arXiv subjects

Yang Han

Publications and source records attributed to Yang Han.

At least 55 records · Page 3Linked to original sources

arXiv2022

Synergies and Prospects for Early Resolution of the Neutrino Mass Ordering

The measurement of neutrino Mass Ordering (MO) is a fundamental element for the understanding of leptonic flavour sector of the Standard Model of Particle Physics. Its determination relies on the precise measurement of $Δm^2_{31}$ and $Δm^2_{32}$ using either neutrino vacuum oscillations, such as the ones studied by medium baseline reactor experiments, or matter effect modified oscillations such as those manifesting in long-baseline neutrino beams (LB$ν$B) or atmospheric neutrino experiments. Despite existing MO indication today, a fully resolved MO measurement ($\geq$5$σ$) is most likely to await for the next generation of neutrino experiments: JUNO, whose stand-alone sensitivity is $\sim$3$σ$, or LB$ν$B experiments (DUNE and Hyper-Kamiokande). Upcoming atmospheric neutrino experiments are also expected to provide precious information. In this work, we study the possible context for the earliest full MO resolution. A firm resolution is possible even before 2028, exploiting mainly vacuum oscillation, upon the combination of JUNO and the current generation of LB$ν$B experiments (NOvA and T2K). This opportunity is possible thanks to a powerful synergy boosting the overall sensitivity where the sub-percent precision of $Δm^2_{32}$ by LB$ν$B experiments is found to be the leading order term for the MO earliest discovery. We also found that the comparison between matter and vacuum driven oscillation results enables unique discovery potential for physics beyond the Standard Model.

hep-ph↗

arXiv2022

Audio-Visual Wake Word Spotting System For MISP Challenge 2021

This paper presents the details of our system designed for the Task 1 of Multimodal Information Based Speech Processing (MISP) Challenge 2021. The purpose of Task 1 is to leverage both audio and video information to improve the environmental robustness of far-field wake word spotting. In the proposed system, firstly, we take advantage of speech enhancement algorithms such as beamforming and weighted prediction error (WPE) to address the multi-microphone conversational audio. Secondly, several data augmentation techniques are applied to simulate a more realistic far-field scenario. For the video information, the provided region of interest (ROI) is used to obtain visual representation. Then the multi-layer CNN is proposed to learn audio and visual representations, and these representations are fed into our two-branch attention-based network which can be employed for fusion, such as transformer and conformed. The focal loss is used to fine-tune the model and improve the performance significantly. Finally, multiple trained models are integrated by casting vote to achieve our final 0.091 score.

cs.SD↗

arXiv2022

BV spaces and the perimeters related to Schrodinger operators with inverse-square potentials and applications to the rank-one theorem

For $a \ge - {( \frac{d}{2}- 1)^2} $ and $2σ= {d - 2}-( {{{(d - 2)}^2} + 4a})^{1/2}$, let $$\begin{cases}\mathcal{H}_{a}= - Δ+ \frac{a} {{{{ | x |}^2}}},\\ \mathcal{\widetilde{H}}_σ= 2\big( { - Δ+ \frac{σ^2} {{{{ | x |}^2}}}}\big)\end{cases}$$ be two Schrödinger operators with inverse-square potentials. In this paper, on the domain $Ω\subset {\mathbb {R}^d}\backslash \{ 0\}, d\geq 2,$ %apart from the origin, the ${\mathcal{H} _a}$-BV space $\mathcal{B} {\mathcal{V} _{{\mathcal{H} _a}}}(Ω)$ and the ${\mathcal{\widetilde{H}}_σ}$-BV space $\mathcal{B} {\mathcal{V} _{{\mathcal{\widetilde H} _σ}}}(Ω)$ related to $\mathcal{H}_{a}$ and $\mathcal{\widetilde{H}}_σ$ are introduced, respectively. We investigate a series of basic properties of $\mathcal{B} {\mathcal{V} _{{\mathcal{H} _a}}}(Ω)$ and $\mathcal{B} {\mathcal{V} _{{\mathcal{\widetilde H} _σ}}}(Ω)$. Furthermore, we prove that ${\mathcal{\widetilde{H}}_σ}$-restricted BV functions can be characterized equivalently via their subgraphs. As applications, we derive the rank-one theorem for ${\mathcal{\widetilde{H}}_σ}$-restricted BV functions.

math.FA↗

arXiv2022

Static and Dynamic Models for Multivariate Distribution Forecasts: Proper Scoring Rule Tests of Factor-Quantile vs. Multivariate GARCH Models

A plethora of static and dynamic models exist to forecast Value-at-Risk and other quantile-related metrics used in financial risk management. Industry practice tends to favour simpler, static models such as historical simulation or its variants whereas most academic research centres on dynamic models in the GARCH family. While numerous studies examine the accuracy of multivariate models for forecasting risk metrics, there is little research on accurately predicting the entire multivariate distribution. Yet this is an essential element of asset pricing or portfolio optimization problems having non-analytic solutions. We approach this highly complex problem using a variety of proper multivariate scoring rules to evaluate over 100,000 forecasts of eight-dimensional multivariate distributions: of exchange rates, interest rates and commodity futures. This way we test the performance of static models, viz. empirical distribution functions and a new factor-quantile model, with commonly used dynamic models in the asymmetric multivariate GARCH class.

stat.ME↗

arXiv2022

AKB-48: A Real-World Articulated Object Knowledge Base

Human life is populated with articulated objects. A comprehensive understanding of articulated objects, namely appearance, structure, physics property, and semantics, will benefit many research communities. As current articulated object understanding solutions are usually based on synthetic object dataset with CAD models without physics properties, which prevent satisfied generalization from simulation to real-world applications in visual and robotics tasks. To bridge the gap, we present AKB-48: a large-scale Articulated object Knowledge Base which consists of 2,037 real-world 3D articulated object models of 48 categories. Each object is described by a knowledge graph ArtiKG. To build the AKB-48, we present a fast articulation knowledge modeling (FArM) pipeline, which can fulfill the ArtiKG for an articulated object within 10-15 minutes, and largely reduce the cost for object modeling in the real world. Using our dataset, we propose AKBNet, a novel integral pipeline for Category-level Visual Articulation Manipulation (C-VAM) task, in which we benchmark three sub-tasks, namely pose estimation, object reconstruction and manipulation. Dataset, codes, and models will be publicly available at https://liuliu66.github.io/articulationobjects/.

cs.CV↗

arXiv2021

Shuffle Transformer with Feature Alignment for Video Face Parsing

This is a short technical report introducing the solution of the Team TCParser for Short-video Face Parsing Track of The 3rd Person in Context (PIC) Workshop and Challenge at CVPR 2021. In this paper, we introduce a strong backbone which is cross-window based Shuffle Transformer for presenting accurate face parsing representation. To further obtain the finer segmentation results, especially on the edges, we introduce a Feature Alignment Aggregation (FAA) module. It can effectively relieve the feature misalignment issue caused by multi-resolution feature aggregation. Benefiting from the stronger backbone and better feature aggregation, the proposed method achieves 86.9519% score in the Short-video Face Parsing track of the 3rd Person in Context (PIC) Workshop and Challenge, ranked the first place.

cs.CV↗

arXiv2021

Symmetry criteria for Hochschild extensions

We give two sufficient and necessary conditions for a Hochschild extension of a finite dimensional algebra by its dual bimodule and a Hochschild 2-cocycle to be a symmetric algebra.

math.RA↗

arXiv2021

Deep-AIR: A Hybrid CNN-LSTM Framework for Air Quality Modeling in Metropolitan Cities

Air pollution has long been a serious environmental health challenge, especially in metropolitan cities, where air pollutant concentrations are exacerbated by the street canyon effect and high building density. Whilst accurately monitoring and forecasting air pollution are highly crucial, existing data-driven models fail to fully address the complex interaction between air pollution and urban dynamics. Our Deep-AIR, a novel hybrid deep learning framework that combines a convolutional neural network with a long short-term memory network, aims to address this gap to provide fine-grained city-wide air pollution estimation and station-wide forecast. Our proposed framework creates 1x1 convolution layers to strengthen the learning of cross-feature spatial interaction between air pollution and important urban dynamic features, particularly road density, building density/height, and street canyon effect. Using Hong Kong and Beijing as case studies, Deep-AIR achieves a higher accuracy than our baseline models. Our model attains an accuracy of 67.6%, 77.2%, and 66.1% in fine-grained hourly estimation, 1-hr, and 24-hr air pollution forecast for Hong Kong, and an accuracy of 65.0%, 75.3%, and 63.5% for Beijing. Our saliency analysis has revealed that for Hong Kong, street canyon and road density are the best estimators for NO2, while meteorology is the best estimator for PM2.5.

cs.LG↗

arXiv2021

Mass production and characterization of 3-inch PMTs for the JUNO experiment

26,000 3-inch photomultiplier tubes (PMTs) have been produced for Jiangmen Underground Neutrino Observatory (JUNO) by the Hainan Zhanchuang Photonics Technology Co., Ltd (HZC) company in China and passed all acceptance tests with only 15 tubes rejected. The mass production began in 2018 and elapsed for about 2 years at a rate of $\sim$1,000~PMTs per month. The characterization of the PMTs was performed in the factory concurrently with production as a joint effort between HZC and JUNO. Fifteen performance parameters were tracked at different sampling rates, and novel working strategies were implemented to improve quality assurance. This constitutes the largest sample of 3-inch PMTs ever produced and studied in detail to date.

physics.ins-det↗

arXiv2021

Evaluating the Discrimination Ability of Proper Multivariate Scoring Rules

Proper scoring rules are commonly applied to quantify the accuracy of distribution forecasts. Given an observation they assign a scalar score to each distribution forecast, with the the lowest expected score attributed to the true distribution. The energy and variogram scores are two rules that have recently gained some popularity in multivariate settings because their computation does not require a forecast to have parametric density function and so they are broadly applicable. Here we conduct a simulation study to compare the discrimination ability between the energy score and three variogram scores. Compared with other studies, our simulation design is more realistic because it is supported by a historical data set containing commodity prices, currencies and interest rates, and our data generating processes include a diverse selection of models with different marginal distributions, dependence structure, and calibration windows. This facilitates a comprehensive comparison of the performance of proper scoring rules in different settings. To compare the scores we use three metrics: the mean relative score, error rate and a generalised discrimination heuristic. Overall, we find that the variogram score with parameter p=0.5 outperforms the energy score and the other two variogram scores.

stat.ME↗

arXiv2020

Hirzebruch-Riemann-Roch and Lefschetz type formulas for finite dimensional algebras

The Hirzebuch-Riemann-Roch (HRR) and Lefschetz type formulas for finite dimensional elementary algebras of finite global dimension are explicitly given. They have cohomological, homological, Hochschild cohomological and Hochschild homological four versions, and module, bimodule, module complex and bimodule complex four levels. For this, the dimension matrix of a bimodule (complex) and the trace matrix of a bimodule (complex) endomorphism are introduced. It is shown that Shklyarov pairing, Chern character and Hattori-Stallings trace can be concretely expressed by Cartan matrix, dimension vector and trace vector in this situation. Furthermore, the HRR and Lefschetz type formulas for finite dimensional elementary algebras of finite global dimension and dg algebras are compared.

math.RT↗

arXiv2020

A Further Study of Unsupervised Pre-training for Transformer Based Speech Recognition

Building a good speech recognition system usually requires large amounts of transcribed data, which is expensive to collect. To tackle this problem, many unsupervised pre-training methods have been proposed. Among these methods, Masked Predictive Coding achieved significant improvements on various speech recognition datasets with BERT-like Masked Reconstruction loss and Transformer backbone. However, many aspects of MPC have not been fully investigated. In this paper, we conduct a further study on MPC and focus on three important aspects: the effect of pre-training data speaking style, its extension on streaming model, and how to better transfer learned knowledge from pre-training stage to downstream tasks. Experiments reveled that pre-training data with matching speaking style is more useful on downstream recognition tasks. A unified training objective with APC and MPC provided 8.46% relative error reduction on streaming model trained on HKUST. Also, the combination of target data adaption and layer-wise discriminative training helped the knowledge transfer of MPC, which achieved 3.99% relative error reduction on AISHELL over a strong baseline.

eess.AS↗

arXiv2020

Transferability of neural network potentials for varying stoichiometry: phonons and thermal conductivity of Mn$_x$Ge$_y$ compounds

Germanium manganese compounds exhibit a variety of stable and metastable phases with different stoichiometry. These materials entail interesting electronic, magnetic and thermal properties both in their bulk form and as heterostructures. Here we develop and validate a transferable machine learning potential, based on the high-dimensional neural network formalism, to enable the study of Mn$_x$Ge$_y$ materials over a wide range of compositions. We show that a neural network potential fitted on a minimal training set reproduces successfully the structural and vibrational properties and the thermal conductivity of systems with different local chemical environments, and it can be used to predict phononic effects in nanoscale heterostructures.

cond-mat.mtrl-sci↗

arXiv2020

Deep-AIR: A Hybrid CNN-LSTM Framework forFine-Grained Air Pollution Forecast

Poor air quality has become an increasingly critical challenge for many metropolitan cities, which carries many catastrophicphysical and mental consequences on human health and quality of life. However, accurately monitoring and forecasting air qualityremains a highly challenging endeavour. Limited by geographically sparse data, traditional statistical models and newly emergingdata-driven methods of air quality forecasting mainly focused on the temporal correlation between the historical temporal datasets of airpollutants. However, in reality, both distribution and dispersion of air pollutants are highly location-dependant. In this paper, we proposea novel hybrid deep learning model that combines Convolutional Neural Networks (CNN) and Long Short Term Memory (LSTM)together to forecast air quality at high-resolution. Our model can utilize the spatial correlation characteristic of our air pollutant datasetsto achieve higher forecasting accuracy than existing deep learning models of air pollution forecast.

eess.SP↗

arXiv2019

Exact Hochschild extensions and deformed Calabi-Yau completions

We introduce the Hochschild extensions of dg algebras, which are $A_\infty$-algebras. We show that all exact Hochschild extensions are symmetric Hochschild extensions, more precisely, every exact Hochschild extension of a finite dimensional complete typical dg algebra is a symmetric $A_\infty$-algebra. Moreover, we prove that the Koszul dual of trivial extension is Calabi-Yau completion and the Koszul dual of exact Hochschild extension is deformed Calabi-Yau completion, more precisely, the Koszul dual of the trivial extension of a finite dimensional complete dg algebra is the Calabi-Yau completion of its Koszul dual, and the Koszul dual of an exact Hochschild extension of a finite dimensional complete typical dg algebra is the deformed Calabi-Yau completion of its Koszul dual.

math.RA↗

arXiv2018

Hochschild (co)homologies of dg $K$-rings and their Koszul duals

We formulate the (co)bar construction theory of dg $K$-(co)rings and the calculus theory of the Hochschild homology and cohomology of dg $K$-rings. As applications, we compare the Hochschild (co)homologies of a complete typical dg $K$-ring and its Koszul dual. Moreover, we show that the Koszul dual of a finite dimensional complete typical $d$-symmetric dg $K$-ring is a $d$-Calabi-Yau dg algebra whose Hochschild cohomology is a Batalin-Vilkovisky algebra. Furthermore, we prove that the Hochschild cohomologies of a finite dimensional complete typical $d$-symmetric dg $K$-ring and its Koszul dual are isomorphic as Batalin-Vilkovisky algebras. In conclusion, we found a connection between the Batalin-Vilkovisky algebra structures on the Hochschild cohomologies of $d$-Calabi-Yau dg algebras and $d$-symmetric dg $K$-rings.

math.KT↗

arXiv2017

Brauer-Thrall type theorems for derived module categories

The numerical invariants (global) cohomological length, (global) cohomological width, and (global) cohomological range of complexes (algebras) are introduced. Cohomological range leads to the concepts of derived bounded algebras and strongly derived unbounded algebras naturally. The first and second Brauer-Thrall type theorems for the bounded derived category of a finite-dimensional algebra over an algebraically closed field are obtained. The first Brauer-Thrall type theorem says that derived bounded algebras are just derived finite algebras. The second Brauer-Thrall type theorem says that an algebra is either derived discrete or strongly derived unbounded, but not both. Moreover, piecewise hereditary algebras and derived discrete algebras are characterized as the algebras of finite global cohomological width and finite global cohomological length respectively.

math.RT↗

arXiv2016

The effect of temperature dependent tissue parameters on acoustic radiation force induced displacements

Multiple ultrasound elastography techniques rely on acoustic radiation force (ARF) in monitoring high-intensity focused ultrasound (HIFU) therapy. However, ARF is dependent on tissue attenuation and sound speed, both of which are also known to change with temperature making the therapy monitoring more challenging. Furthermore, the viscoelastic properties of tissue are also temperature dependent, which affects the displacements induced by ARF. The aim of this study is to quantify the temperature dependent changes in the acoustic and viscoelastic properties of liver and investigate their effect on ARF induced displacements by using both experimental methods and simulations. Furthermore, the temperature dependent viscoelastic properties of liver are experimentally measured over a frequency range of 0.1-200 Hz at temperatures reaching 80 C, and both conventional and fractional Zener models are used to fit the data. The fractional Zener model was found to fit better with the experimental viscoelasticity data with respect to the conventional model with up to two magnitudes lower sum of squared errors (SSE). The characteristics of experimental displacement data were also seen in the simulations due to the changes in attenuation coefficient and lesion development. At low temperatures before thermal ablation, attenuation was found to affect the displacement amplitude. At higher temperature, the decrease in displacement amplitude occurs approximately at 60-70 C due to the combined effect of viscoelasticity changes and lesion growth overpowering the effect of attenuation. The results suggest that it is necessary to monitor displacement continuously during HIFU therapy in order to ascertain when ablation occurs.

physics.med-ph↗