Thong Nguyen

A bstract Using variational autoencoders trained on known physics processes, we develop a one-sided threshold test to isolate previously unseen processes as outlier events. Since the autoencoder training does not depend on any specific new physics signature, the proposed procedure doesn’t make specific assumptions on the nature of new physics. An event selection based on this algorithm would be complementary to classic LHC searches, typically based on model-dependent hypothesis testing. Such an algorithm would deliver a list of anomalous events, that the experimental collaborations could further scrutinize and even release as a catalog, similarly to what is typically done in other scientific domains. Event topologies repeating in this dataset could inspire new-physics model building and new experimental searches. Running in the trigger system of the LHC experiments, such an application could identify anomalous events that would be otherwise lost, extending the scientific reach of the LHC.

Abstract We investigate the performance of a jet identification algorithm based on interaction networks (JEDI-net) to identify all-hadronic decays of high-momentum heavy particles produced at the LHC and distinguish them from ordinary jets originating from the hadronization of quarks and gluons. The jet dynamics are described as a set of one-to-one interactions between the jet constituents. Based on a representation learned from these interactions, the jet is associated to one of the considered categories. Unlike other architectures, the JEDI-net models achieve their performance without special handling of the sparse input jet representation, extensive pre-processing, particle ordering, or specific assumptions regarding the underlying detector geometry. The presented models give better results with less model parameters, offering interesting prospects for LHC applications.

Abstract We apply an Adversarially Learned Anomaly Detection (ALAD) algorithm to the problem of detecting new physics processes in proton–proton collisions at the Large Hadron Collider. Anomaly detection based on ALAD matches performances reached by Variational Autoencoders, with a substantial improvement in some cases. Training the ALAD algorithm on 4.4 fb $$^{-1}$$ <mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML"> <mml:msup> <mml:mrow /> <mml:mrow> <mml:mo>-</mml:mo> <mml:mn>1</mml:mn> </mml:mrow> </mml:msup> </mml:math> of 8 TeV CMS Open Data, we show how a data-driven anomaly detection and characterization would work in real life, re-discovering the top quark by identifying the main features of the $$t \bar{t}$$ <mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML"> <mml:mrow> <mml:mi>t</mml:mi> <mml:mover> <mml:mrow> <mml:mi>t</mml:mi> </mml:mrow> <mml:mrow> <mml:mo>¯</mml:mo> </mml:mrow> </mml:mover> </mml:mrow> </mml:math> experimental signature at the LHC.

We develop an algorithm based on an interaction network to identify high-transverse-momentum Higgs bosons decaying to bottom quark-antiquark pairs and distinguish them from ordinary jets that reflect the configurations of quarks and gluons at short distances. The algorithm's inputs are features of the reconstructed charged particles in a jet and the secondary vertices associated with them. Describing the jet shower as a combination of particle-to-particle and particle-to-vertex interactions, the model is trained to learn a jet representation on which the classification problem is optimized. The algorithm is trained on simulated samples of realistic LHC collisions, released by the CMS Collaboration on the CERN Open Data Portal. The interaction network achieves a drastic improvement in the identification performance with respect to state-of-the-art algorithms.

To study the physics of fundamental particles and their interactions, the Large Hadron Collider was constructed at CERN, where protons collide to create new particles measured by detectors. Collisions occur at a frequency of 40 MHz, and with an event size of roughly 1 MB it is impossible to read out and store the generated amount of data from the detector and therefore a multi-tiered, real-time filtering system is required. In this paper, we show how to adapt and deploy deep-learning-based autoencoders for the unsupervised detection of new physics signatures in the challenging environment of a real-time event selection system at the Large Hadron Collider. The first-stage filter, implemented on custom electronics, decides within a few microseconds whether an event should be kept or discarded. At this stage, the rate is reduced from 40 MHz to about 100 kHz. We demonstrate the deployment of an unsupervised selection algorithm on this custom electronics, running in as little as 80 ns and enhancing the signal-over-background ratio by three orders of magnitude. This work enables the practical deployment of these networks during the next data-taking campaign of the Large Hadron Collider. The Large Hadron Collider produces 40 million collision events per second, most of which need to be discarded by a real-time filtering system. Unsupervised deep learning algorithms are developed and deployed on custom electronics to search for rare events indicating new physics, rather than for specific events led by theory.

Learned sparse retrieval (LSR) is a family of first-stage retrieval methods that are trained to generate sparse lexical representations of queries and documents for use with an inverted index. Many LSR methods have been recently introduced, with Splade models achieving state-of-the-art performance on MSMarco. Despite similarities in their model architectures, many LSR methods show substantial differences in effectiveness and efficiency. Differences in the experimental setups and configurations used make it difficult to compare the methods and derive insights. In this work, we analyze existing LSR methods and identify key components to establish an LSR framework that unifies all LSR methods under the same perspective. We then reproduce all prominent methods using a common codebase and re-train them in the same environment, which allows us to quantify how components of the framework affect effectiveness and efficiency. We find that (1) including document term weighting is most important for a method's effectiveness, (2) including query weighting has a small positive impact, and (3) document expansion and query expansion have a cancellation effect. As a result, we show how removing query expansion from a state-of-the-art model can reduce latency significantly while maintaining effectiveness on MSMarco and TripClick benchmarks. Our code is publicly available (Code: https://github.com/thongnt99/learned-sparse-retrieval ).

This paper proposes a new ECG signal compression algorithm using a discrete symmetric wavelet transform. This proposed compression scheme may find applications in digital Holter recording, in ECG signal archiving and in ECG data transmission through communication channels. Using the new method, a compression ratio of 8 to 1 can be achieved with PRD=3.9%, in contrast to the AZTEC compression ratio of 6.8 to 1 with PRD=10.0% and the fan algorithm compression ratio of 7;4 to 1 with PRD=8.1%.

We investigate how a Generative Adversarial Network could be used to generate a list of particle four-momenta from LHC proton collisions, allowing one to define a generative model that could abstract from the irregularities of typical detector geometries. As an example of application, we show how such an architecture could be used as a generator of LHC parasitic collisions (pileup). We present two approaches to generate the events: unconditional generator and generator conditioned on missing transverse energy. We assess generation performances in a realistic LHC data-analysis environment, with a pileup mitigation algorithm applied.

Evidence is accumulating that sleep duration is related to blood pressure (BP) and hypertensive status, but the strength of the association varies by age, and findings are inconsistent for adolescents. This cross-sectional study tested the hypothesis that sleep duration, both during the night and during naps, would be negatively associated with ambulatory systolic BP (SBP) and diastolic BP (DBP) measured over 24 hours in adolescents.In this ethnically diverse (37% non-Hispanic black, 31% Hispanic, 29% non-Hispanic white, 3% other), school-based sample of 366 adolescents aged 11-16 years, ambulatory BP was measured every 30 minutes for 24 hours on a school day; actigraphy was used to measure sleep duration. Covariables included demographic factors, anthropometric indices, physical activity, and position and location at the time of each BP measurement. Mixed models were used to test day and night sleep duration as predictors of 24-hour SBP and DBP, controlling for covariables.The mean sleep duration was 6.83 (SD = 1.36) hours at night, and 7.23 (SD = 1.67) hours over 24 hours. Controlling for duration of sleep during the day and covariables, each additional hour of nighttime sleep was associated with lower SBP (-0.57; P < 0.0001); controlling for nighttime sleep duration and covariables, each additional hour of daytime sleep was associated with lower SBP (-0.73; P < 0.001) and lower DBP (-0.50; P < 0.001).Longer sleep duration was significantly associated with lower ambulatory SBP and DBP in adolescents. The findings have potential implications for cardiovascular health in this age group.

The purpose of this study of school-age children was to estimate prevalence and interrelationships of overweight, central adiposity, and hypertension. It included 1,070 children in kindergarten through sixth grade (67% Hispanic, 26% African American, mean age = 8.9 years). Measures included body mass index (BMI), waist circumference (WC), systolic and/or diastolic hypertension identified by measurements on three separate occasions. Percentage overweight (BMI >or=95th percentile) was 28.7%, 17.9% were at risk of overweight, 28.8% had WC >or=90th percentile, and 9.4% had elevated (>or=90th percentile) systolic and/or diastolic blood pressure (BP). If we had screened only for BMI and examined those with BMI >or=85th percentile or underweight for hypertension, we would have missed 26% of the children with persistently elevated BP. WC explained variance in elevated BP not explained by BMI (p < .001). Measurement of WC is easily incorporated in a school-based screening protocol.

We show how an event topology classification based on deep learning could be used to improve the purity of data samples selected in real time at the Large Hadron Collider. We consider different data representations, on which different kinds of multi-class classifiers are trained. Both raw data and high-level features are utilized. In the considered examples, a filter based on the classifier’s score can be trained to retain $$\sim 99\%$$ of the interesting events and reduce the false-positive rate by more than one order of magnitude. By operating such a filter as part of the online event selection infrastructure of the LHC experiments, one could benefit from a more flexible and inclusive selection strategy while reducing the amount of downstream resources wasted in processing false positives. The saved resources could translate into a reduction of the detector operation cost or into an effective increase of storage and processing capabilities, which could be reinvested to extend the physics reach of the LHC experiments.

Automated methods have been widely used to identify and analyze mental health conditions (e.g., depression) from various sources of information, including social media.Yet, deployment of such models in real-world healthcare applications faces challenges including poor out-of-domain generalization and lack of trust in black box models.In this work, we propose approaches for depression detection that are constrained to different degrees by the presence of symptoms described in PHQ9, a questionnaire used by clinicians in the depression screening process.In dataset-transfer experiments on three social media datasets, we find that grounding the model in PHQ9's symptoms substantially improves its ability to generalize to out-of-distribution data compared to a standard BERT-based approach.Furthermore, this approach can still perform competitively on indomain data.These results and our qualitative analyses suggest that grounding model predictions in clinically-relevant symptoms can improve generalizability while producing a model that is easier to inspect.

Learned Sparse Retrieval (LSR) is a group of neural methods designed to encode queries and documents into sparse lexical vectors. These vectors can be efficiently indexed and retrieved using an inverted index. While LSR has shown promise in text retrieval, its potential in multi-modal retrieval remains largely unexplored. Motivated by this, in this work, we explore the application of LSR in the multi-modal domain, i.e., we focus on Multi-Modal Learned Sparse Retrieval (MLSR). We conduct experiments using several MLSR model configurations and evaluate the performance on the image suggestion task. We find that solving the task solely based on the image content is challenging. Enriching the image content with its caption improves the model performance significantly, implying the importance of image captions to provide fine-grained concepts and context information of images. Our approach presents a practical and effective solution for training LSR retrieval models in multi-modal settings.

Learned sparse retrieval (LSR) is a family of neural methods that encode queries and documents into sparse lexical vectors that can be indexed and retrieved efficiently with an inverted index. We explore the application of LSR to the multi-modal domain, with a focus on text-image retrieval. While LSR has seen success in text retrieval, its application in multimodal retrieval remains underexplored. Current approaches like LexLIP and STAIR require complex multi-step training on massive datasets. Our proposed approach efficiently transforms dense vectors from a frozen dense model into sparse lexical vectors. We address issues of high dimension co-activation and semantic deviation through a new training algorithm, using Bernoulli random variables to control query expansion. Experiments with two dense models (BLIP, ALBEF) and two datasets (MSCOCO, Flickr30k) show that our proposed algorithm effectively reduces co-activation and semantic deviation. Our best-performing sparsified model outperforms state-of-the-art text-image LSR models with a shorter training time and lower GPU memory requirements. Our approach offers an effective solution for training LSR retrieval models in multimodal settings. Our code and model checkpoints are available at github.com/thongnt99/lsr-multimodal

Fully fine-tuning pretrained large-scale transformer models has become a popular paradigm for video-language modeling tasks, such as temporal language grounding and video-language summarization. With a growing number of tasks and limited training data, such full fine-tuning approach leads to costly model storage and unstable training. To overcome these shortcomings, we introduce lightweight adapters to the pre-trained model and only update them at fine-tuning time. However, existing adapters fail to capture intrinsic temporal relations among video frames or textual words. Moreover, they neglect the preservation of critical task-related information that flows from the raw video-language input into the adapter’s low-dimensional space. To address these issues, we first propose a novel REcurrent ADapter (READ) that employs recurrent computation to enable temporal modeling capability. Second, we propose Partial Video-Language Alignment (PVLA) objective via the use of partial optimal transport to maintain task-related information flowing into our READ modules. We validate our READ-PVLA framework through extensive experiments where READ-PVLA significantly outperforms all existing fine-tuning strategies on multiple low-resource temporal language grounding and video-language summarization benchmarks.

We present a fast-simulation application based on a deep neural network, designed to create large analysis-specific datasets. Taking as an example the generation of W + jet events produced in s= 13 TeV proton-proton collisions, we train a neural network to model detector resolution effects as a transfer function acting on an analysis-specific set of relevant features, computed at generation level, i.e., in absence of detector effects. Based on this model, we propose a novel fast-simulation workflow that starts from a large amount of generator-level events to deliver large analysis-specific samples. The adoption of this approach would result in about an order-of-magnitude reduction in computing and storage requirements for the collision simulation workflow. This strategy could help the high energy physics community to face the computing challenges of the future High-Luminosity LHC.

Abstract Topic models aim to discover latent topics and infer what topics a document contains in an unsupervised fashion. They have been prevalent for decades with wide applications like text analysis. Recently the rise of neural networks has facilitated a new research field---Neural Topic Models (NTMs). Different from conventional models, NTMs directly optimize parameters without model-specific derivations. This endows NTMs with better scalability and flexibility, resulting in significant research attention and plentiful new methods and applications. In this paper, we present a thorough survey on neural topic models carefully and widely. In detail, we systematically organize current NTM methods according to their network structures and introduce the NTMs for various scenarios like short texts and cross-lingual documents. Then we discuss developed popular applications based on NTMs. We finally highlight the challenges of NTMs to motivate future research.

The rapid expansion of user medical records across global systems presents not only opportunities but also new challenges in maintaining effective application models that ensure user privacy, controllability, and the ability to commercialize patient medical records. Moreover, the proliferation of data analysis models in healthcare institutions necessitates the decentralization and restorability of medical record data. It is imperative that user medical data collected from these systems can be easily analyzed and utilized even years after collection, without the risk of data loss due to numerous factors. Additionally, medical information must be authorized by the data owner, granting patients the right to accept or decline data usage requests from medical research agencies. In response, we propose an innovative solution for implementing a decentralized system utilizing an EVM-compatible blockchain and IPFS for decentralized storage. To ensure privacy and control, we employ Proxy Re-Encryption (PRE), a cryptographic authorized method, within the medical data marketplace. Our proposed architecture significantly reduces costs associated with granting read access to healthcare research agencies by minimizing the encryption and decryption time of stored records. Furthermore, it empowers users with enhanced control over their health data through tamperproof blockchain smart contracts and IPFS, safeguarding the integrity and privacy of their medical records.

Download This Paper Open PDF in Browser Add Paper to My Library Share: Permalink Using these links will ensure access to this page indefinitely Copy URL Copy DOI

The paper describes a method of classifying natural textures based on maximum likelihood parameter estimation technique. The wavelet transform (WT) is used to represent the textural images in multiresolution. Co-occurrence matrices are then computed for the different scales of the wavelet transform and textural features are obtained from the co-occurrence matrices. Then a maximum likelihood classifier is designed using a set of training texture samples. Ten different Brodot textures have been classified using this procedure with an average classification accuracy of 99.7.< <ETX xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">&gt;</ETX>

In recent years, several studies have demonstrated the benefit of using deep learning to solve typical tasks related to high energy physics data taking and analysis. In particular, generative adversarial networks are a good candidate to supplement the simulation of the detector response in a collider environment. Training of neural network models has been made tractable with the improvement of optimization methods and the advent of GP-GPU well adapted to tackle the highly-parallelizable task of training neural nets. Despite these advancements, training of large models over large data sets can take days to weeks. Even more so, finding the best model architecture and settings can take many expensive trials. To get the best out of this new technology, it is important to scale up the available network-training resources and, consequently, to provide tools for optimal large-scale distributed training. In this context, our development of a new training workflow, which scales on multi-node/multi-GPU architectures with an eye to deployment on high performance computing machines is described. We describe the integration of hyper parameter optimization with a distributed training framework using Message Passing Interface, for models defined in keras [12] or pytorch [13]. We present results on the speedup of training generative adversarial networks trained on a data set composed of the energy deposition from electron, photons, charged and neutral hadrons in a fine grained digital calorimeter.

We investigated the relationships among environmental features of physical activity friendliness, socioeconomic indicators, and prevalence of obesity (BMI status), central adiposity (waist circumference, waist-height ratio), and hypertension.

Abstract Deep Learning techniques are being studied for different applications by the HEP community: in this talk, we discuss the case of detector simulation. The need for simulated events, expected in the future for LHC experiments and their High Luminosity upgrades, is increasing dramatically and requires new fast simulation solutions. Here we present updated results on the development of 3DGAN, one of the first examples using three-dimensional convolutional Generative Adversarial Networks to simulate high granularity electromagnetic calorimeters. In particular, we report on two main aspects: results on the simulation of a more general, realistic physics use case and on data parallel strategies to distribute the training process across multiple nodes on public cloud resources.

Cross-lingual topic models have been prevalent for cross-lingual text analysis by revealing aligned latent topics. However, most existing methods suffer from producing repetitive topics that hinder further analysis and performance decline caused by low-coverage dictionaries. In this paper, we propose the Cross-lingual Topic Modeling with Mutual Information (InfoCTM). Instead of the direct alignment in previous work, we propose a topic alignment with mutual information method. This works as a regularization to properly align topics and prevent degenerate topic representations of words, which mitigates the repetitive topic issue. To address the low-coverage dictionary issue, we further propose a cross-lingual vocabulary linking method that finds more linked cross-lingual words for topic alignment beyond the translations of a given dictionary. Extensive experiments on English, Chinese, and Japanese datasets demonstrate that our method outperforms state-of-the-art baselines, producing more coherent, diverse, and well-aligned topics and showing better transferability for cross-lingual classification tasks.

Understanding and comparing the behavior of retrieval models is a fundamental challenge that requires going beyond examining average effectiveness and per-query metrics, because these do not reveal key differences in how ranking models' behavior impacts individual results. DiffIR is a new open-source web tool to assist with qualitative ranking analysis by visually 'diffing' system rankings at the individual result level for queries where behavior significantly diverges. Using one of several configurable similarity measures, it identifies queries for which the rankings of models compared have important differences in individual rankings and provides a visual web interface to compare the rankings side-by-side. DiffIR additionally supports a model-specific visualization approach based on custom term importance weight files. These support studying the behavior of interpretable models, such as neural retrieval methods that produce document scores based on a similarity matrix or based on a single document passage. Observations from this tool can complement neural probing approaches like ABNIRML to generate quantitative tests. We provide an illustrative use case of DiffIR by studying the qualitative differences between recently developed neural ranking models on a standard TREC benchmark dataset.

This paper investigates the design of M-band linear phase wavelet filter banks (M>2), and explores their application to image coding. The generalized LOT description of M-band linear-phase paraunitary filter banks is used to parametrize the M-band linear-phase orthogonal wavelets. It is proven that an M-band linear-phase orthogonal wavelet of even length cannot have more than one vanishing moment. Since this limits the effectiveness of the resulting wavelet filters, we next suggest methods for the construction of linear-phase biorthogonal M-band wavelet lowpass filters, generalizing prior 2-band constructions. However, one cannot guarantee that an arbitrary lowpass filter pair can be completed to a full perfect-reconstruction filter bank. Finally, the new linear-phase orthogonal wavelet filter banks are compared with known wavelet filters with regard to their performance in a transform-based image coder.

(1) Determine the prevalence of overweight and high blood pressure (BP) among middle and high school students over a 2-year period and, (2) measure the cost and initial outcomes of screening.Cost and outcome description using a cross-sectional design sample. The target population was 12- to 19-year-old healthy students attending grades 7 through 12 at 3 proximal schools located in a large urban school district in Texas.Of 2,338 students screened, 925 (39.6%) had a body mass index (BMI)>or=85th percentile and 504 (21.6%) had BMIs>or=95th percentile for age and gender. There were 346 students (14.8%) with BMIs>or=85th percentile and systolic blood pressure (SBP)>or=95th percentile for age, gender, and height. The cost of the 2-year screening program was $66,442, and the cost per student was $28. The cost to identify a student with increased BMI or high SBP was $72 and $107, respectively.This study offered an objective framework to examine the cost and outcomes of screening children for overweight and increased BP. The study has implications for discussion and informed decision making about school-based screening for these conditions.

We discuss a model-independent strategy for boosting new physics searches with the help of an unsupervised anomaly detection algorithm. Prior to a search, each input event is preprocessed by the algorithm - a variational autoencoder (VAE). Based on the loss assigned to each event, input data can be split into a background control sample and a signal enriched sample. Following this strategy, one can enhance the sensitivity to new physics with no assumption on the underlying new physics signature. Our results show that a typical BSM search on the signal enriched group is more sensitive than an equivalent search on the original dataset.

We apply an Adversarially Learned Anomaly Detection (ALAD) algorithm to the problem of detecting new physics processes in proton-proton collisions at the Large Hadron Collider. Anomaly detection based on ALAD matches performances reached by Variational Autoencoders, with a substantial improvement in some cases. Training the ALAD algorithm on 4.4 fb-1 of 8 TeV CMS Open Data, we show how a data-driven anomaly detection and characterization would work in real life, re-discovering the top quark by identifying the main features of the t-tbar experimental signature at the LHC.

Automated methods have been widely used to identify and analyze mental health conditions (e.g., depression) from various sources of information, including social media. Yet, deployment of such models in real-world healthcare applications faces challenges including poor out-of-domain generalization and lack of trust in black box models. In this work, we propose approaches for depression detection that are constrained to different degrees by the presence of symptoms described in PHQ9, a questionnaire used by clinicians in the depression screening process. In dataset-transfer experiments on three social media datasets, we find that grounding the model in PHQ9's symptoms substantially improves its ability to generalize to out-of-distribution data compared to a standard BERT-based approach. Furthermore, this approach can still perform competitively on in-domain data. These results and our qualitative analyses suggest that grounding model predictions in clinically-relevant symptoms can improve generalizability while producing a model that is easier to inspect.

\(h^\pm \to \tau^\pm \nu\) signal events reconstructed by inclusive single-muon selection. Events are represented as an array of physics-motivated high-level features. Details are given in https://arxiv.org/abs/1811.10276

Recent empirical studies show that adversarial topic models (ATM) can successfully capture semantic patterns of the document by differentiating a document with another dissimilar sample. However, utilizing that discriminative-generative architecture has two important drawbacks: (1) the architecture does not relate similar documents, which has the same document-word distribution of salient words; (2) it restricts the ability to integrate external information, such as sentiments of the document, which has been shown to benefit the training of neural topic model. To address those issues, we revisit the adversarial topic architecture in the viewpoint of mathematical analysis, propose a novel approach to re-formulate discriminative goal as an optimization problem, and design a novel sampling method which facilitates the integration of external variables. The reformulation encourages the model to incorporate the relations among similar samples and enforces the constraint on the similarity among dissimilar ones; while the sampling method, which is based on the internal input and reconstructed output, helps inform the model of salient words contributing to the main topic. Experimental results show that our framework outperforms other state-of-the-art neural topic models in three common benchmark datasets that belong to various domains, vocabulary sizes, and document lengths in terms of topic coherence.

The general factorization of a linear-phase paraunitary filter bank (LPPUFB) is revisited and we introduce a class of lapped orthogonal transforms with extended overlap (GenLOT). In this formulation, the discrete cosine transform (DCT) is the order-1 GenLOT, the lapped orthogonal transform is the order-2 GenLOT, and so on, for any filter length which is an integer multiple of the block size. All GenLOTs are based on the DCT and have fast implementation algorithms. The degrees of freedom in the design of GenLOTs are described and design examples are presented along with some practical applications.< <ETX xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">&gt;</ETX>

Learned sparse retrieval (LSR) is a family of first-stage retrieval methods that are trained to generate sparse lexical representations of queries and documents for use with an inverted index. Many LSR methods have been recently introduced, with Splade models achieving state-of-the-art performance on MSMarco. Despite similarities in their model architectures, many LSR methods show substantial differences in effectiveness and efficiency. Differences in the experimental setups and configurations used make it difficult to compare the methods and derive insights. In this work, we analyze existing LSR methods and identify key components to establish an LSR framework that unifies all LSR methods under the same perspective. We then reproduce all prominent methods using a common codebase and re-train them in the same environment, which allows us to quantify how components of the framework affect effectiveness and efficiency. We find that (1) including document term weighting is most important for a method's effectiveness, (2) including query weighting has a small positive impact, and (3) document expansion and query expansion have a cancellation effect. As a result, we show how removing query expansion from a state-of-the-art model can reduce latency significantly while maintaining effectiveness on MSMarco and TripClick benchmarks. Our code is publicly available at https://github.com/thongnt99/learned-sparse-retrieval

Cross-lingual topic models have been prevalent for cross-lingual text analysis by revealing aligned latent topics. However, most existing methods suffer from producing repetitive topics that hinder further analysis and performance decline caused by low-coverage dictionaries. In this paper, we propose the Cross-lingual Topic Modeling with Mutual Information (InfoCTM). Instead of the direct alignment in previous work, we propose a topic alignment with mutual information method. This works as a regularization to properly align topics and prevent degenerate topic representations of words, which mitigates the repetitive topic issue. To address the low-coverage dictionary issue, we further propose a cross-lingual vocabulary linking method that finds more linked cross-lingual words for topic alignment beyond the translations of a given dictionary. Extensive experiments on English, Chinese, and Japanese datasets demonstrate that our method outperforms state-of-the-art baselines, producing more coherent, diverse, and well-aligned topics and showing better transferability for cross-lingual classification tasks.

Multimodal Review Helpfulness Prediction (MRHP) aims to rank product reviews based on predicted helpfulness scores and has been widely applied in e-commerce via presenting customers with useful reviews. Previous studies commonly employ fully-connected neural networks (FCNNs) as the final score predictor and pairwise loss as the training objective. However, FCNNs have been shown to perform inefficient splitting for review features, making the model difficult to clearly differentiate helpful from unhelpful reviews. Furthermore, pairwise objective, which works on review pairs, may not completely capture the MRHP goal to produce the ranking for the entire review list, and possibly induces low generalization during testing. To address these issues, we propose a listwise attention network that clearly captures the MRHP ranking context and a listwise optimization objective that enhances model generalization. We further propose gradient-boosted decision tree as the score predictor to efficaciously partition product reviews' representations. Extensive experiments demonstrate that our method achieves state-of-the-art results and polished generalization performance on two large-scale MRHP benchmark datasets.

Learned sparse retrieval (LSR) is a family of neural retrieval methods that transform queries and documents into sparse weight vectors aligned with a vocabulary. While LSR approaches like Splade work well for short passages, it is unclear how well they handle longer documents. We investigate existing aggregation approaches for adapting LSR to longer documents and find that proximal scoring is crucial for LSR to handle long documents. To leverage this property, we proposed two adaptations of the Sequential Dependence Model (SDM) to LSR: ExactSDM and SoftSDM. ExactSDM assumes only exact query term dependence, while SoftSDM uses potential functions that model the dependence of query terms and their expansion terms (i.e., terms identified using a transformer's masked language modeling head). Experiments on the MSMARCO Document and TREC Robust04 datasets demonstrate that both ExactSDM and SoftSDM outperform existing LSR aggregation approaches for different document length constraints. Surprisingly, SoftSDM does not provide any performance benefits over ExactSDM. This suggests that soft proximity matching is not necessary for modeling term dependence in LSR. Overall, this study provides insights into handling long documents with LSR, proposing adaptations that improve its performance.

We present Semantic Interpreter, a natural language-friendly AI system for productivity software such as Microsoft Office that leverages large language models (LLMs) to execute user intent across application features. While LLMs are excellent at understanding user intent expressed as natural language, they are not sufficient for fulfilling application-specific user intent that requires more than text-to-text transformations. We therefore introduce the Office Domain Specific Language (ODSL), a concise, high-level language specialized for performing actions in and interacting with entities in Office applications. Semantic Interpreter leverages an Analysis-Retrieval prompt construction method with LLMs for program synthesis, translating natural language user utterances to ODSL programs that can be transpiled to application APIs and then executed. We focus our discussion primarily on a research exploration for Microsoft PowerPoint.

Topic models have been prevalent for decades with various applications. However, existing topic models commonly suffer from the notorious topic collapsing: discovered topics semantically collapse towards each other, leading to highly repetitive topics, insufficient topic discovery, and damaged model interpretability. In this paper, we propose a new neural topic model, Embedding Clustering Regularization Topic Model (ECRTM). Besides the existing reconstruction error, we propose a novel Embedding Clustering Regularization (ECR), which forces each topic embedding to be the center of a separately aggregated word embedding cluster in the semantic space. This enables each produced topic to contain distinct word semantics, which alleviates topic collapsing. Regularized by ECR, our ECRTM generates diverse and coherent topics together with high-quality topic distributions of documents. Extensive experiments on benchmark datasets demonstrate that ECRTM effectively addresses the topic collapsing issue and consistently surpasses state-of-the-art baselines in terms of topic quality, topic distributions of documents, and downstream classification tasks.

Generative retrieval is a promising new neural retrieval paradigm that aims to optimize the retrieval pipeline by performing both indexing and retrieval with a single transformer model. However, this new paradigm faces challenges with updating the index and scaling to large collections. In this paper, we analyze two prominent variants of generative retrieval and show that they can be conceptually viewed as bi-encoders for dense retrieval. Specifically, we analytically demonstrate that the generative retrieval process can be decomposed into dot products between query and document vectors, similar to dense retrieval. This analysis leads us to propose a new variant of generative retrieval, called Tied-Atomic, which addresses the updating and scaling issues by incorporating techniques from dense retrieval. In experiments on two datasets, NQ320k and the full MSMARCO, we confirm that this approach does not reduce retrieval effectiveness while enabling the model to scale to large collections.

Abstract With the burgeoning amount of data of image-text pairs and diversity of Vision-and-Language (V&amp;L) tasks, scholars have introduced an abundance of deep learning models in this research domain. Furthermore, in recent years, transfer learning has also shown tremendous success in Computer Vision for tasks such as Image Classification, Object Detection, etc., and in Natural Language Processing for Question Answering, Machine Translation, etc. Inheriting the spirit of Transfer Learning, research works in V&amp;L have devised multiple pretraining techniques on large-scale datasets in order to enhance the performance of downstream tasks. The aim of this article is to provide a comprehensive revision of contemporary V&amp;L pretraining models. In particular, we categorize and delineate pretrain-ing approaches, along with the summary of state-of-the-art vision-and-language pretrained models. Moreover, a list of training datasets and downstream tasks is supplied to further polish the perspective into V&amp;L pretraining. Lastly, we decided to take a further step to discuss numerous directions for future research.

Multimodal Review Helpfulness Prediction (MRHP) aims to rank product reviews based on predicted helpfulness scores and has been widely applied in e-commerce via presenting customers with useful reviews.Previous studies commonly employ fully-connected neural networks (FCNNs) as the final score predictor and pairwise loss as the training objective.However, FCNNs have been shown to perform inefficient splitting for review features, making the model difficult to clearly differentiate helpful from unhelpful reviews.Furthermore, pairwise objective, which works on review pairs, may not completely capture the MRHP goal to produce the ranking for the entire review list, and possibly induces low generalization during testing.To address these issues, we propose a listwise attention network that clearly captures the MRHP ranking context and a listwise optimization objective that enhances model generalization.We further propose gradient-boosted decision tree as the score predictor to efficaciously partition product reviews' representations.Extensive experiments demonstrate that our method achieves state-of-the-art results and polished generalization performance on two large-scale MRHP benchmark datasets.

The Snowmass report on rare processes and precision measurements recommended Mu2e-II and a next generation muon facility at Fermilab (Advanced Muon Facility) as priorities for the frontier. The Workshop on a future muon program at FNAL was held in March 2023 to discuss design studies for Mu2e-II, organizing efforts for the next generation muon facility, and identify synergies with other efforts (e.g., muon collider). Topics included high-power targetry, status of R&D for Mu2e-II, development of compressor rings, FFA and concepts for muon experiments (conversion, decays, muonium and other opportunities) at AMF. This document summarizes the workshop discussions with a focus on future R&D tasks needed to realize these concepts.

Assessment of potential risk of sludge dispersion of Tho-Quang fishing port is essential to ensure compliance with the Technical Regulations on assessment of dredges and identification of dredged sludge areas in Circular No. 28/2019/TT-BTNMT dated December 31, 2019 of the Ministry of Natural Resources and Environment. However, there are still many discussion regarding the possibility of spreading waste after submerging in the proposed area and potentially affecting the natural ecosystem of the vicinity off the coast of Son-Tra peninsula. The results of this study will contribute to the scientific theoretical basis to be able to quantitatively evaluate the spread of sewage sludge extracted from the dredging of Tho-Quang boat lock. In this study, we will apply 3D numerical modeling to simulate dredged sludge waste propagation with different discharge scenarios in terms of submergence depth, submerged sludge discharge as well as other hydro-meteorological scenarios in order to evaluate the potential risks of sewage sludge spreading to the nature reserve in the Son-Tra peninsula, Danang city. The results show that the influence of tidal current plays a very important role in the spread of sludge. Under the effect of tidal currents, submerged sludge can spread up to 10 km from the discharge center with a concentration of 5 mg/l.

Counterfactual statements, which describe events that did not or cannot take place, are beneficial to numerous NLP applications. Hence, we consider the problem of counterfactual detection (CFD) and seek to enhance the CFD models. Previous models are reliant on clue phrases to predict counterfactuality, so they suffer from significant performance drop when clue phrase hints do not exist during testing. Moreover, these models’ prediction also biases towards non-counterfactual over the counterfactual class. To address these issues, we propose to integrate neural topic model into the CFD model to capture the global semantics of the input statement. We continue to causally intervene the hidden representations of the CFD model to balance the effect of the class labels. Extensive experiments show that our approach outperforms previous state-of-the-art CFD and bias-resolving methods in both the CFD and other bias-sensitive tasks.

A portable, light weight, low-cost, dual-axis solar turtle prototype with dynamic self-tracking solar panel was designed and fabricated using 3D printed, machined, and purchased components to charge a battery large enough to run multiple devices. The base and body of the design housing is a locking lid rolling cooler that insures if the unit is tipped over the components inside will not be damaged. The design also makes the unit slightly weather resistant. Utilizing a fixed main post made from lightweight strong material allows for the main panel to have a height of 5 feet off the ground or 1 foot off the base of the design for the solar cell to operate. PVC was chosen due to its hollow center and strong and slightly flexible body that is lightweight. All parts needed for the prototype design were purchased along with some parts being 3D printed. Some aluminum milled pieces were used in the final design due to the large amount of load needed to handle at certain points of the build. Implementing a practical "yes/no" function will verify proper angle and exposure of the panel towards the sun, using miniature solar panels that will be eclipsed by the larger solar cell whenever directly facing the sun itself. The dual-axis solar turtle built is marketable and more efficient than static panels due to auto-tracking. When tested, it collected 19.9% more solar energy than static panels.

Evidence for the light-by-light scattering process, γγ→γγ, in ultraperipheral PbPb collisions at a centre-of-mass energy per nucleon pair of 5.02TeV is reported. The analysis is conducted using a data sample corresponding to an integrated luminosity of 390μb^(-1) recorded by the CMS experiment at the LHC. Light-by-light scattering processes are selected in events with two photons exclusively produced, each with transverse energy E_T^γ > 2GeV, pseudorapidity |η^γ| 5GeV, diphoton transverse momentum p_T^(γγ) < 1GeV, and diphoton acoplanarity below 0.01. After all selection criteria are applied, 14 events are observed, compared to expectations of 9.0±0.9(theo) events for the signal and 4.0±1.2(stat) for the background processes. The excess observed in data relative to the background-only expectation corresponds to a significance of 3.7 standard deviations, and has properties consistent with those expected for the light-by-light scattering signal. The measured fiducial light-by-light scattering cross section, σ_(fid)(γγ→γγ) = 120±46(stat)±28(syst)±12(theo)nb, is consistent with the standard model prediction. The mγγ distribution is used to set new exclusion limits on the production of pseudoscalar axion-like particles, via the γγ→a→γγ process, in the mass range m_a = 5-90 GeV.

\(LQ \to b \tau\) signal events reconstructed by inclusive single-muon selection. Events are represented as an array of physics-motivated high-level features. Details are given in https://arxiv.org/abs/1811.10276

\(A \to 4\ell\) signal events reconstructed by inclusive single-muon selection. Events are represented as an array of physics-motivated high-level features. Details are given in https://arxiv.org/abs/1811.10276

\(h^0 \to \tau^+ \tau^-\) signal events reconstructed by inclusive single-muon selection. Events are represented as an array of physics-motivated high-level features. Details are given in https://arxiv.org/abs/1811.10276

Modern Review Helpfulness Prediction systems are dependent upon multiple modalities, typically texts and images. Unfortunately, those contemporary approaches pay scarce attention to polish representations of cross-modal relations and tend to suffer from inferior optimization. This might cause harm to model's predictions in numerous cases. To overcome the aforementioned issues, we propose Multimodal Contrastive Learning for Multimodal Review Helpfulness Prediction (MRHP) problem, concentrating on mutual information between input modalities to explicitly elaborate cross-modal relations. In addition, we introduce Adaptive Weighting scheme for our contrastive learning approach in order to increase flexibility in optimization. Lastly, we propose Multimodal Interaction module to address the unalignment nature of multimodal data, thereby assisting the model in producing more reasonable multimodal representations. Experimental results show that our method outperforms prior baselines and achieves state-of-the-art results on two publicly available benchmark datasets for MRHP problem.

At SC16, the SCC teams participated in a new application area: the Reproducibility Challenge. In this paper we report on our efforts to reproduce results presented in a paper titled “A Parallel Connectivity Algorithm for de Bruijn Graphs in Metagenomic Applications,” which shows that the parallel graph-based algorithm developed scales to over a thousand cores, and runs faster than traditional Breadth First Search algorithms. In general, using the smaller competition test data sets on over 128 processors, we were able to reproduce some, but not all, of the reported results: we were unable to run the D1 data set on 128 cores and 2GB/core memory; our results did show similar timing trends for the different algorithm variations; we were able to observe the trend of communication dominating the computation time; and the AP and AP_LB versions of our runs on smaller datasets only show a small time improvement in our graphs, which is similar but not exactly what was described within the paper. We believe that cluster architecture, required memory, network tuning, and number of processors available impacted our ability to exactly reproduce the results of the paper.

This paper presents the lattice structure of a class of dyadic symmetric wavelets with integer coefficients. These wavelets are found by factorization of the maximally-flat halfband filter. The proposed lattice structure has modular structure and is very robust with respect to coefficient quantization. Moreover, the number of internal bit is low and the overall structure is suitable for VLSI implementation.

Filter banks have been shown to be efficient in several emerging signal communication applications. A new class of time-frequency spread coders for transmultiplexer systems using multirate filter banks is presented. As compared with conventional filter banks designed with stopband attenuation and passband flatness criteria, the user coders with the new algorithms are designed with time and frequency spread criteria. In new the algorithms, the filters are achieved through a cascade of lattice structures and delay chains. Along with the time-frequency property and the reconstruction property, the intercode/intracode correlation property are included in the design algorithms to be optimized. The designed coders are used in the application of digital watermarking of images and demonstrate good performance in JPEG encoding at different qualities.

Alongshore oscillatory flows over an elongated topographic feature next to a vertical wall for a homogeneous, rotating fluid were investigated by means of numerical and laboratory experiments. The physical experiments were conducted in the Grenoble 13-m diameter rotating tank, in which an elongated obstacle of limited longitudinal extent was placed along the vertical sidewall. The background oscillating motion was obtained by periodically varying the platform angular velocity. Fluid motions were visualized and quantified by direct velocity measurements and particle tracking. The numerical model employed was a tridimensional model developed by Haidvogel et al. It consists of the traditional primitive equations, that is, the Navier-Stokes equations for a rotating fluid with the addition of the hydrostatic, Boussinesq, and incompressibility approximations. (The experiments described here employ the homogeneous version.) The numerical formulation uses finite differences in the horizontal and spectral representation in the vertical dimensions. Both the laboratory and numerical experiments show that in the range of dimensionless parameters considered, two distinct flow regimes, based on general properties of the rectified flow patterns observed, can be defined. It is further shown that the flow regime designation depends principally on the magnitude of the temporal Rossby number, Rot, defined as the ratio of the flow oscillation to the background rotation frequency. Good qualitative and quantitative agreement is found between the laboratory experiments and the numerical model for such observables as the spatial distribution of rectified flow patterns. Several other flow observables are defined and their relation with the system parameters delineated.

[1666-13-3] C12H10Se2 (MW 312.14) InChI = 1S/C12H10Se2/c1-3-7-11(8-4-1)13-14-12-9-5-2-6-10-12/h1-10H InChIKey = YWWZCHLUQSHMCL-UHFFFAOYSA-N (source of phenylseleno functionality, reactive with wide variety of organic nucleophiles,2-13 electrophiles,14-24 and radicals;25-27 source of phenylseleno radicals28-31) Physical Data: mp 63–64 °C. Solubility: sol MeOH, EtOH, ether, THF, toluene, and most common organic solvents; insol water. Form Supplied in: yellow powder; widely available. Preparative Methods: synthesized by reaction of Phenylmagnesium Bromide with Selenium, followed by treatment with Bromine.32 This method is superior to older methods33 in that it does not generate the toxic byproducts H2Se and PhSeH. Handling, Storage, and Precautions: is air stable, has a faint odor, and is not appreciably hygroscopic. Organoselenides are reputed to be highly toxic, although not as toxic as inorganic selenium compounds such as H2Se or SeO2.1b Use in a fume hood.

\(t \bar t\) background events reconstructed by inclusive single-muon selection. Events are represented as an array of physics-motivated high-level features. Details are given in https://arxiv.org/abs/1811.10276

QCD multijet background events reconstructed by inclusive single-muon selection. Events are represented as an array of physics-motivated high-level features. Details are given in https://arxiv.org/abs/1811.10276

\(Z \to \ell \ell\) background events reconstructed by inclusive single-muon selection. Events are represented as an array of physics-motivated high-level features. Details are given in https://arxiv.org/abs/1811.10276

\(W \to \ell \nu\) background events reconstructed by inclusive single-muon selection. Events are represented as an array of physics-motivated high-level features. Details are given in https://arxiv.org/abs/1811.10276

The authors describe a method of classifying natural textures based on the maximum likelihood parameter estimation technique. The novelty of the technique lies in the use of textural features that are derived from the subbands of a wavelet transformed image via the co-occurrence matrices. A maximum likelihood classifier is designed using a set of training texture samples. Ten different Brodotz textures have been classified using this procedure with an average classification accuracy of 99.7%. The main emphasis is to apply this technique to the classification of underwater acoustic signals. A time–frequency plot is obtained for each segment of the acoustic signal and then converted to an intensity pattern. The textural classification scheme is then applied to the intensity patterns of the acoustic signals. Eight different underwater acoustic signals have been classified by this procedure with an average accuracy of 99.99%.

This paper introduces a family of symmetric biorthogonal M-band wavelets with rational and integer coefficients and up to 2 degrees of regularity. Unlike previous approaches, we construct M-band wavelets by adding simple pre/post-filtering modules along the block boundaries of the traditional block DCT framework. Integer- and rational-coefficient solutions are then obtained from placing further restrictions on each involved component. In other words, we address the design and implementation of fast, efficient pre/post-filters that help improve the polynomial-representing and capturing capability of traditional block transforms such as the DCT. Fast symmetric integer-mapping M-band wavelets can be easily generated by iterating our decomposition scheme on the lowpass subband. Several design examples are presented to demonstrate the validity of the proposed theory.

ABSTRACT Application of openhole sand control technology is becoming mandatory in the field, particularly with the given uncertainty in geomechanics, challenges to wellbore integrity while drilling, and sand production during the life of the well. The completion equipment readiness and success of the installation can be challenging in the event of extending the horizontal section to accommodate geological heterogeneity and maximizing well productivity. This paper discusses operational excellence recorded in Well A, in the Thang Long Field, offshore Vietnam, from well design perspectives ensuring maximum reservoir contact to outcome of well completion. The well was targeted in the Oligocene reservoir, a thin oil rim with large gas cap overlay, and required drilling and completion for 1126 m horizontal length of 8 1/2-in. open hole. The completion design included multiple swellable packers for isolation of unwanted zones, 6 5/8-in. basepipe sand screens for the production zones, and a fluid loss control device to help prevent undesirable losses. Several torque and drag simulations were performed to help predict potential threats that could be encountered during completion string deployment or during space out of the inner wash pipe string. One apparent challenge of this completion design was to deploy the lower completion string to total depth (TD) per stringent reservoir requirements, resulting in an approximate 1126 m length of the string in the horizontal section. Another task was to facilitate manipulating 1130 m of wash pipe inside the completion string to locate the seal assemblies accurately at the corresponding seal bore extension positions for effective acidizing treatment. Although these were long sections of completion string and wash pipe, the quality of acidizing stimulation to effectively remove mud cake should not be compromised to ensure positive production rates. During operations, the completion string was run to target depth without any issue, and the wash pipe was spaced out and manipulated correctly. These operations subsequently led to a successful acidizing treatment and the proper closure of the flapper type fluid loss device. The completion design and operation were concluded successfully, significantly contributing to field production performance to date. The novelty of the completion design and installation is the ability to deploy an 1126-m lower completion in long, highly deviated and horizontal openhole section coupled with acid stimulation in reasonable time and as per plan.

This work focuses on the use of image processing methods to detect and classify underwater acoustic signals. The time-frequency spectra of underwater acoustic signals are usually converted to lofargrams for display purposes. These lofargrams exhibit texture-like characteristics. Moving targets exhibit ramps while ambient noise has a noisy pattern. Hence, these can be detected using textural pattern classification methods. More specifically, textural features such as contrast, entropy, inverse difference moment, etc., are computed from the co-occurrence matrices of the lofargrams. A maximum likelihood classifier is designed to classify the different patterns in the lofargrams. We have successfully classified eight different narrowband underwater acoustic signals with an average classification accuracy of 99.99%.

Aspects of human behavior in cyber security allow more natural security to the user.This research focuses the appearance of anticipating cyber threats and their abstraction hierarchy levels on the mental picture levels of human.The study concerns the modeling of the behaviors of mental states of an individual under cyber attacks.The mental state of agents being not observable, we propose a non-stationary hidden Markov chain approach to model the agent mental behaviors.A renewal process based on a nonparametric estimation is also considered to investigate the spending time in a given mental state.In these approaches, the effects of the complexity of the cyber attacks are taken into account in the models.

With the burgeoning amount of data of image-text pairs and diversity of Vision-and-Language (V\&L) tasks, scholars have introduced an abundance of deep learning models in this research domain. Furthermore, in recent years, transfer learning has also shown tremendous success in Computer Vision for tasks such as Image Classification, Object Detection, etc., and in Natural Language Processing for Question Answering, Machine Translation, etc. Inheriting the spirit of Transfer Learning, research works in V\&L have devised multiple pretraining techniques on large-scale datasets in order to enhance the performance of downstream tasks. The aim of this article is to provide a comprehensive revision of contemporary V\&L pretraining models. In particular, we categorize and delineate pretraining approaches, along with the summary of state-of-the-art vision-and-language pretrained models. Moreover, a list of training datasets and downstream tasks is supplied to further polish the perspective into V\&L pretraining. Lastly, we decided to take a further step to discuss numerous directions for future research.

Modern Review Helpfulness Prediction systems are dependent upon multiple modalities, typically texts and images.Unfortunately, those contemporary approaches pay scarce attention to polish representations of cross-modal relations and tend to suffer from inferior optimization.This might cause harm to model's predictions in numerous cases.To overcome the aforementioned issues, we propose Multimodal Contrastive Learning for Multimodal Review Helpfulness Prediction (MRHP) problem, concentrating on mutual information between input modalities to explicitly elaborate cross-modal relations.In addition, we introduce Adaptive Weighting scheme for our contrastive learning approach in order to increase flexibility in optimization.Lastly, we propose Multimodal Interaction module to address the unalignment nature of multimodal data, thereby assisting the model in producing more reasonable multimodal representations.Experimental results show that our method outperforms prior baselines and achieves state-of-the-art results on two publicly available benchmark datasets for MRHP problem.

Deep learning plays a significant role in jet tagging. Interaction network / message passing network are parameter efficient. The proposed network out-performs some other deep learning approaches. There is promising direction for future taggers and other problems.

We apply an Adversarially Learned Anomaly Detection (ALAD) algorithm to the problem of detecting new physics processes in proton-proton collisions at the Large Hadron Collider. Anomaly detection based on ALAD matches performances reached by Variational Autoencoders, with a substantial improvement in some cases. Training the ALAD algorithm on 4.4 fb-1 of 8 TeV CMS Open Data, we show how a data-driven anomaly detection and characterization would work in real life, re-discovering the top quark by identifying the main features of the t-tbar experimental signature at the LHC.

\(LQ \to b \tau\) signal events reconstructed by inclusive single-muon selection. Events are represented as an array of physics-motivated high-level features. Details are given in https://arxiv.org/abs/1811.10276

\(Z \to \ell \ell\) background events reconstructed by inclusive single-muon selection. Events are represented as an array of physics-motivated high-level features. Details are given in https://arxiv.org/abs/1811.10276

\(h^0 \to \tau^+ \tau^-\) signal events reconstructed by inclusive single-muon selection. Events are represented as an array of physics-motivated high-level features. Details are given in https://arxiv.org/abs/1811.10276

\(h^\pm \to \tau^\pm \nu\) signal events reconstructed by inclusive single-muon selection. Events are represented as an array of physics-motivated high-level features. Details are given in https://arxiv.org/abs/1811.10276

Topology Classification Dataset: raw images of simulated proton-proton LHC collisions at 13 TeV Full description at https://arxiv.org/abs/1807.00083

QCD multijet background events reconstructed by inclusive single-muon selection. Events are represented as an array of physics-motivated high-level features. Details are given in https://arxiv.org/abs/1811.10276

\(W \to \ell \nu\) background events reconstructed by inclusive single-muon selection. Events are represented as an array of physics-motivated high-level features. Details are given in https://arxiv.org/abs/1811.10276

Topology Classification Dataset: abstract images of simulated proton-proton LHC collisions at 13 TeV Full description at https://arxiv.org/abs/1807.00083

\(A \to 4\ell\) signal events reconstructed by inclusive single-muon selection. Events are represented as an array of physics-motivated high-level features. Details are given in https://arxiv.org/abs/1811.10276

Topology Classification Dataset: abstract images of simulated proton-proton LHC collisions at 13 TeV Full description at https://arxiv.org/abs/1807.00083

\(t \bar t\) background events reconstructed by inclusive single-muon selection. Events are represented as an array of physics-motivated high-level features. Details are given in https://arxiv.org/abs/1811.10276

The trochanteric fractures usually occur in elderly people who are often associated with osteoporosis; the impacting forces are commonly low energy. The displacement of fragments can be much or little, but it can cause pain a lot.

Current state-of-the-art cross-lingual summarization models employ multi-task learning paradigm, which works on a shared vocabulary module and relies on the self-attention mechanism to attend among tokens in two languages. However, correlation learned by self-attention is often loose and implicit, inefficient in capturing crucial cross-lingual representations between languages. The matter worsens when performing on languages with separate morphological or structural features, making the cross-lingual alignment more challenging, resulting in the performance drop. To overcome this problem, we propose a novel Knowledge-Distillation-based framework for Cross-Lingual Summarization, seeking to explicitly construct cross-lingual correlation by distilling the knowledge of the monolingual summarization teacher into the cross-lingual summarization student. Since the representations of the teacher and the student lie on two different vector spaces, we further propose a Knowledge Distillation loss using Sinkhorn Divergence, an Optimal-Transport distance, to estimate the discrepancy between those teacher and student representations. Due to the intuitively geometric nature of Sinkhorn Divergence, the student model can productively learn to align its produced cross-lingual hidden states with monolingual hidden states, hence leading to a strong correlation between distant languages. Experiments on cross-lingual summarization datasets in pairs of distant languages demonstrate that our method outperforms state-of-the-art models under both high and low-resourced settings.

Current state-of-the-art cross-lingual summarization models employ multi-task learning paradigm, which works on a shared vocabulary module and relies on the self-attention mechanism to attend among tokens in two languages. However, correlation learned by self-attention is often loose and implicit, inefficient in capturing crucial cross-lingual representations between languages. The matter worsens when performing on languages with separate morphological or structural features, making the cross-lingual alignment more challenging, resulting in the performance drop. To overcome this problem, we propose a novel Knowledge-Distillation-based framework for Cross-Lingual Summarization, seeking to explicitly construct cross-lingual correlation by distilling the knowledge of the monolingual summarization teacher into the cross-lingual summarization student. Since the representations of the teacher and the student lie on two different vector spaces, we further propose a Knowledge Distillation loss using Sinkhorn Divergence, an Optimal-Transport distance, to estimate the discrepancy between those teacher and student representations. Due to the intuitively geometric nature of Sinkhorn Divergence, the student model can productively learn to align its produced cross-lingual hidden states with monolingual hidden states, hence leading to a strong correlation between distant languages. Experiments on cross-lingual summarization datasets in pairs of distant languages demonstrate that our method outperforms state-of-the-art models under both high and low-resourced settings.

Empirical data-driven models are increasingly being used to simulate shoreline evolution over time scales ranging from days to decades. For the empirical shoreline evolution modeling, the combined longshore and cross-shore modeling demonstrate the ability to improve the prediction skills for shoreline changes. This study focuses on the feasibility of the combined model with a case study of the embayed beach shoreline of Nha Trang, Vietnam. Nha Trang’s climate is dominated by the tropical monsoon climate, so the beach changes here are strongly affected by this typical climate. The combined model applied in this study is a coupling of the cross-shore model and a longshore model. The combined model can provide the seasonal shoreline position fluctuations induced by the longshore contribution. Besides, it is realized that the very mild waves in summer induce strong accretion which the model cannot simulate. Therefore, a small modification of the equilibrium Dean number (Ω0) in the ShoreFor cross-shore model is suggested to improve the accretion simulation.