Search

Scientific workflows are used to analyze large amounts of data. These workflows comprise numerous tasks, many of which are executed repeatedly, running the same custom program on different inputs. Users specify resource allocations for each task, which must be sufficient for all inputs to prevent task failures. As a result, task memory allocations tend to be overly conservative, wasting precious cluster resources, limiting overall parallelism, and increasing workflow makespan. In this paper, we first benchmark a state-of-the-art method on four real-life workflows from the nf-core workflow repository. This analysis reveals that certain assumptions underlying current prediction methods, which typically were evaluated only on simulated workflows, cannot generally be confirmed for real workflows and executions. We then present Ponder, a new online task-sizing strategy that considers and chooses between different methods to cater to different memory demand patterns. We implemented Ponder for Nextflow and made the code publicly available. In an experimental evaluation that also considers the impact of memory predictions on scheduling, Ponder improves Memory Allocation Quality on average by 71.0% and makespan by 21.8% in comparison to a state-of-the-art method. Moreover, Ponder produces 93.8% fewer task failures.

With the exponential growth of the life science literature, biomedical text mining (BTM) has become an essential technology for accelerating the extraction of insights from publications. Identifying named entities (e.g., diseases, drugs, or genes) in texts and their linkage to reference knowledge bases are crucial steps in BTM pipelines to enable information aggregation from different documents. However, tools for these two steps are rarely applied in the same context in which they were developed. Instead, they are applied in the wild, i.e., on application-dependent text collections different from those used for the tools' training, varying, e.g., in focus, genre, style, and text type. This raises the question of whether the reported performance of BTM tools can be trusted for downstream applications. Here, we report on the results of a carefully designed cross-corpus benchmark for named entity extraction, where tools were applied systematically to corpora not used during their training. Based on a survey of 28 published systems, we selected five for an in-depth analysis on three publicly available corpora encompassing four different entity types. Comparison between tools results in a mixed picture and shows that, in a cross-corpus setting, the performance is significantly lower than the one reported in an in-corpus setting. HunFlair2 showed the best performance on average, being closely followed by PubTator. Our results indicate that users of BTM tools should expect diminishing performances when applying them in the wild compared to original publications and show that further research is necessary to make BTM tools more robust.

Biomedical entity linking (BEL) is the task of grounding entity mentions to a knowledge base (KB). A popular approach to the task are name-based methods, i.e. those identifying the most appropriate name in the KB for a given mention, either via dense retrieval or autoregressive modeling. However, as these methods directly return KB names, they cannot cope with homonyms, i.e. different KB entities sharing the exact same name. This significantly affects their performance, especially for KBs where homonyms account for a large amount of entity mentions (e.g. UMLS and NCBI Gene). We therefore present BELHD (Biomedical Entity Linking with Homonym Disambiguation), a new name-based method that copes with this challenge. Specifically, BELHD builds upon the BioSyn (Sung et al.,2020) model introducing two crucial extensions. First, it performs a preprocessing of the KB in which it expands homonyms with an automatically chosen disambiguating string, thus enforcing unique linking decisions. Second, we introduce candidate sharing, a novel strategy to select candidates for contrastive learning that enhances the overall training signal. Experiments with 10 corpora and five entity types show that BELHD improves upon state-of-the-art approaches, achieving the best results in 6 out 10 corpora with an average improvement of 4.55pp recall@1. Furthermore, the KB preprocessing is orthogonal to the core prediction model and thus can also improve other methods, which we exemplify for GenBioEL (Yuan et al, 2022), a generative name-based BEL approach. Code is available at: link added upon publication.

Scientific workflow systems are increasingly popular for expressing and executing complex data analysis pipelines over large datasets, as they offer reproducibility, dependability, and scalability of analyses by automatic parallelization on large compute clusters. However, implementing workflows is difficult due to the involvement of many black-box tools and the deep infrastructure stack necessary for their execution. Simultaneously, user-supporting tools are rare, and the number of available examples is much lower than in classical programming languages. To address these challenges, we investigate the efficiency of Large Language Models (LLMs), specifically ChatGPT, to support users when dealing with scientific workflows. We performed three user studies in two scientific domains to evaluate ChatGPT for comprehending, adapting, and extending workflows. Our results indicate that LLMs efficiently interpret workflows but achieve lower performance for exchanging components or purposeful workflow extensions. We characterize their limitations in these challenging scenarios and suggest future research directions.

Ubiquitous sensors today emit high frequency streams of numerical measurements that reflect properties of human, animal, industrial, commercial, and natural processes. Shifts in such processes, e.g. caused by external events or internal state changes, manifest as changes in the recorded signals. The task of streaming time series segmentation (STSS) is to partition the stream into consecutive variable-sized segments that correspond to states of the observed processes or entities. The partition operation itself must in performance be able to cope with the input frequency of the signals. We introduce ClaSS, a novel, efficient, and highly accurate algorithm for STSS. ClaSS assesses the homogeneity of potential partitions using self-supervised time series classification and applies statistical tests to detect significant change points (CPs). In our experimental evaluation using two large benchmarks and six real-world data archives, we found ClaSS to be significantly more precise than eight state-of-the-art competitors. Its space and time complexity is independent of segment sizes and linear only in the sliding window size. We also provide ClaSS as a window operator with an average throughput of 1k data points per second for the Apache Flink streaming engine.

Many resource management techniques for task scheduling, energy and carbon efficiency, and cost optimization in workflows rely on a-priori task runtime knowledge. Building runtime prediction models on historical data is often not feasible in practice as workflows, their input data, and the cluster infrastructure change. Online methods, on the other hand, which estimate task runtimes on specific machines while the workflow is running, have to cope with a lack of measurements during start-up. Frequently, scientific workflows are executed on heterogeneous infrastructures consisting of machines with different CPU, I/O, and memory configurations, further complicating predicting runtimes due to different task runtimes on different machine types. This paper presents Lotaru, a method for locally predicting the runtimes of scientific workflow tasks before they are executed on heterogeneous compute clusters. Crucially, our approach does not rely on historical data and copes with a lack of training data during the start-up. To this end, we use microbenchmarks, reduce the input data to quickly profile the workflow locally, and predict a task's runtime with a Bayesian linear regression based on the gathered data points from the local workflow execution and the microbenchmarks. Due to its Bayesian approach, Lotaru provides uncertainty estimates that can be used for advanced scheduling methods on distributed cluster infrastructures. In our evaluation with five real-world scientific workflows, our method outperforms two state-of-the-art runtime prediction baselines and decreases the absolute prediction error by more than 12.5%. In a second set of experiments, the prediction performance of our method, using the predicted runtimes for state-of-the-art scheduling, carbon reduction, and cost prediction, enables results close to those achieved with perfect prior knowledge of runtimes.

Biomedical entity linking (BEL) is the task of grounding entity mentions to a knowledge base. It plays a vital role in information extraction pipelines for the life sciences literature. We review recent work in the field and find that, as the task is absent from existing benchmarks for biomedical text mining, different studies adopt different experimental setups making comparisons based on published numbers problematic. Furthermore, neural systems are tested primarily on instances linked to the broad coverage knowledge base UMLS, leaving their performance to more specialized ones, e.g. genes or variants, understudied. We therefore developed BELB, a Biomedical Entity Linking Benchmark, providing access in a unified format to 11 corpora linked to 7 knowledge bases and spanning six entity types: gene, disease, chemical, species, cell line and variant. BELB greatly reduces preprocessing overhead in testing BEL systems on multiple corpora offering a standardized testbed for reproducible experiments. Using BELB we perform an extensive evaluation of six rule-based entity-specific systems and three recent neural approaches leveraging pre-trained language models. Our results reveal a mixed picture showing that neural approaches fail to perform consistently across entity types, highlighting the need of further studies towards entity-agnostic models.

Porting a scientific data analysis workflow (DAW) to a cluster infrastructure, a new software stack, or even only a new dataset with some notably different properties is often challenging. Despite the structured definition of the steps (tasks) and their interdependencies during a complex data analysis in the DAW specification, relevant assumptions may remain unspecified and implicit. Such hidden assumptions often lead to crashing tasks without a reasonable error message, poor performance in general, non-terminating executions, or silent wrong results of the DAW, to name only a few possible consequences. Searching for the causes of such errors and drawbacks in a distributed compute cluster managed by a complex infrastructure stack, where DAWs for large datasets typically are executed, can be tedious and time-consuming. We propose validity constraints (VCs) as a new concept for DAW languages to alleviate this situation. A VC is a constraint specifying some logical conditions that must be fulfilled at certain times for DAW executions to be valid. When defined together with a DAW, VCs help to improve the portability, adaptability, and reusability of DAWs by making implicit assumptions explicit. Once specified, VC can be controlled automatically by the DAW infrastructure, and violations can lead to meaningful error messages and graceful behaviour (e.g., termination or invocation of repair mechanisms). We provide a broad list of possible VCs, classify them along multiple dimensions, and compare them to similar concepts one can find in related fields. We also provide a first sketch for VCs' implementation into existing DAW infrastructures.

Scientific workflow management systems (SWMSs) and resource managers together ensure that tasks are scheduled on provisioned resources so that all dependencies are obeyed, and some optimization goal, such as makespan minimization, is achieved. In practice, however, there is no clear separation of scheduling responsibilities between an SWMS and a resource manager because there exists no agreed-upon separation of concerns between their different components. This has two consequences. First, the lack of a standardized API to exchange scheduling information between SWMSs and resource managers hinders portability. It incurs costly adaptations when a component should be replaced by a different one (e.g., an SWMS with another SWMS on the same resource manager). Second, due to overlapping functionalities, current installations often actually have two schedulers, both making partial scheduling decisions under incomplete information, leading to suboptimal workflow scheduling. In this paper, we propose a simple REST interface between SWMSs and resource managers, which allows any SWMS to pass dynamic workflow information to a resource manager, enabling maximally informed scheduling decisions. We provide an implementation of this API as an example, using Nextflow as an SWMS and Kubernetes as a resource manager. Our experiments with nine real-world workflows show that this strategy reduces makespan by up to 25.1% and 10.8% on average compared to the standard Nextflow/Kubernetes configuration. Furthermore, a more widespread implementation of this API would enable leaner code bases, a simpler exchange of components of workflow systems, and a unified place to implement new scheduling algorithms.

A time series is a sequence of sequentially ordered real values in time. Time series classification (TSC) is the task of assigning a time series to one of a set of predefined classes, usually based on a model learned from examples. Dictionary-based methods for TSC rely on counting the frequency of certain patterns in time series and are important components of the currently most accurate TSC ensembles. One of the early dictionary-based methods was WEASEL, which at its time achieved SotA results while also being very fast. However, it is outperformed both in terms of speed and accuracy by other methods. Furthermore, its design leads to an unpredictably large memory footprint, making it inapplicable for many applications. In this paper, we present WEASEL 2.0, a complete overhaul of WEASEL based on two recent advancements in TSC: Dilation and ensembling of randomized hyper-parameter settings. These two techniques allow WEASEL 2.0 to work with a fixed-size memory footprint while at the same time improving accuracy. Compared to 15 other SotA methods on the UCR benchmark set, WEASEL 2.0 is significantly more accurate than other dictionary methods and not significantly worse than the currently best methods. Actually, it achieves the highest median accuracy over all data sets, and it performs best in 5 out of 12 problem classes. We thus believe that WEASEL 2.0 is a viable alternative for current TSC and also a potentially interesting input for future ensembles.

Scientific workflows typically comprise a multitude of different processing steps which often are executed in parallel on different partitions of the input data. These executions, in turn, must be scheduled on the compute nodes of the computational infrastructure at hand. This assignment is complicated by the facts that (a) tasks typically have highly heterogeneous resource requirements and (b) in many infrastructures, compute nodes offer highly heterogeneous resources. In consequence, predictions of the runtime of a given task on a given node, as required by many scheduling algorithms, are often rather imprecise, which can lead to sub-optimal scheduling decisions. We propose Reshi, a method for recommending task-node assignments during workflow execution that can cope with heterogeneous tasks and heterogeneous nodes. Reshi approaches the problem as a regression task, where task-node pairs are modeled as feature vectors over the results of dedicated micro benchmarks and past task executions. Based on these features, Reshi trains a regression tree model to rank and recommend nodes for each ready-to-run task, which can be used as input to a scheduler. For our evaluation, we benchmarked 27 AWS machine types using three representative workflows. We compare Reshi's recommendations with three state-of-the-art schedulers. Our evaluation shows that Reshi outperforms HEFT by a mean makespan reduction of 7.18% and 18.01% assuming a mean task runtime prediction error of 15%., Comment: Paper accepted in 41st IEEE International Performance Computing and Communications Conference (IPCCC 2022)

The study of natural and human-made processes often results in long sequences of temporally-ordered values, aka time series (TS). Such processes often consist of multiple states, e.g. operating modes of a machine, such that state changes in the observed processes result in changes in the distribution of shape of the measured values. Time series segmentation (TSS) tries to find such changes in TS post-hoc to deduce changes in the data-generating process. TSS is typically approached as an unsupervised learning problem aiming at the identification of segments distinguishable by some statistical property. Current algorithms for TSS require domain-dependent hyper-parameters to be set by the user, make assumptions about the TS value distribution or the types of detectable changes which limits their applicability. Common hyperparameters are the measure of segment homogeneity and the number of change points, which are particularly hard to tune for each data set. We present ClaSP, a novel, highly accurate, hyper-parameter-free and domain-agnostic method for TSS. ClaSP hierarchically splits a TS into two parts. A change point is determined by training a binary TS classifier for each possible split point and selecting the one split that is best at identifying subsequences to be from either of the partitions. ClaSP learns its main two model-parameters from the data using two novel bespoke algorithms. In our experimental evaluation using a benchmark of 107 data sets, we show that ClaSP outperforms the state of the art in terms of accuracy and is fast and scalable. Furthermore, we highlight properties of ClaSP using several real-world case studies.

A time series motif intuitively is a short time series that repeats itself approximately the same within a larger time series. Such motifs often represent concealed structures, such as heart beats in an ECG recording, the riff in a pop song, or sleep spindles in EEG sleep data. Motif discovery (MD) is the task of finding such motifs in a given input series. As there are varying definitions of what exactly a motif is, a number of different algorithms exist. As central parameters they all take the length l of the motif and the maximal distance r between the motif's occurrences. In practice, however, especially suitable values for r are very hard to determine upfront, and found motifs show a high variability even for very similar r values. Accordingly, finding an interesting motif requires extensive trial-and-error. In this paper, we present a different approach to the MD problem. We define k-Motiflets as the set of exactly k occurrences of a motif of length l, whose maximum pairwise distance is minimal. This turns the MD problem upside-down: The central parameter of our approach is not the distance threshold r, but the desired number of occurrence k of the motif, which we show is considerably more intuitive and easier to set. Based on this definition, we present exact and approximate algorithms for finding k-Motiflets and analyze their complexity. To further ease the use of our method, we describe statistical tools to automatically determine meaningful values for its input parameters. By evaluation on several real-world data sets and comparison to four SotA MD algorithms, we show that our proposed algorithm is both quantitatively superior to its competitors, finding larger motif sets at higher similarity, and qualitatively better, leading to clearer and easier to interpret motifs without any need for manual tuning.

Many scientific workflow scheduling algorithms need to be informed about task runtimes a-priori to conduct efficient scheduling. In heterogeneous cluster infrastructures, this problem becomes aggravated because these runtimes are required for each task-node pair. Using historical data is often not feasible as logs are typically not retained indefinitely and workloads as well as infrastructure changes. In contrast, online methods, which predict task runtimes on specific nodes while the workflow is running, have to cope with the lack of example runs, especially during the start-up. In this paper, we present Lotaru, a novel online method for locally estimating task runtimes in scientific workflows on heterogeneous clusters. Lotaru first profiles all nodes of a cluster with a set of short-running and uniform microbenchmarks. Next, it runs the workflow to be scheduled on the user's local machine with drastically reduced data to determine important task characteristics. Based on these measurements, Lotaru learns a Bayesian linear regression model to predict a task's runtime given the input size and finally adjusts the predicted runtime specifically for each task-node pair in the cluster based on the micro-benchmark results. Due to its Bayesian approach, Lotaru can also compute robust uncertainty estimates and provides them as an input for advanced scheduling methods. Our evaluation with five real-world scientific workflows and different datasets shows that Lotaru significantly outperforms the baselines in terms of prediction errors for homogeneous and heterogeneous clusters., Comment: paper accepted in 34th International Conference on Scientific and Statistical Database Management (SSDBM 2022)

Interpretability and uncertainty quantification in machine learning can provide justification for decisions, promote scientific discovery and lead to a better understanding of model behavior. Symbolic regression provides inherently interpretable machine learning, but relatively little work has focused on the use of symbolic regression on noisy data and the accompanying necessity to quantify uncertainty. A new Bayesian framework for genetic-programming-based symbolic regression (GPSR) is introduced that uses model evidence (i.e., marginal likelihood) to formulate replacement probability during the selection phase of evolution. Model parameter uncertainty is automatically quantified, enabling probabilistic predictions with each equation produced by the GPSR algorithm. Model evidence is also quantified in this process, and its use is shown to increase interpretability, improve robustness to noise, and reduce overfitting when compared to a conventional GPSR implementation on both numerical and physical experiments.

Ruptures of the largest earthquakes can last between a few seconds and several minutes. An early assessment of the final earthquake size is essential for early warning systems. However, it is still unclear when in the rupture history this final size can be predicted. Here we introduce a probabilistic view of rupture evolution - how likely is the event to become large - allowing for a clear and well-founded answer with implications for earthquake physics and early warning. We apply our approach to real time magnitude estimation based on either moment rate functions or broadband teleseismic P arrivals. In both cases, we find strong and principled evidence against early rupture predictability because differentiation between differently sized ruptures only occurs once half of the rupture has been observed. Even then, it is impossible to foresee future asperities. Our results hint towards a universal initiation behavior for small and large ruptures., Comment: 7 pages main text, 15 pages supplement

In reliability analysis, methods used to estimate failure probability are often limited by the costs associated with model evaluations. Many of these methods, such as multifidelity importance sampling (MFIS), rely upon a computationally efficient, surrogate model like a Gaussian process (GP) to quickly generate predictions. The quality of the GP fit, particularly in the vicinity of the failure region(s), is instrumental in supplying accurately predicted failures for such strategies. We introduce an entropy-based GP adaptive design that, when paired with MFIS, provides more accurate failure probability estimates and with higher confidence. We show that our greedy data acquisition strategy better identifies multiple failure regions compared to existing contour-finding schemes. We then extend the method to batch selection, without sacrificing accuracy. Illustrative examples are provided on benchmark data as well as an application to an impact damage simulator for National Aeronautics and Space Administration (NASA) spacesuits., Comment: 28 pages, 11 figures

Multi-model Monte Carlo methods, such as multi-level Monte Carlo (MLMC) and multifidelity Monte Carlo (MFMC), allow for efficient estimation of the expectation of a quantity of interest given a set of models of varying fidelities. Recently, it was shown that the MLMC and MFMC estimators are both instances of the approximate control variates (ACV) framework [Gorodetsky et al. 2020]. In that same work, it was also shown that hand-tailored ACV estimators could outperform MLMC and MFMC for a variety of model scenarios. Because there is no reason to believe that these hand-tailored estimators are the best among a myriad of possible ACV estimators, a more general approach to estimator construction is pursued in this work. First, a general form of the ACV estimator variance is formulated. Then, the formulation is utilized to generate parametrically-defined estimators. These parametrically-defined estimators allow for an optimization to be pursued over a larger domain of possible ACV estimators. The parametrically-defined estimators are tested on a large set of model scenarios, and it is found that the broader search domain enabled by parametrically-defined estimators leads to greater variance reduction.

The landscape of workflow systems for scientific applications is notoriously convoluted with hundreds of seemingly equivalent workflow systems, many isolated research claims, and a steep learning curve. To address some of these challenges and lay the groundwork for transforming workflows research and development, the WorkflowsRI and ExaWorks projects partnered to bring the international workflows community together. This paper reports on discussions and findings from two virtual "Workflows Community Summits" (January and April, 2021). The overarching goals of these workshops were to develop a view of the state of the art, identify crucial research challenges in the workflows community, articulate a vision for potential community efforts, and discuss technical approaches for realizing this vision. To this end, participants identified six broad themes: FAIR computational workflows; AI workflows; exascale challenges; APIs, interoperability, reuse, and standards; training and education; and building a workflows community. We summarize discussions and recommendations for each of these themes., Comment: arXiv admin note: substantial text overlap with arXiv:2103.09181

While standard generative adversarial networks (GANs) rely solely on training data to learn unknown probability distributions, physics-informed GANs (PI-GANs) encode physical laws in the form of stochastic partial differential equations (PDEs) using auto differentiation. By relating observed data to unobserved quantities of interest through PDEs, PI-GANs allow for the estimation of underlying probability distributions without their direct measurement (i.e. inverse problems). The scalable nature of GANs allows high-dimensional, spatially-dependent probability distributions (i.e., random fields) to be inferred, while incorporating prior information through PDEs allows the training datasets to be relatively small. In this work, PI-GANs are demonstrated for the application of elastic modulus estimation in mechanical testing. Given measured deformation data, the underlying probability distribution of spatially-varying elastic modulus (stiffness) is learned. Two feed-forward deep neural network generators are used to model the deformation and material stiffness across a two dimensional domain. Wasserstein GANs with gradient penalty are employed for enhanced stability. In the absence of explicit training data, it is demonstrated that the PI-GAN learns to generate realistic, physically-admissible realizations of material stiffness by incorporating the PDE that relates it to the measured deformation. It is shown that the statistics (mean, standard deviation, point-wise distributions, correlation length) of these generated stiffness samples have good agreement with the true distribution., Comment: 14 pages, 6 figures

Scientific workflows are a cornerstone of modern scientific computing, and they have underpinned some of the most significant discoveries of the last decade. Many of these workflows have high computational, storage, and/or communication demands, and thus must execute on a wide range of large-scale platforms, from large clouds to upcoming exascale HPC platforms. Workflows will play a crucial role in the data-oriented and post-Moore's computing landscape as they democratize the application of cutting-edge research techniques, computationally intensive methods, and use of new computing platforms. As workflows continue to be adopted by scientific projects and user communities, they are becoming more complex. Workflows are increasingly composed of tasks that perform computations such as short machine learning inference, multi-node simulations, long-running machine learning model training, amongst others, and thus increasingly rely on heterogeneous architectures that include CPUs but also GPUs and accelerators. The workflow management system (WMS) technology landscape is currently segmented and presents significant barriers to entry due to the hundreds of seemingly comparable, yet incompatible, systems that exist. Another fundamental problem is that there are conflicting theoretical bases and abstractions for a WMS. Systems that use the same underlying abstractions can likely be translated between, which is not the case for systems that use different abstractions. More information: https://workflowsri.org/summits/technical

Precise real time estimates of earthquake magnitude and location are essential for early warning and rapid response. While recently multiple deep learning approaches for fast assessment of earthquakes have been proposed, they usually rely on either seismic records from a single station or from a fixed set of seismic stations. Here we introduce a new model for real-time magnitude and location estimation using the attention based transformer networks. Our approach incorporates waveforms from a dynamically varying set of stations and outperforms deep learning baselines in both magnitude and location estimation performance. Furthermore, it outperforms a classical magnitude estimation algorithm considerably and shows promising performance in comparison to a classical localization algorithm. In this work, we furthermore conduct a comprehensive study of the requirements on training data, the training procedures and the typical failure modes using three diverse and large scale data sets. Our analysis gives several key insights. First, we can precisely pinpoint the effect of large training data; for example, a four times larger training set reduces the required time for real time assessment by a factor of four. Second, the basic model systematically underestimates large magnitude events. This issue can be mitigated by incorporating events from other regions into the training through transfer learning. Third, location estimation is highly precise in areas with sufficient training data, but is strongly degraded for events outside the training distribution. Our analysis suggests that these characteristics are not only present for our model, but for most deep learning models for fast assessment published so far. They result from the black box modeling and their mitigation will likely require imposing physics derived constraints on the neural network., Comment: 24 pages main text, 20 pages appendix. This article has been accepted for publication in Geophysical Journal International Published by Oxford University Press on behalf of the Royal Astronomical Society

Earthquakes are major hazards to humans, buildings and infrastructure. Early warning methods aim to provide advance notice of incoming strong shaking to enable preventive action and mitigate seismic risk. Their usefulness depends on accuracy, the relation between true, missed and false alerts, and timeliness, the time between a warning and the arrival of strong shaking. Current approaches suffer from apparent aleatoric uncertainties due to simplified modelling or short warning times. Here we propose a novel early warning method, the deep-learning based transformer earthquake alerting model (TEAM), to mitigate these limitations. TEAM analyzes raw, strong motion waveforms of an arbitrary number of stations at arbitrary locations in real-time, making it easily adaptable to changing seismic networks and warning targets. We evaluate TEAM on two regions with high seismic hazard, Japan and Italy, that are complementary in their seismicity. On both datasets TEAM outperforms existing early warning methods considerably, offering accurate and timely warnings. Using domain adaptation, TEAM even provides reliable alerts for events larger than any in the training data, a property of highest importance as records from very large events are rare in many regions., Comment: 13 pages main text, 16 pages appendix. This article has been accepted for publication in Geophysical Journal International Published by Oxford University Press on behalf of the Royal Astronomical Society

Tables are a popular and efficient means of presenting structured information. They are used extensively in various kinds of documents including web pages. Tables display information as a two-dimensional matrix, the semantics of which is conveyed by a mixture of structure (rows, columns), headers, caption, and content. Recent research has started to consider tables as first class objects, not just as an addendum to texts, yielding interesting results for problems like table matching, table completion, or value imputation. All of these problems inherently rely on an accurate measure for the semantic similarity of two tables. We present TabSim, a novel method to compute table similarity scores using deep neural networks. Conceptually, TabSim represents a table as a learned concatenation of embeddings of its caption, its content, and its structure. Given two tables in this representation, a Siamese neural network is trained to compute a score correlating with the tables' semantic similarity. To train and evaluate our method, we created a gold standard corpus consisting of 1500 table pairs extracted from biomedical articles and manually scored regarding their degree of similarity, and adopted two other corpora originally developed for a different yet similar task. Our evaluation shows that TabSim outperforms other table similarity measures on average by app. 7% pp F1-score in a binary similarity classification setting and by app. 1.5% pp in a ranking scenario.

Summary: Named Entity Recognition (NER) is an important step in biomedical information extraction pipelines. Tools for NER should be easy to use, cover multiple entity types, highly accurate, and robust towards variations in text genre and style. To this end, we propose HunFlair, an NER tagger covering multiple entity types integrated into the widely used NLP framework Flair. HunFlair outperforms other state-of-the-art standalone NER tools with an average gain of 7.26 pp over the next best tool, can be installed with a single command and is applied with only four lines of code. Availability: HunFlair is freely available through the Flair framework under an MIT license: https://github.com/flairNLP/flair and is compatible with all major operating systems. Contact:{weberple,saengema,alan.akbik}@informatik.hu-berlin.de, Comment: - Corrected author list - Updated project link

Dwarf spheroidal galaxy satellites of the Milky Way are prime targets for indirect detection of dark matter with gamma rays due to their proximity, high dark matter content and absence of non-thermal emission processes. Recently, the Dark Energy Survey (DES) revealed the existence of new ultra-faint dwarf spheroidal galaxies in the southern-hemisphere sky, therefore ideally located for ground-based observations with the imaging atmospheric Cherenkov telescope array H.E.S.S. We present a search for very-high-energy ($E\gtrsim100$ GeV) gamma-ray emission using H.E.S.S. observations carried out recently towards Reticulum II, Tucana II, Tucana III, Tucana IV and Grus II satellites. No significant very-high-energy gamma-ray excess is found from the observations on any individual object nor in the combined analysis of all the datasets. Using the most recent modeling of the dark matter distribution in the dwarf galaxy halo, we compute for the first time on DES satellites individual and combined constraints from Cherenkov telescope observations on the annihilation cross section of dark matter particles in the form of Weakly Interacting Massive Particles. The combined 95% C.L. observed upper limits reach $\langle \sigma v \rangle \simeq 1 \times 10^{-23}$ cm$^3$s$^{-1}$ in the $W^+W^-$ channel and $4 \times 10^{-26}$ cm$^3$s$^{-1}$ in the $\gamma\gamma$ channels for a dark matter mass of 1.5 TeV. The H.E.S.S. constraints well complement the results from Fermi-LAT, HAWC, MAGIC and VERITAS and are currently the most stringent in the $\gamma\gamma$ channels in the multi-GeV/multi-TeV mass range., Comment: 20 pages, 12 figures

The nearby radio galaxy Centaurus A belongs to a class of Active Galaxies that are very luminous at radio wavelengths. The majority of these galaxies show collimated relativistic outflows known as jets, that extend over hundreds of thousands of parsecs for the most powerful sources. Accretion of matter onto the central super-massive black hole is believed to fuel these jets and power their emission, with the radio emission being related to the synchrotron radiation of relativistic electrons in magnetic fields. The origin of the extended X-ray emission seen in the kiloparsec-scale jets from these sources is still a matter of debate, although Cen A's X-ray emission has been suggested to originate in electron synchrotron processes. The other possible explanation is Inverse Compton (IC) scattering with CMB soft photons. Synchrotron radiation needs ultra-relativistic electrons ($\sim50$ TeV), and given their short cooling times, requires some continuous re-acceleration mechanism to be active. IC scattering, on the other hand, does not require very energetic electrons, but requires jets that stay highly relativistic on large scales ($\geq$1 Mpc) and that remain well-aligned with the line of sight. Some recent evidence disfavours inverse Compton-CMB models, although other evidence seems to be compatible with them. In principle, the detection of extended gamma-ray emission, directly probing the presence of ultra-relativistic electrons, could distinguish between these options, but instruments have hitherto been unable to resolve the relevant structures. At GeV energies there is also an unusual spectral hardening in Cen A, whose explanation is unclear. Here we report observations of Cen A at TeV energies that resolve its large-scale jet. We interpret the data as evidence for the acceleration of ultra-relativistic electrons in the jet, and favour the synchrotron explanation for the X-rays., Comment: 32 pages, 5 figure

The analysis of next-generation sequencing (NGS) data requires complex computational workflows consisting of dozens of autonomously developed yet interdependent processing steps. Whenever large amounts of data need to be processed, these workflows must be executed on a parallel and/or distributed systems to ensure reasonable runtime. Porting a workflow developed for a particular system on a particular hardware infrastructure to another system or to another infrastructure is non-trivial, which poses a major impediment to the scientific necessities of workflow reproducibility and workflow reusability. In this work, we describe our efforts to port a state-of-the-art workflow for the detection of specific variants in whole-exome sequencing of mice. The workflow originally was developed in the scientific workflow system snakemake for execution on a high-performance cluster controlled by Sun Grid Engine. In the project, we ported it to the scientific workflow system SaasFee that can execute workflows on (multi-core) stand-alone servers or on clusters of arbitrary sizes using the Hadoop. The purpose of this port was that also owners of low-cost hardware infrastructures, for which Hadoop was made for, become able to use the workflow. Although both the source and the target system are called scientific workflow systems, they differ in numerous aspects, ranging from the workflow languages to the scheduling mechanisms and the file access interfaces. These differences resulted in various problems, some expected and more unexpected, that had to be resolved before the workflow could be run with equal semantics. As a side-effect, we also report cost/runtime ratios for a state-of-the-art NGS workflow on very different hardware platforms: A comparably cheap stand-alone server (80 threads), a mid-cost, mid-sized cluster (552 threads), and a high-end HPC system (3784 threads).

The detection of the first electromagnetic counterpart to the binary neutron star (BNS) merger remnant GW170817 established the connection between short $\gamma$-ray bursts and BNS mergers. It also confirmed the forging of heavy elements in the ejecta (a so-called kilonova) via the r-process nucleosynthesis. The appearance of non-thermal radio and X-ray emission, as well as the brightening, which lasted more than 100 days, were somewhat unexpected. Current theoretical models attempt to explain this temporal behavior as either originating from a relativistic off-axis jet or a kilonova-like outflow. In either scenario, there is some ambiguity regarding how much energy is transported in the non-thermal electrons versus the magnetic field of the emission region. Combining the VLA (radio) and Chandra (X-ray) measurements with observations in the GeV-TeV domain can help break this ambiguity, almost independently of the assumed origin of the emission. Here we report for the first time on deep H.E.S.S. observations of GW170817 / GRB 170817A between 124 and 272 days after the BNS merger with the full H.E.S.S. array of telescopes, as well as on an updated analysis of the prompt (<5 days) observations with the upgraded H.E.S.S. phase-I telescopes. We discuss implications of the H.E.S.S. measurement for the magnetic field in the context of different source scenarios., Comment: 7 pages, 2 figures, 1 table

We report on the detection of very-high-energy (VHE; $E > 100$ GeV) $\gamma$-ray emission from the BL Lac objects KUV 00311-1938 and PKS 1440-389 with the High Energy Stereoscopic System (H.E.S.S.). H.E.S.S. observations were accompanied or preceded by multi-wavelength observations with Fermi/LAT, XRT and UVOT on board the Swift satellite, and ATOM. Based on an extrapolation of the Fermi/LAT spectrum towards the VHE $\gamma$-ray regime, we deduce a 95% confidence level upper limit on the unknown redshift of KUV 00311-1938 of z < 0.98, and of PKS 1440-389 of z < 0.53. When combined with previous spectroscopy results the redshift of KUV 00311-1938 is constrained to $0.51 \leq z < 0.98$ and for PKS 1440-389 to $0.14 \lessapprox z < 0.53$., Comment: Accepted for publication in MNRAS. 13 pages, 4 figures

Aims. Colliding wind binary systems have long been suspected to be high-energy (HE; 100 MeV < E < 100 GeV) {\gamma}-ray emitters. {\eta} Car is the most prominent member of this object class and is confirmed to emit phase-locked HE {\gamma} rays from hundreds of MeV to ~100 GeV energies. This work aims to search for and characterise the very-high-energy (VHE; E >100 GeV) {\gamma}-ray emission from {\eta} Car around the last periastron passage in 2014 with the ground-based High Energy Stereoscopic System (H.E.S.S.). Methods. The region around {\eta} Car was observed with H.E.S.S. between orbital phase p = 0.78 - 1.10, with a closer sampling at p {\approx} 0.95 and p {\approx} 1.10 (assuming a period of 2023 days). Optimised hardware settings as well as adjustments to the data reduction, reconstruction, and signal selection were needed to suppress and take into account the strong, extended, and inhomogeneous night sky background (NSB) in the {\eta} Car field of view. Tailored run-wise Monte-Carlo simulations (RWS) were required to accurately treat the additional noise from NSB photons in the instrument response functions. Results. H.E.S.S. detected VHE {\gamma}-ray emission from the direction of {\eta} Car shortly before and after the minimum in the X-ray light-curve close to periastron. Using the point spread function provided by RWS, the reconstructed signal is point-like and the spectrum is best described by a power law. The overall flux and spectral index in VHE {\gamma} rays agree within statistical and systematic errors before and after periastron. The {\gamma}-ray spectrum extends up to at least ~400 GeV. This implies a maximum magnetic field in a leptonic scenario in the emission region of 0.5 Gauss. No indication for phase-locked flux variations is detected in the H.E.S.S. data., Comment: 9 pages, 4 figures, 3 tables, in press with A&A

PSR B1259-63/LS 2883 is a gamma-ray binary system consisting of a pulsar in an eccentric orbit around a bright Oe stellar-type companion star that features a dense circumstellar disc. The high- and very-high-energy (HE, VHE) gamma-ray emission from PSR B1259-63/LS 2883 around the times of its periastron passage are characterised, in particular, at the time of the HE gamma-ray flares reported to have occurred in 2011, 2014, and 2017. Spectra and light curves were derived from observations conducted with the H.E.S.S.-II array in 2014 and 2017. A local double-peak profile with asymmetric peaks in the VHE light curve is measured, with a flux minimum at the time of periastron $t_p$ and two peaks coinciding with the times at which the neutron star crosses the companion's circumstellar disc ($\sim t_p \pm 16$ d). A high VHE gamma-ray flux is also observed at the times of the HE gamma-ray flares ($\sim t_p + 30$ d) and at phases before the first disc crossing ($\sim t_p - 35$ d). PSR B1259-63/LS 2883 displays periodic flux variability at VHE gamma-rays without clear signatures of super-orbital modulation in the time span covered by H.E.S.S. observations. In contrast, the photon index of the measured power-law spectra remains unchanged within uncertainties for about 200 d around periastron. Lower limits on exponential cut-off energies up to $\sim 40$ TeV are placed. At HE gamma-rays, PSR B1259-63/LS 2883 has now been detected also before and after periastron, close to the disc crossing times. Repetitive flares with distinct variability patterns are detected in this energy range. Such outbursts are not observed at VHEs, although a relatively high emission level is measured. The spectra obtained in both energy regimes displays a similar slope, although a common physical origin either in terms of a related particle population, emission mechanism, or emitter location is ruled out., Comment: 16 pages, 7 figures Accepted for publication in A&A

Gamma-ray bursts (GRBs) are brief flashes of gamma rays, considered to be the most energetic explosive phenomena in the Universe. The emission from GRBs comprises a short (typically tens of seconds) and bright prompt emission, followed by a much longer afterglow phase. During the afterglow phase, the shocked outflow -- produced by the interaction between the ejected matter and the circumburst medium -- slows down, and a gradual decrease in brightness is observed. GRBs typically emit most of their energy via gamma-rays with energies in the kiloelectronvolt-to-megaelectronvolt range, but a few photons with energies of tens of gigaelectronvolts have been detected by space-based instruments. However, the origins of such high-energy (above one gigaelectronvolt) photons and the presence of very-high-energy (more than 100 gigaelectronvolts) emission have remained elussive. Here we report observations of very-high-energy emission in the bright GRB 180720B deep in the GRB afterglow -ten hours after the end of the prompt emission phase, when the X-ray flux had already decayed by four orders of magnitude. Two possible explanations exist for the observed radiation: inverse Compton emission and synchrotron emission of ultrarelativistic electrons. Our observations show that the energy fluxes in the X-ray and gamma-ray range and their photon indices remain comparable to each other throughout the afterglow. This discovery places distinct constraints on the GRB environment for both emission mechanisms, with the inverse Compton explanation alleviating the particle energy requirements for the emission observed at late times. The late timing of this detection has consequences for the future observations of GRBs at the highest energies., Comment: Preprint version of Nature paper. Contacts: E.Ruiz-Velasco, F. Aharonian, E.Bissaldi, C.Hoischen, R.D Parsons, Q.Piel, A.Taylor, D.Khangulyan

Flat-spectrum radio-quasars (FSRQs) are rarely detected at very-high-energies (VHE; E>100 GeV) due to their low-frequency-peaked SEDs. At present, only 6 FSRQs are known to emit VHE photons, representing only 7% of the VHE extragalactic catalog. Following the detection of MeV-GeV gamma-ray flaring activity from the FSRQ PKS 0736+017 (z=0.189) with Fermi, the H.E.S.S. array of Cherenkov telescopes triggered ToO observations on February 18, 2015, with the goal of studying the gamma-ray emission in the VHE band. H.E.S.S. ToO observations were carried out during the nights of February 18, 19, 21, and 24, 2015. Together with Fermi-LAT, the multi-wavelength coverage of the flare includes Swift observations in soft-X-rays and optical/UV, and optical monitoring (photometry and spectro-polarimetry) by the Steward Observatory, the ATOM, the KAIT and the ASAS-SN telescope. VHE emission from PKS 0736+017 was detected with H.E.S.S. during the night of February 19, 2015, only. Fermi data indicate the presence of a gamma-ray flare, peaking at the time of the H.E.S.S. detection, with a flux doubling time-scale of around six hours. The gamma-ray flare was accompanied by at least a 1 mag brightening of the non-thermal optical continuum. No simultaneous observations at longer wavelengths are available for the night of the H.E.S.S. detection. The gamma-ray observations with H.E.S.S. and Fermi are used to put constraints on the location of the gamma-ray emitting region during the flare: it is constrained to be just outside the radius of the broad-line-region with a bulk Lorentz factor $\simeq 20$, or at the level of the radius of the dusty torus with Gamma > 60. PKS 0736+017 is the seventh FSRQ known to emit VHE photons and, at z=0.189, is the nearest so far. The location of the gamma-ray emitting region during the flare can be tightly constrained thanks to opacity, variability, and collimation arguments., Comment: In press in A&A. 13 pages, 5 figures. Abstract abridged

The Crab nebula is one of the most studied cosmic particle accelerators, shining brightly across the entire electromagnetic spectrum up to very high-energy gamma rays. It is known from radio to gamma-ray observations that the nebula is powered by a pulsar, which converts most of its rotational energy losses into a highly relativistic outflow. This outflow powers a pulsar wind nebula (PWN), a region of up to 10~light-years across, filled with relativistic electrons and positrons. These particles emit synchrotron photons in the ambient magnetic field and produce very high-energy gamma rays by Compton up-scattering of ambient low-energy photons. While the synchrotron morphology of the nebula is well established, it was up to now not known in which region the very high-energy gamma rays are emitted. Here we report that the Crab nebula has an angular extension at gamma-ray energies of 52 arcseconds (assuming a Gaussian source width), significantly larger than at X-ray energies. This result closes a gap in the multi-wavelength coverage of the nebula, revealing the emission region of the highest energy gamma rays. These gamma rays are a new probe of a previously inaccessible electron and positron energy range. We find that simulations of the electromagnetic emission reproduce our new measurement, providing a non-trivial test of our understanding of particle acceleration in the Crab nebula., Comment: 23 pages, 5 figures

Early time series classification (eTSC) is the problem of classifying a time series after as few measurements as possible with the highest possible accuracy. The most critical issue of any eTSC method is to decide when enough data of a time series has been seen to take a decision: Waiting for more data points usually makes the classification problem easier but delays the time in which a classification is made; in contrast, earlier classification has to cope with less input data, often leading to inferior accuracy. The state-of-the-art eTSC methods compute a fixed optimal decision time assuming that every times series has the same defined start time (like turning on a machine). However, in many real-life applications measurements start at arbitrary times (like measuring heartbeats of a patient), implying that the best time for taking a decision varies heavily between time series. We present TEASER, a novel algorithm that models eTSC as a two two-tier classification problem: In the first tier, a classifier periodically assesses the incoming time series to compute class probabilities. However, these class probabilities are only used as output label if a second-tier classifier decides that the predicted label is reliable enough, which can happen after a different number of measurements. In an evaluation using 45 benchmark datasets, TEASER is two to three times earlier at predictions than its competitors while reaching the same or an even higher classification accuracy. We further show TEASER's superior performance using real-life use cases, namely energy monitoring, and gait detection.

Rule-based models are attractive for various tasks because they inherently lead to interpretable and explainable decisions and can easily incorporate prior knowledge. However, such systems are difficult to apply to problems involving natural language, due to its linguistic variability. In contrast, neural models can cope very well with ambiguity by learning distributed representations of words and their composition from data, but lead to models that are difficult to interpret. In this paper, we describe a model combining neural networks with logic programming in a novel manner for solving multi-hop reasoning tasks over natural language. Specifically, we propose to use a Prolog prover which we extend to utilize a similarity function over pretrained sentence encoders. We fine-tune the representations for the similarity function via backpropagation. This leads to a system that can apply rule-based reasoning to natural language, and induce domain-specific rules from training data. We evaluate the proposed system on two different question answering tasks, showing that it outperforms two baselines -- BIDAF (Seo et al., 2016a) and FAST QA (Weissenborn et al., 2017b) on a subset of the WikiHop corpus and achieves competitive results on the MedHop data set (Welbl et al., 2017)., Comment: ACL 2019