial domain. To accurately characterize "data-level dependence'' in analysis of extreme events, we propose a mixture model that achieves flexible dependence properties and allows high-dimensional inference for extremes of spatial processes. We modify the popular random scale construction that multiplies a Gaussian random field by a single radial variable; we allow the radial variable to vary smoothly across space and add non-stationarity to the Gaussian process. As the level of extremeness increases, this single model exhibits both asymptotic independence at long ranges and either asymptotic dependence or independence at short ranges. We make joint inference on the dependence model and a marginal model using a copula approach within a Bayesian hierarchical model. Three different simulation scenarios show close to nominal frequentist coverage rates. Lastly, we apply the model to a dataset of extreme summertime precipitation over the central United States. We find that the joint tail of precipitation exhibits non-stationary dependence structure that cannot be captured by limiting extreme value models or current state-of-the-art sub-asymptotic models.

Video and audio are closely correlated modalities that humans naturally perceive together. While recent advancements have enabled the generation of audio or video from text, producing both modalities simultaneously still typically relies on either a cascaded process or multi-modal contrastive encoders. These approaches, however, often lead to suboptimal results due to inherent information losses during inference and conditioning. In this paper, we introduce SyncFlow, a system that is capable of simultaneously generating temporally synchronized audio and video from text. The core of SyncFlow is the proposed dual-diffusion-transformer (d-DiT) architecture, which enables joint video and audio modelling with proper information fusion. To efficiently manage the computational cost of joint audio and video modelling, SyncFlow utilizes a multi-stage training strategy that separates video and audio learning before joint fine-tuning. Our empirical evaluations demonstrate that SyncFlow produces audio and video outputs that are more correlated than baseline methods with significantly enhanced audio quality and audio-visual correspondence. Moreover, we demonstrate strong zero-shot capabilities of SyncFlow, including zero-shot video-to-audio generation and adaptation to novel video resolutions without further training.

Spatial omics assays allow for the molecular characterisation of cells in their spatial context. Notably, the two main technological streams, imaging-based and high-throughput sequencing-based, can give rise to very different data modalities. The characteristics of the two data types are well known in adjacent fields such as spatial statistics as point patterns and lattice data, and there is a wide range of tools available. This paper discusses the application of spatial statistics to spatially-resolved omics data and in particular, discusses various advantages, challenges, and nuances. This work is accompanied by a vignette, pasta, that showcases the usefulness of spatial statistics in biology using several R packages.

With the emergence of audio-language models, constructing large-scale paired audio-language datasets has become essential yet challenging for model development, primarily due to the time-intensive and labour-heavy demands involved. While large language models (LLMs) have improved the efficiency of synthetic audio caption generation, current approaches struggle to effectively extract and incorporate detailed audio information. In this paper, we propose an automated pipeline that integrates audio-language models for fine-grained content extraction, LLMs for synthetic caption generation, and a contrastive language-audio pretraining (CLAP) model-based refinement process to improve the quality of captions. Specifically, we employ prompt chaining techniques in the content extraction stage to obtain accurate and fine-grained audio information, while we use the refinement process to mitigate potential hallucinations in the generated captions. Leveraging the AudioSet dataset and the proposed approach, we create AudioSetCaps, a dataset comprising 1.9 million audio-caption pairs, the largest audio-caption dataset at the time of writing. The models trained with AudioSetCaps achieve state-of-the-art performance on audio-text retrieval with R@1 scores of 46.3% for text-to-audio and 59.7% for audio-to-text retrieval and automated audio captioning with the CIDEr score of 84.8. As our approach has shown promising results with AudioSetCaps, we create another dataset containing 4.1 million synthetic audio-language pairs based on the Youtube-8M and VGGSound datasets. To facilitate research in audio-language learning, we have made our pipeline, datasets with 6 million audio-language pairs, and pre-trained models publicly available at https://github.com/JishengBai/AudioSetCaps.

Physical vapor deposition can prepare organic glasses with high kinetic stability. When heated, these glassy solids slowly transform into the supercooled liquid in a process known as rejuvenation. In this study, we anneal vapor-deposited glasses of methyl-m-toluate (MMT) for six hours at 0.98Tg to observe rejuvenation using dielectric spectroscopy. Glasses of moderate stability exhibited partial or full rejuvenation in six hours. For highly stable glasses, prepared at substrate temperatures of 0.85Tg and 0.80Tg, the six-hour annealing time is ~2% of the estimated transformation time, and no change in the onset temperature for the {\alpha} relaxation process was observed, as expected. Surprisingly, for these highly stable glasses, annealing resulted in significant increases in the storage component of the dielectric susceptibility, without corresponding increases in the loss component. These changes are interpreted to indicate that short-term annealing rejuvenates a high frequency relaxation (e.g., the boson peak) within the stable glass. We compare these results to computer simulations of the rejuvenation of highly stable glasses generated by the swap Monte Carlo algorithm. The in silico glasses, in contrast to the experiment, show no evidence of rejuvenation within the stable glass at times shorter than the alpha relaxation process.

This paper initiates the study of the Einstein equation on homogeneous supermanifolds. First, we produce explicit curvature formulas for graded Riemannian metrics on these spaces. Next, we present a construction of homogeneous supermanifolds by means of Dynkin diagrams, resembling the construction of generalised flag manifolds in classical (non-super) theory. We describe the Einstein metrics on several classes of spaces obtained through this approach. Our results provide examples of compact homogeneous supermanifolds on which the Einstein equation has no solutions, discrete families of solutions, and continuous families of Ricci-flat solutions among invariant metrics. These examples demonstrate that the finiteness conjecture from classical homogeneous geometry fails on supermanifolds, and challenge the intuition furnished by Bochner's vanishing theorem., Comment: 46 pages

Sound event localization and detection (SELD) has seen substantial advancements through learning-based methods. These systems, typically trained from scratch on specific datasets, have shown considerable generalization capabilities. Recently, deep neural networks trained on large-scale datasets have achieved remarkable success in the sound event classification (SEC) field, prompting an open question of whether these advancements can be extended to develop general-purpose SELD models. In this paper, leveraging the power of pre-trained SEC models, we propose pre-trained SELD networks (PSELDNets) on large-scale synthetic datasets. These synthetic datasets, generated by convolving sound events with simulated spatial room impulse responses (SRIRs), contain 1,167 hours of audio clips with an ontology of 170 sound classes. These PSELDNets are transferred to downstream SELD tasks. When we adapt PSELDNets to specific scenarios, particularly in low-resource data cases, we introduce a data-efficient fine-tuning method, AdapterBit. PSELDNets are evaluated on a synthetic-test-set using collected SRIRs from TAU Spatial Room Impulse Response Database (TAU-SRIR DB) and achieve satisfactory performance. We also conduct our experiments to validate the transferability of PSELDNets to three publicly available datasets and our own collected audio recordings. Results demonstrate that PSELDNets surpass state-of-the-art systems across all publicly available datasets. Given the need for direction-of-arrival estimation, SELD generally relies on sufficient multi-channel audio clips. However, incorporating the AdapterBit, PSELDNets show more efficient adaptability to various tasks using minimal multi-channel or even just monophonic audio clips, outperforming the traditional fine-tuning approaches., Comment: 13 pages, 9 figures. The code is available at https://github.com/Jinbo-Hu/PSELDNets

The Gaussian process (GP) is a widely used probabilistic machine learning method for stochastic function approximation, stochastic modeling, and analyzing real-world measurements of nonlinear processes. Unlike many other machine learning methods, GPs include an implicit characterization of uncertainty, making them extremely useful across many areas of science, technology, and engineering. Traditional implementations of GPs involve stationary kernels (also termed covariance functions) that limit their flexibility and exact methods for inference that prevent application to data sets with more than about ten thousand points. Modern approaches to address stationarity assumptions generally fail to accommodate large data sets, while all attempts to address scalability focus on approximating the Gaussian likelihood, which can involve subjectivity and lead to inaccuracies. In this work, we explicitly derive an alternative kernel that can discover and encode both sparsity and nonstationarity. We embed the kernel within a fully Bayesian GP model and leverage high-performance computing resources to enable the analysis of massive data sets. We demonstrate the favorable performance of our novel kernel relative to existing exact and approximate GP methods across a variety of synthetic data examples. Furthermore, we conduct space-time prediction based on more than one million measurements of daily maximum temperature and verify that our results outperform state-of-the-art methods in the Earth sciences. More broadly, having access to exact GPs that use ultra-scalable, sparsity-discovering, nonstationary kernels allows GP methods to truly compete with a wide variety of machine learning methods.

We demonstrate that neural networks can be used to improve search strategies, over existing strategies, in LHC searches for light electroweak-charged scalars that decay to a muon and a heavy invisible fermion. We propose a new search involving a neural network discriminator as a final cut and show that different signal regions can be defined using networks trained on different subsets of signal samples (distinguishing low-mass and high-mass regions). We also present a workflow using publicly-available analysis tools, that can lead, from background and signal simulation, to network training, through to finding projections for limits using an analysis and ${\tt ONNX}$ libraries to interface network and recasting tools. We provide an estimate of the sensitivity of our search from Run 2 LHC data, and projections for higher luminosities, showing a clear advantage over previous methods., Comment: 13 pages, 4 figures

We used attitude data from the Mars Ingenuity helicopter with a simple steady-state model to estimate windspeeds and directions at altitudes of 3 meters up to 24 meters, the first time winds at such altitudes have been probed on Mars. We compared our estimates to concurrent wind data at 1.5 m height from the meteorology package MEDA onboard the Mars 2020 Perseverance rover and to predictions from meteorological models. Wind directions inferred from the Ingenuity data agreed to within uncertainties with the directions measured by MEDA, when the latter were available, but deviated from model-predicted directions by as much as 180 deg in some cases. Also, the inferred windspeeds are often much higher than expected. For example, meteorological predictions tailored to the time and location of Ingenuity's 59th flight suggest Ingenuity should not have seen windspeeds above about 15 m/s, but we inferred speeds reaching nearly 25 m/s. By contrast, the 61st flight was at a similar time and season and showed weaker winds then the 59th flight, suggesting winds shaped by transient phenomena. For flights during which we have MEDA data to compare to, inferred windspeeds imply friction velocities exceeding 1 m/s and roughness lengths of more than 10 cm based on a boundary layer model that incorporates convective instability, which seem implausibly large. These results suggest Ingenuity was probing winds sensitive to aerodynamic conditions hundreds of meters upwind instead of the conditions very near Mars 2020, but they may also reflect a need for updated boundary layer wind models. An improved model for Ingenuity's aerodynamic response that includes the effects of transient winds may also modify our results. In any case, the work here provides a foundation for exploration of planetary boundary layers using drones and suggests important future avenues for research and development., Comment: Accepted by PSJ

This paper introduces briefly the history and growth of the Detection and Classification of Acoustic Scenes and Events (DCASE) challenge, workshop, research area and research community. Created in 2013 as a data evaluation challenge, DCASE has become a major research topic in the Audio and Acoustic Signal Processing area. Its success comes from a combination of factors: the challenge offers a large variety of tasks that are renewed each year; and the workshop offers a channel for dissemination of related work, engaging a young and dynamic community. At the same time, DCASE faces its own challenges, growing and expanding to different areas. One of the core principles of DCASE is open science and reproducibility: publicly available datasets, baseline systems, technical reports and workshop publications. While the DCASE challenge and workshop are independent of IEEE SPS, the challenge receives annual endorsement from the AASP TC, and the DCASE community contributes significantly to the ICASSP flagship conference and the success of SPS in many of its activities., Comment: Submitted to ICASSP 2025

Intracranial aneurysms (IAs) that rupture result in significant morbidity and mortality. While traditional risk models such as the PHASES score are useful in clinical decision making, machine learning (ML) models offer the potential to provide more accuracy. In this study, we compared the performance of four different machine learning algorithms Random Forest (RF), XGBoost (XGB), Support Vector Machine (SVM), and Multi Layer Perceptron (MLP) on clinical and radiographic features to predict rupture status of intracranial aneurysms. Among the models, RF achieved the highest accuracy (85%) with balanced precision and recall, while MLP had the lowest overall performance (accuracy of 63%). Fractal dimension ranked as the most important feature for model performance across all models.

Benchmarking in bioinformatics is a process of designing, running and disseminating rigorous performance evaluations of methods (software). Benchmarking systems facilitate the benchmarking process by providing an entrypoint to store, coordinate and execute concrete benchmarks. We describe an alpha version of a new benchmarking system, Omnibenchmark, to facilitate benchmark formalization and execution in solo and community efforts. Omnibenchmark provides a benchmark definition syntax (in a configuration YAML file), a dynamic workflow generation based on Snakemake, S3-compatible storage handling, and reproducible software environments using EasyBuild, lmod, Apptainer or conda. Tutorials and installation instructions are available from https://omnibenchmark.org., Comment: 7 pages, 1 figure, 1 table

Benchmarking, which involves collecting reference datasets and demonstrating method performances, is a requirement for the development of new computational tools, but also becomes a domain of its own to achieve neutral comparisons of methods. Although a lot has been written about how to design and conduct benchmark studies, this Perspective sheds light on a wish list for a computational platform to orchestrate benchmark studies. We discuss various ideas for organizing reproducible software environments, formally defining benchmarks, orchestrating standardized workflows, and how they interface with computing infrastructure., Comment: 19 pages, 2 figures, 1 table

This paper presents a residential audio dataset to support sound event detection research for smart home applications aimed at promoting wellbeing for older adults. The dataset is constructed by deploying audio recording systems in the homes of 8 participants aged 55-80 years for a 7-day period. Acoustic characteristics are documented through detailed floor plans and construction material information to enable replication of the recording environments for AI model deployment. A novel automated speech removal pipeline is developed, using pre-trained audio neural networks to detect and remove segments containing spoken voice, while preserving segments containing other sound events. The resulting dataset consists of privacy-compliant audio recordings that accurately capture the soundscapes and activities of daily living within residential spaces. The paper details the dataset creation methodology, the speech removal pipeline utilizing cascaded model architectures, and an analysis of the vocal label distribution to validate the speech removal process. This dataset enables the development and benchmarking of sound event detection models tailored specifically for in-home applications.

Language-queried audio source separation (LASS) focuses on separating sounds using textual descriptions of the desired sources. Current methods mainly use discriminative approaches, such as time-frequency masking, to separate target sounds and minimize interference from other sources. However, these models face challenges when separating overlapping soundtracks, which may lead to artifacts such as spectral holes or incomplete separation. Rectified flow matching (RFM), a generative model that establishes linear relations between the distribution of data and noise, offers superior theoretical properties and simplicity, but has not yet been explored in sound separation. In this work, we introduce FlowSep, a new generative model based on RFM for LASS tasks. FlowSep learns linear flow trajectories from noise to target source features within the variational autoencoder (VAE) latent space. During inference, the RFM-generated latent features are reconstructed into a mel-spectrogram via the pre-trained VAE decoder, followed by a pre-trained vocoder to synthesize the waveform. Trained on 1,680 hours of audio data, FlowSep outperforms the state-of-the-art models across multiple benchmarks, as evaluated with subjective and objective metrics. Additionally, our results show that FlowSep surpasses a diffusion-based LASS model in both separation quality and inference efficiency, highlighting its strong potential for audio source separation tasks. Code, pre-trained models and demos can be found at: https://audio-agi.github.io/FlowSep_demo/ ., Comment: Accepted by ICASSP 2025, camera-ready version

Audio Event Recognition (AER) traditionally focuses on detecting and identifying audio events. Most existing AER models tend to detect all potential events without considering their varying significance across different contexts. This makes the AER results detected by existing models often have a large discrepancy with human auditory perception. Although this is a critical and significant issue, it has not been extensively studied by the Detection and Classification of Sound Scenes and Events (DCASE) community because solving it is time-consuming and labour-intensive. To address this issue, this paper introduces the concept of semantic importance in AER, focusing on exploring the differences between human perception and model inference. This paper constructs a Multi-Annotated Foreground Audio Event Recognition (MAFAR) dataset, which comprises audio recordings labelled by 10 professional annotators. Through labelling frequency and variance, the MAFAR dataset facilitates the quantification of semantic importance and analysis of human perception. By comparing human annotations with the predictions of ensemble pre-trained models, this paper uncovers a significant gap between human perception and model inference in both semantic identification and existence detection of audio events. Experimental results reveal that human perception tends to ignore subtle or trivial events in the event semantic identification, while model inference is easily affected by events with noises. Meanwhile, in event existence detection, models are usually more sensitive than humans., Comment: Dataset homepage: https://github.com/Voltmeter00/MAFAR

We extend the so-called hyperchargeless Higgs triplet model to include a weak triplet of Dirac fermions and a Dirac fermion $X$ transforming trivially under the Standard Model gauge group. We are motivated in part by a collection of anomalies that can be interpreted as a new scalar state with mass of approximately 152 GeV coupling to $W$ but not $Z$ bosons, which may be consistent with the electrically neutral triplet scalar in this model (provided that it mixes very slightly with the Standard Model Higgs boson). Meanwhile, the lightest neutral fermion in the model is stable and can be thermal dark matter with the correct relic abundance if it mixes lightly with the neutral triplet fermion, such that the dark matter is composed mostly of $X$. Because $X$ couples to the Standard Model only through a Yukawa-like interaction with the pair of triplets, this model falls into the frustrated dark matter paradigm. Finally, the spectrum of exotic fermions in this model can exhibit the strong compression favored by the current excess in the monojet channel, and evades multijet constraints in the region favored by monojets. In this work we explore this model's phenomenology and compare the parameter space regions best suited to each of the aforementioned excesses and constraints., Comment: 11 pages, 5 figures

We use {\it ab initio} calculations to understand the current-induced optical response and orbital moment accumulation at the surfaces of metallic films. These two quantities are related by a sum rule that equates the circular dichroic absorption integrated over frequency to the gauge-invariant self-rotation contribution to the orbital magnetization, $\vec{M}_{\rm SR}$. In typical ferromagnets, $\vec{M}_{\rm SR}$ is a good approximation to the total orbital magnetization. We compute the current-induced $\vec{M}_{\rm SR}$ for a Pt thin film and compare it to the current-induced orbital moment accumulation calculated with the atom-centered approximation (ACA). We find significant differences: the size of $\vec{M}_{\rm SR}$ is, in general, larger than the ACA orbital moment accumulation by an order of magnitude and includes substantial finite-size effects. The differences between the two quantities caution against interpreting optical measurements with models utilizing the ACA. Finally, we compute the total $\vec{M}_{\rm SR}$ and ACA orbital moment accumulation as a function of layer thickness. For both quantities, the length scale at which the total surface accumulation saturates is on the order of the mean free path and longer than the length scale of their spatial profiles.

This study aimed to determine if ChatGPT's large language models could match the scoring accuracy of human and machine scores from the ASAP competition. The investigation focused on various prediction models, including linear regression, random forest, gradient boost, and boost. ChatGPT's performance was evaluated against human raters using quadratic weighted kappa (QWK) metrics. Results indicated that while ChatGPT's gradient boost model achieved QWKs close to human raters for some data sets, its overall performance was inconsistent and often lower than human scores. The study highlighted the need for further refinement, particularly in handling biases and ensuring scoring fairness. Despite these challenges, ChatGPT demonstrated potential for scoring efficiency, especially with domain-specific fine-tuning. The study concludes that ChatGPT can complement human scoring but requires additional development to be reliable for high-stakes assessments. Future research should improve model accuracy, address ethical considerations, and explore hybrid models combining ChatGPT with empirical methods., Comment: 35 pages, 8 tables, 2 Figures, 27 references

Climate change detection and attribution (D&A) is concerned with determining the extent to which anthropogenic activities have influenced specific aspects of the global climate system. D&A fits within the broader field of causal inference, the collection of statistical methods that identify cause and effect relationships. There are a wide variety of methods for making attribution statements, each of which require different types of input data and focus on different types of weather and climate events and each of which are conditional to varying extents. Some methods are based on Pearl causality while others leverage Granger causality, and the causal framing provides important context for how the resulting attribution conclusion should be interpreted. However, while Granger-causal attribution analyses have become more common, there is no clear statement of their strengths and weaknesses and no clear consensus on where and when Granger-causal perspectives are appropriate. In this prospective paper, we provide a formal definition for Granger-based approaches to trend and event attribution and a clear comparison with more traditional methods for assessing the human influence on extreme weather and climate events. Broadly speaking, Granger-causal attribution statements can be constructed quickly from observations and do not require computationally-intesive dynamical experiments. These analyses also enable rapid attribution, which is useful in the aftermath of a severe weather event, and provide multiple lines of evidence for anthropogenic climate change when paired with Pearl-causal attribution. Confidence in attribution statements is increased when different methodologies arrive at similar conclusions. Moving forward, we encourage the D&A community to embrace hybrid approaches to climate change attribution that leverage the strengths of both Granger and Pearl causality.

The last decade has seen numerous record-shattering heatwaves in all corners of the globe. In the aftermath of these devastating events, there is interest in identifying worst-case thresholds or upper bounds that quantify just how hot temperatures can become. Generalized Extreme Value theory provides a data-driven estimate of extreme thresholds; however, upper bounds may be exceeded by future events, which undermines attribution and planning for heatwave impacts. Here, we show how the occurrence and relative probability of observed events that exceed a priori upper bound estimates, so-called "impossible" temperatures, has changed over time. We find that many unprecedented events are actually within data-driven upper bounds, but only when using modern spatial statistical methods. Furthermore, there are clear connections between anthropogenic forcing and the "impossibility" of the most extreme temperatures. Robust understanding of heatwave thresholds provides critical information about future record-breaking events and how their extremity relates to historical measurements.

A goal of computational studies of protein-protein interfaces (PPIs) is to predict the binding site between two monomers that form a heterodimer. The simplest version of this problem is to rigidly re-dock the bound forms of the monomers, which involves generating computational models of the heterodimer and then scoring them to determine the most native-like models. Scoring functions have been assessed previously using rank- and classification-based metrics, however, these methods are sensitive to the number and quality of models in the scoring function training set. We assess the accuracy of seven PPI scoring functions by comparing their scores to a measure of structural similarity to the x-ray crystal structure (i.e. the DockQ score) for a non-redundant set of heterodimers from the Protein Data Bank. For each heterodimer, we generate re-docked models uniformly sampled over DockQ and calculate the Spearman correlation between the PPI scores and DockQ. For some targets, the scores and DockQ are highly correlated; however, for many targets, there are weak correlations. Several physical features can explain the difference between difficult- and easy-to-score targets. For example, strong correlations exist between the score and DockQ for targets with highly intertwined monomers and many interface contacts. We also develop a new score based on only three physical features that matches or exceeds the performance of current PPI scoring functions. These results emphasize that PPI prediction can be improved by focusing on correlations between the PPI score and DockQ and incorporating more discriminating physical features into PPI scoring functions., Comment: 21 pages, 7 figures

The broadcasting industry is increasingly adopting IP techniques, revolutionising both live and pre-recorded content production, from news gathering to live music events. IP broadcasting allows for the transport of audio and video signals in an easily configurable way, aligning with modern networking techniques. This shift towards an IP workflow allows for much greater flexibility, not only in routing signals but with the integration of tools using standard web development techniques. One possible tool could include the use of live audio tagging, which has a number of uses in the production of content. These include from automated closed captioning to identifying unwanted sound events within a scene. In this paper, we describe the process of containerising an audio tagging model into a microservice, a small segregated code module that can be integrated into a multitude of different network setups. The goal is to develop a modular, accessible, and flexible tool capable of seamless deployment into broadcasting workflows of all sizes, from small productions to large corporations. Challenges surrounding latency of the selected audio tagging model and its effect on the usefulness of the end product are discussed., Comment: Submitted to DCASE 2024 Workshop

Significant improvement has been achieved in automated audio captioning (AAC) with recent models. However, these models have become increasingly large as their performance is enhanced. In this work, we propose a knowledge distillation (KD) framework for AAC. Our analysis shows that in the encoder-decoder based AAC models, it is more effective to distill knowledge into the encoder as compared with the decoder. To this end, we incorporate encoder-level KD loss into training, in addition to the standard supervised loss and sequence-level KD loss. We investigate two encoder-level KD methods, based on mean squared error (MSE) loss and contrastive loss, respectively. Experimental results demonstrate that contrastive KD is more robust than MSE KD, exhibiting superior performance in data-scarce situations. By leveraging audio-only data into training in the KD framework, our student model achieves competitive performance, with an inference speed that is 19 times faster\footnote{An online demo is available at \url{https://huggingface.co/spaces/wsntxxn/efficient_audio_captioning}}., Comment: Interspeech 2024

Universal sound separation (USS) is a task of separating mixtures of arbitrary sound sources. Typically, universal separation models are trained from scratch in a supervised manner, using labeled data. Self-supervised learning (SSL) is an emerging deep learning approach that leverages unlabeled data to obtain task-agnostic representations, which can benefit many downstream tasks. In this paper, we propose integrating a self-supervised pre-trained model, namely the audio masked autoencoder (A-MAE), into a universal sound separation system to enhance its separation performance. We employ two strategies to utilize SSL embeddings: freezing or updating the parameters of A-MAE during fine-tuning. The SSL embeddings are concatenated with the short-time Fourier transform (STFT) to serve as input features for the separation model. We evaluate our methods on the AudioSet dataset, and the experimental results indicate that the proposed methods successfully enhance the separation performance of a state-of-the-art ResUNet-based USS model.

Collusion between students in online exams is a major problem that undermines the integrity of the exam results. Although there exist methods that use exam data to identify pairs of students who have likely copied each other's answers, these methods are restricted to specific formats of multiple-choice exams. Here we present a statistical algorithm, Q-SID, that efficiently detects groups of students who likely have colluded, i.e., collusion groups, with error quantification. Q-SID uses graded numeric question scores only, so it works for many formats of multiple-choice and non-multiple-choice exams. Q-SID reports two false-positive rates (FPRs) for each collusion group: (1) empirical FPR, whose null data are from 36 strictly proctored exam datasets independent of the user-input exam data and (2) synthetic FPR, whose null data are simulated from a copula-based probabilistic model, which is first fitted to the user-input exam data and then modified to have no collusion. On 34 unproctored exam datasets, including two benchmark datasets with true positives and negatives verified by textural analysis, we demonstrate that Q-SID is a collusion detection algorithm with powerful and robust performance across exam formats, numbers of questions and students, and exam complexity.

Generative models have shown significant achievements in audio generation tasks. However, existing models struggle with complex and detailed prompts, leading to potential performance degradation. We hypothesize that this problem stems from the simplicity and scarcity of the training data. This work aims to create a large-scale audio dataset with rich captions for improving audio generation models. We first develop an automated pipeline to generate detailed captions by transforming predicted visual captions, audio captions, and tagging labels into comprehensive descriptions using a Large Language Model (LLM). The resulting dataset, Sound-VECaps, comprises 1.66M high-quality audio-caption pairs with enriched details including audio event orders, occurred places and environment information. We then demonstrate that training the text-to-audio generation models with Sound-VECaps significantly improves the performance on complex prompts. Furthermore, we conduct ablation studies of the models on several downstream audio-language tasks, showing the potential of Sound-VECaps in advancing audio-text representation learning. Our dataset and models are available online from here https://yyua8222.github.io/Sound-VECaps-demo/., Comment: 5 pages with 1 appendix, accepted by ICASSP 2025

Capillary droplets form due to surface tension when two immiscible fluids are mixed. We describe the motion of gravity-driven capillary droplets flowing through narrow constrictions and obstacle arrays in both simulations and experiments. Our new capillary deformable particle model recapitulates the shape and velocity of single oil droplets in water as they pass through narrow constrictions in microfluidic chambers. Using this experimentally validated model, we simulate the flow and clogging of single capillary droplets in narrow channels and obstacle arrays and find several important results. First, the capillary droplet speed profile is nonmonotonic as the droplet exits the narrow orifice, and we can tune the droplet properties so that the speed overshoots the terminal speed far from the constriction. Second, in obstacle arrays, we find that extremely deformable droplets can wrap around obstacles, which leads to decreased average droplet speed in the continuous flow regime and increased probability for clogging in the regime where permanent clogs form. Third, the wrapping mechanism causes the clogging probability in obstacle arrays to become nonmonotonic with surface tension $\Gamma$. At large $\Gamma$, the droplets are nearly rigid and the clogging probability is large since the droplets can not squeeze through the gaps between obstacles. With decreasing $\Gamma$, the clogging probability decreases as the droplets become more deformable. However, in the small-$\Gamma$ limit the clogging probability increases, since the droplets are extremely deformable and cause clogs as they wrap around the obstacles. The results from these studies are important for developing a predictive understanding of capillary droplet flows through complex and confined geometries.

This is the manual for the version 2 of HackAnalysis, a powerful, lightweight, versatile and, most importantly, hackable, recasting tool. New features in this version include: compressed event format storage for ultra-fast development; integration of new physics and mathematics routines via RestFrames and Eigen; automatic computation of systematic uncertainties; an interface to ONNX for neural networks; easy implementation of new models via QNUMBERS blocks; a python package for interfacing to pyhf, spey and in-built statistics routines for fast computation of experimental limits; improved integration with BSMArt for fast scanning, and a new batch running/convergence check. Several new (electroweakino) analyses are included with this release., Comment: 40 pages, 2 figures, 7 tables. Comments welcome. Code available at https://goodsell.pages.in2p3.fr/hackanalysis/

The trends and reactions of the general public towards global events can be analyzed using data from social platforms, including Twitter. The number of tweets has been reported to help detect variations in communication traffic within subsets like countries, age groups and industries. Similarly, publicly accessible data and (in particular) data from social media about agricultural issues provide a great opportunity for obtaining instantaneous snapshots of farmer opinions and a method to track changes in opinion through temporal analysis. In this paper we hypothesize that the presence of keywords like precision agriculture, digital agriculture, Internet of Things (IoT), BigData, remote sensing, GPS, etc., in tweets could serve as an indicator of discussions centered around interest in modern farming practices. We extracted relevant tweets using keywords such as IoT, BigData and Geographical Information System (GIS), and then analyzed their geographical origin and frequency of their mention. We analyzed the Twitter data for the period of 1st -11th January 2018 to understand these trends and the factors affecting them. These factors, such as special events, projects, biogeography, etc., were further analyzed using tweet sources and trending hashtags from the database. The regions with the highest interest in the keywords were United States, Egypt, Brazil, Japan and China. A comparison of frequency of keywords revealed IoT as the most tweeted word (77.6%) in the downloaded data. The most used language was English followed by Spanish, Japanese and French. Periodical tweets on IoT from an account handled by IoT project on Twitter and Seminars on IoT in January in Santa Catarina (Brazil) were found to be the underlying factors for the observed trends., Comment: 7 pages, conference accepted