arXiv subjects

Jiajun Zhang

Publications and source records attributed to Jiajun Zhang.

At least 217 records · Page 12Linked to original sources

arXiv2020

The parameter-free Finger-Of-God model and its application to 21cm intensity mapping

Using the galaxy catalog built from ELUCID N-body simulation and the semi-analytical galaxy formation model, we have built a mock HI intensity mapping map. We have implemented the Finger-of-God (FoG) effect in the map by considering the galaxy HI gas velocity dispersion. By comparing the HI power spectrum in the redshift space with the measurement from IllustrisTNG simulation, we have found that such FoG effect can explain the discrepancy between current mock map built from N-body simulation and Illustris TNG simulation. Then we built a parameter-free FoG model and a shot-noise model to calculate the HI power spectrum. We found that our model can accurately fit both the monopole and quadrupole moments of the HI matter power spectrum. Our method of building the mock HI intensity map and the parameter-free FoG model will be widely useful for the up-coming 21cm intensity mapping experiments, such as CHIME, Tianlai, BINGO, FAST and SKA. It is also crucial for us to study the non-linear effects in 21cm intensity mapping.

astro-ph.CO↗

arXiv2020

Weak equivalence principle, swampland and $H_0$ tension with fast single radio bursts FRB 180924 and FRB 190523

Two new fast single radio bursts FRB 180924 and FRB 190523 well localized to massive galaxies have opened a new window to probe and characterize how cosmic baryons are allocated between galaxies, their surroundings and intergalactic medium. We are motivated by testing Einstein's weak equivalence principle with these two cosmic transients which have accurate redshifts. Using photons with different energies emitted by FRB 180924, we obtain, so far, the most stringent bound $Δγ<2.16\times10^{-10}$ for non-repeating FRBs with accurate redshifts when only considering the gravitational potential of the Milk Way. If using the gravitational potential of the Laniakea supercluster instead of the Milk Way one, we also obtain the strictest bound $Δγ<1.06\times10^{-14}$ to date. In light of rapid progress of FRB cosmology, towards the next two decades, we give an universal limitation $Δγ<8.24\times10^{-22}$ from photons with different energies emitted by single FRBs with accurate redshifts. Moreover, we analyze detailedly the effects of various astrophysical parameters on the precision of weak equivalence principle. We also estimate the abilities of single FRBs with known redshifts to test the validity of swampland criterion, and to distinguish which value of $H_0$ is preferred.

astro-ph.CO↗

arXiv2020

Neural Machine Translation: Challenges, Progress and Future

Machine translation (MT) is a technique that leverages computers to translate human languages automatically. Nowadays, neural machine translation (NMT) which models direct mapping between source and target languages with deep neural networks has achieved a big breakthrough in translation performance and become the de facto paradigm of MT. This article makes a review of NMT framework, discusses the challenges in NMT, introduces some exciting recent progresses and finally looks forward to some potential future research trends. In addition, we maintain the state-of-the-art methods for various NMT tasks at the website https://github.com/ZNLP/SOTA-MT.

cs.CL↗

arXiv2020

A new moving group in the Local Arm

We present a new moving group clustered in kinematics, spatial position and elemental abundances. Its spatial position is around the center of the Local Arm of the Milky Way. A convergent point method was taken to select candidate member stars.\textbf{ Among 206 candidate member stars, 74 are pre-main-sequence stars and some of them have stellar disks.} We presume those pre-main sequence stars belong to Orion nebula. We suggest this moving group is caused by density wave of the Local Arm passing by.

astro-ph.GA↗

arXiv2019

Synchronous Speech Recognition and Speech-to-Text Translation with Interactive Decoding

Speech-to-text translation (ST), which translates source language speech into target language text, has attracted intensive attention in recent years. Compared to the traditional pipeline system, the end-to-end ST model has potential benefits of lower latency, smaller model size, and less error propagation. However, it is notoriously difficult to implement such a model without transcriptions as intermediate. Existing works generally apply multi-task learning to improve translation quality by jointly training end-to-end ST along with automatic speech recognition (ASR). However, different tasks in this method cannot utilize information from each other, which limits the improvement. Other works propose a two-stage model where the second model can use the hidden state from the first one, but its cascade manner greatly affects the efficiency of training and inference process. In this paper, we propose a novel interactive attention mechanism which enables ASR and ST to perform synchronously and interactively in a single model. Specifically, the generation of transcriptions and translations not only relies on its previous outputs but also the outputs predicted in the other task. Experiments on TED speech translation corpora have shown that our proposed model can outperform strong baselines on the quality of speech translation and achieve better speech recognition performances as well.

cs.CL↗

arXiv2019

Modelling cosmic ray electron physics in cosmological smoothed particle hydrodynamics simulation

Cosmic ray electron (CRE) acceleration and cooling are important physical processes in astrophysics. We develop an approximative framework to treat CRE physics in the parallel smoothed particle hydrodynamics code Gadget-3. In our methodology, the CRE spectrum of each fluid element is approximated by a single power-law distribution with spatially varying amplitude, upper cut-off, lower cut-off, and spectral index. We consider diffusive shock acceleration to be the source of injection, and oppositely the sinking processes is attributed to synchrotron radiation, inverse Compton scatters, and Coulomb scatters. The adiabatic gains and losses are also included. We show that our formalism produces the energy and pressure with an accuracy of $ > 90\%$ for a free cooling CRE spectrum. Both slope and intensity of the radio emission computed from the CRE population given by our method in cosmological hydro-simulation coincide well with observations, and our results also show that relaxed clusters have lower fluxes. Finally, we investigate several impacts of the CRE processes on the cosmological hydro-simulation, we find that: (1) the pressure of the CRE spectrum is very small and can be ignored in hydro-simulation, (2) the impacts of the CRE processes on the gas phase-space state of hydro-simulation is up to $3\%$, (3) the CRE processes induce a $5\%$ influence on the mass function in the mass range $10^{12} -10^{13} h^{-1} M_{\odot}$, (4) The gas temperature of massive galaxy cluster is influenced by the CRE processes up to $\sim 10\%$.

astro-ph.HE↗

arXiv2019

Chinese Spelling Error Detection Using a Fusion Lattice LSTM

Spelling error detection serves as a crucial preprocessing in many natural language processing applications. Due to the characteristics of Chinese Language, Chinese spelling error detection is more challenging than error detection in English. Existing methods are mainly under a pipeline framework, which artificially divides error detection process into two steps. Thus, these methods bring error propagation and cannot always work well due to the complexity of the language environment. Besides existing methods only adopt character or word information, and ignore the positive effect of fusing character, word, pinyin1 information together. We propose an LF-LSTM-CRF model, which is an extension of the LSTMCRF with word lattices and character-pinyin-fusion inputs. Our model takes advantage of the end-to-end framework to detect errors as a whole process, and dynamically integrates character, word and pinyin information. Experiments on the SIGHAN data show that our LF-LSTM-CRF outperforms existing methods with similar external resources consistently, and confirm the feasibility of adopting the end-to-end framework and the availability of integrating of character, word and pinyin information.

cs.CL↗

arXiv2019

Elemental abundances distributions in (R, $V_ϕ$) plane with LAMOST DR5 and Gaia DR2

Since Gaia DR2 was released, many velocity structures in the disk have been revealed such as large scale ridge-like patterns in the phase space. Both kinematic information and stellar elemental abundances are needed to reveal their evolution history. We have used labels from the APOGEE survey to predict elemental abundances for a huge amount of low resolution spectra from the LAMOST survey. Deep learning with artificial neural networks can automatically draw on physically sensible features in the spectrum for their predictions. Abundances of 12 individual elements: [C/Fe], [N/Fe], [O/Fe], [Mg/Fe], [Al/Fe], [Si/Fe], [S/Fe], [Cl/Fe], [Ca/Fe], [Ti/Fe], [Mn/Fe] and [Ni/Fe] along with basic stellar labels $T_{\textrm{eff}}$, log $g$, metallicity ([M/H] and [Fe/H]) and [$α$/M] for 1 063 386 stars have been estimated. Then those stars were cross matched with Gaia DR2 data to obtain kinematic parameters. We presented distributions of chemical abundances in the $V_ϕ$ versus \textit{R} coordinate. Our results extend the chemical characterization of the ridges in the (R, $V_ϕ$) plane to about $R = $13 kpc toward the anticenter direction. In addition, radial elemental abundance gradients for disk stars with abs(\textit{z})$ < 0.5$ kpc are investigated and we fitted a line for median abundance values of bins of stars with galactocentric distance between $R > 7.84$ kpc and $R < 15.84$ kpc. The radial metallicity gradients for disk stars are respectively -0.0475 $\pm $0.0015 for $R < 13.09$ kpc and -0.0173 $\pm $0.0028 for $R < 13.09$ kpc. Gradients for other elemental abundances are also obtained, for example, [$α$/M] gradients for disk stars is 0.0030 $\pm $0.0002.

astro-ph.GA↗

arXiv2019

Stellar chromospheric activity and age relation from open clusters in the LAMOST Survey

We identify member stars of more than 90 open clusters in the LAMOST survey. With the method of Fang et al.(2018), the chromospheric activity (CA) indices logR'CaK for 1091 member stars in 82 open clusters and logR'Hα for 1118 member stars in 83 open clusters are calculated. The relations between the average logR'CaK, logR'Hα in each open cluster and its age are investigated in different Teff and [Fe/H] ranges. We find that CA starts to decrease slowly from logt = 6.70 to logt = 8.50, and then decreases rapidly until logt = 9.53. The trend becomes clearer for cooler stars. The quadratic functions between logR' and logt with 4000K < Teff < 5500K are constructed, which can be used to roughly estimate ages of field stars with accuracy about 40% for logR'CaK and 60% for logR'Hα.

astro-ph.SR↗

arXiv2019

NCLS: Neural Cross-Lingual Summarization

Cross-lingual summarization (CLS) is the task to produce a summary in one particular language for a source document in a different language. Existing methods simply divide this task into two steps: summarization and translation, leading to the problem of error propagation. To handle that, we present an end-to-end CLS framework, which we refer to as Neural Cross-Lingual Summarization (NCLS), for the first time. Moreover, we propose to further improve NCLS by incorporating two related tasks, monolingual summarization and machine translation, into the training process of CLS under multi-task learning. Due to the lack of supervised CLS data, we propose a round-trip translation strategy to acquire two high-quality large-scale CLS datasets based on existing monolingual summarization datasets. Experimental results have shown that our NCLS achieves remarkable improvement over traditional pipeline methods on both English-to-Chinese and Chinese-to-English CLS human-corrected test sets. In addition, NCLS with multi-task learning can further significantly improve the quality of generated summaries. We make our dataset and code publicly available here: http://www.nlpr.ia.ac.cn/cip/dataset.htm.

cs.CL↗

arXiv2019

Are You for Real? Detecting Identity Fraud via Dialogue Interactions

Identity fraud detection is of great importance in many real-world scenarios such as the financial industry. However, few studies addressed this problem before. In this paper, we focus on identity fraud detection in loan applications and propose to solve this problem with a novel interactive dialogue system which consists of two modules. One is the knowledge graph (KG) constructor organizing the personal information for each loan applicant. The other is structured dialogue management that can dynamically generate a series of questions based on the personal KG to ask the applicants and determine their identity states. We also present a heuristic user simulator based on problem analysis to evaluate our method. Experiments have shown that the trainable dialogue system can effectively detect fraudsters, and achieve higher recognition accuracy compared with rule-based systems. Furthermore, our learned dialogue strategies are interpretable and flexible, which can help promote real-world applications.

cs.CL↗

arXiv2019

Testing a quintessence model with Yukawa interaction from cosmological observations and N-body simulations

We consider a quintessence model with Yukawa interaction between dark energy and dark matter and constrain this model by employing the recent cosmological data including the updated cosmic microwave background (CMB) measurements from Planck 2015, the weak gravitational lensing measurements from Kilo Degree Survey (KiDS) and redshift-space distortions. We find that an interaction in the dark sector is compatible with observations. The updated Planck data can significantly improve the constraints compared with the previous results from Planck 2013, while the KiDS data has less constraining power than Planck. The Yukawa interaction model is found to be moderately favored by Planck and able to alleviate the discordance between weak lensing measurements and CMB measurements as previously inferred from the standard Lambda cold dark matter model. N-body simulations for Yukawa interaction model is also performed. We find that using the halo density profile is plausible to improve the constraints significantly in the future.

astro-ph.CO↗

arXiv2019

Understanding Memory Modules on Learning Simple Algorithms

Recent work has shown that memory modules are crucial for the generalization ability of neural networks on learning simple algorithms. However, we still have little understanding of the working mechanism of memory modules. To alleviate this problem, we apply a two-step analysis pipeline consisting of first inferring hypothesis about what strategy the model has learned according to visualization and then verify it by a novel proposed qualitative analysis method based on dimension reduction. Using this method, we have analyzed two popular memory-augmented neural networks, neural Turing machine and stack-augmented neural network on two simple algorithm tasks including reversing a random sequence and evaluation of arithmetic expressions. Results have shown that on the former task both models can learn to generalize and on the latter task only the stack-augmented model can do so. We show that different strategies are learned by the models, in which specific categories of input are monitored and different policies are made based on that to change the memory.

cs.LG↗

arXiv2019

Sequence Generation: From Both Sides to the Middle

The encoder-decoder framework has achieved promising process for many sequence generation tasks, such as neural machine translation and text summarization. Such a framework usually generates a sequence token by token from left to right, hence (1) this autoregressive decoding procedure is time-consuming when the output sentence becomes longer, and (2) it lacks the guidance of future context which is crucial to avoid under translation. To alleviate these issues, we propose a synchronous bidirectional sequence generation (SBSG) model which predicts its outputs from both sides to the middle simultaneously. In the SBSG model, we enable the left-to-right (L2R) and right-to-left (R2L) generation to help and interact with each other by leveraging interactive bidirectional attention network. Experiments on neural machine translation (En-De, Ch-En, and En-Ro) and text summarization tasks show that the proposed model significantly speeds up decoding while improving the generation quality compared to the autoregressive Transformer.

cs.CL↗

arXiv2019

Incremental Learning from Scratch for Task-Oriented Dialogue Systems

Clarifying user needs is essential for existing task-oriented dialogue systems. However, in real-world applications, developers can never guarantee that all possible user demands are taken into account in the design phase. Consequently, existing systems will break down when encountering unconsidered user needs. To address this problem, we propose a novel incremental learning framework to design task-oriented dialogue systems, or for short Incremental Dialogue System (IDS), without pre-defining the exhaustive list of user needs. Specifically, we introduce an uncertainty estimation module to evaluate the confidence of giving correct responses. If there is high confidence, IDS will provide responses to users. Otherwise, humans will be involved in the dialogue process, and IDS can learn from human intervention through an online learning module. To evaluate our method, we propose a new dataset which simulates unanticipated user needs in the deployment stage. Experiments show that IDS is robust to unconsidered user actions, and can update itself online by smartly selecting only the most effective training data, and hence attains better performance with less annotation cost.

cs.CL↗

arXiv2019

Memory Consolidation for Contextual Spoken Language Understanding with Dialogue Logistic Inference

Dialogue contexts are proven helpful in the spoken language understanding (SLU) system and they are typically encoded with explicit memory representations. However, most of the previous models learn the context memory with only one objective to maximizing the SLU performance, leaving the context memory under-exploited. In this paper, we propose a new dialogue logistic inference (DLI) task to consolidate the context memory jointly with SLU in the multi-task framework. DLI is defined as sorting a shuffled dialogue session into its original logical order and shares the same memory encoder and retrieval mechanism as the SLU model. Our experimental results show that various popular contextual SLU models can benefit from our approach, and improvements are quite impressive, especially in slot filling.

cs.CL↗

arXiv2019

Towards a Model-Independent Measurement of the Halo Mass Function with Observables

In the CDM paradigm, the halo mass function is a sensitive probe of the cosmic structure. In observations, halo mass is typically estimated from its relation with other observables. The resulting halo mass function is subject to systematic bias, such as the Eddington bias, due to the scatter or uncertainty in the observable - mass relation. Exact correction for the bias is not easy, as predictions for the observables are typically model-dependent in simulations. In this paper, we point out an interesting feature in the halo mass function of the concordence $Λ$CDM model: the total halo mass within each evenly-spaced logarithmic mass bin is approximately the same over a large mass range. We show that this property allows us to construct an almost bias-free halo mass function using only an observable (as a halo mass estimator) and stacked weak lensing measurements as long as the scatter between the true halo mass and the observable-inferred mass has a stable form in logarithmic units. The method is not sensitive to the form of the mass-observable relation. We test the idea using cosmological simulations, and show that the method performs very well for realistic observables.

astro-ph.CO↗

arXiv2019

Synchronous Bidirectional Neural Machine Translation

Existing approaches to neural machine translation (NMT) generate the target language sequence token by token from left to right. However, this kind of unidirectional decoding framework cannot make full use of the target-side future contexts which can be produced in a right-to-left decoding direction, and thus suffers from the issue of unbalanced outputs. In this paper, we introduce a synchronous bidirectional neural machine translation (SB-NMT) that predicts its outputs using left-to-right and right-to-left decoding simultaneously and interactively, in order to leverage both of the history and future information at the same time. Specifically, we first propose a new algorithm that enables synchronous bidirectional decoding in a single model. Then, we present an interactive decoding model in which left-to-right (right-to-left) generation does not only depend on its previously generated outputs, but also relies on future contexts predicted by right-to-left (left-to-right) decoding. We extensively evaluate the proposed SB-NMT model on large-scale NIST Chinese-English, WMT14 English-German, and WMT18 Russian-English translation tasks. Experimental results demonstrate that our model achieves significant improvements over the strong Transformer model by 3.92, 1.49 and 1.04 BLEU points respectively, and obtains the state-of-the-art performance on Chinese-English and English-German translation tasks.

cs.CL↗