BZPEERPapers, people, evidence and research paths
Menu

Discover

SearchFind papers, people, topics, institutions and opportunities from one place.GraphSee how papers, authors, topics and evidence connect.

Research tools

HomeContinue from the papers, people and topics that matter now.CollectionsTurn saved papers into reading lists, evidence maps and shareable context.ResearchReconstruct the path that led you from search to paper to question.CollaborationMove from a paper, topic or expert request into a focused thread.PublishCreate full-text research articles with metadata, formulas and versions.

Network

ExpertsFind specialists whose work and reviews explain why they can help.CommunitiesJoin research groups organized around papers, topics and shared evidence.

Opportunities

ListingsBrowse roles, calls and collaborations connected to real research context.

Account

Log inReturn to your saved papers, graph views and active threads.Create accountStart saving papers, building a research trail and joining teams.
Log inCreate account

Source author record

G. Sun

G. Sun appears in the imported research catalog. Authorship, coauthor and topic links are available while profile ownership is still unclaimed.

ResearcherUnclaimed source record
cond-mat.str-elcond-mat.quant-gasComputation and Languageeess.ASMachine LearningSoundArtificial Intelligenceastro-ph.GAeess.IVhep-exnucl-exphysics.ins-det

Catalog footprint

What is connected

14works
12topics
4close collaborators

Actions

Connect this record

Log in to claim
Open graphBrowse works

Research graph

See the researcher in context

Open full explorer

Inspect adjacent papers, topics, institutions and collaborators without losing the researcher page.

Building this map preview

BZPEER is loading the nearby papers, people, topics and institutions for this page.

Published work

14 published item(s)

preprint2021arXiv

Content-Aware Speaker Embeddings for Speaker Diarisation

Recent speaker diarisation systems often convert variable length speech segments into fixed-length vector representations for speaker clustering, which are known as speaker embeddings. In this paper, the content-aware speaker embeddings (CASE) approach is proposed, which extends the input of the speaker classifier to include not only acoustic features but also their corresponding speech content, via phone, character, and word embeddings. Compared to alternative methods that leverage similar information, such as multitask or adversarial training, CASE factorises automatic speech recognition (ASR) from speaker recognition to focus on modelling speaker characteristics and correlations with the corresponding content units to derive more expressive representations. CASE is evaluated for speaker re-clustering with a realistic speaker diarisation setup using the AMI meeting transcription dataset, where the content information is obtained by performing ASR based on an automatic segmentation. Experimental results showed that CASE achieved a 17.8% relative speaker error rate reduction over conventional methods.

0 citations0 reviewsNo code linkedOpen access
preprint2021arXiv

Transformer Language Models with LSTM-based Cross-utterance Information Representation

The effective incorporation of cross-utterance information has the potential to improve language models (LMs) for automatic speech recognition (ASR). To extract more powerful and robust cross-utterance representations for the Transformer LM (TLM), this paper proposes the R-TLM which uses hidden states in a long short-term memory (LSTM) LM. To encode the cross-utterance information, the R-TLM incorporates an LSTM module together with a segment-wise recurrence in some of the Transformer blocks. In addition to the LSTM module output, a shortcut connection using a fusion layer that bypasses the LSTM module is also investigated. The proposed system was evaluated on the AMI meeting corpus, the Eval2000 and the RT03 telephone conversation evaluation sets. The best R-TLM achieved 0.9%, 0.6%, and 0.8% absolute WER reductions over the single-utterance TLM baseline, and 0.5%, 0.3%, 0.2% absolute WER reductions over a strong cross-utterance TLM baseline on the AMI evaluation set, Eval2000 and RT03 respectively. Improvements on Eval2000 and RT03 were further supported by significance tests. R-TLMs were found to have better LM scores on words where recognition errors are more likely to occur. The R-TLM WER can be further reduced by interpolation with an LSTM-LM.

0 citations0 reviewsNo code linkedOpen access
preprint2020arXiv

Cross-Utterance Language Models with Acoustic Error Sampling

The effective exploitation of richer contextual information in language models (LMs) is a long-standing research problem for automatic speech recognition (ASR). A cross-utterance LM (CULM) is proposed in this paper, which augments the input to a standard long short-term memory (LSTM) LM with a context vector derived from past and future utterances using an extraction network. The extraction network uses another LSTM to encode surrounding utterances into vectors which are integrated into a context vector using either a projection of LSTM final hidden states, or a multi-head self-attentive layer. In addition, an acoustic error sampling technique is proposed to reduce the mismatch between training and test-time. This is achieved by considering possible ASR errors into the model training procedure, and can therefore improve the word error rate (WER). Experiments performed on both AMI and Switchboard datasets show that CULMs outperform the LSTM LM baseline WER. In particular, the CULM with a self-attentive layer-based extraction network and acoustic error sampling achieves 0.6% absolute WER reduction on AMI, 0.3% WER reduction on the Switchboard part and 0.9% WER reduction on the Callhome part of Eval2000 test set over the respective baselines.

0 citations0 reviewsNo code linkedOpen access
preprint2016arXiv

The Global 21-cm Signal in the Context of the High-z Galaxy Luminosity Function

Motivated by recent progress in studies of the high-$z$ Universe, we build a new model for the global 21-cm signal that is explicitly calibrated to measurements of the galaxy luminosity function (LF) and further tuned to match the Thomson scattering optical depth of the cosmic microwave background, $τ_e$. Assuming that the $z \lesssim 8$ galaxy population can be smoothly extrapolated to higher redshifts, the recent decline in best-fit values of $τ_e$ and the inefficient heating induced by X-ray binaries (HMXBs; the presumptive sources of the X-ray background at high-$z$) imply that the entirety of cosmic reionization and reheating occurs at redshifts $z \lesssim 12$. In contrast to past global 21-cm models, whose $z \sim 20$ ($ν\sim 70$ MHz) absorption features and strong $\sim 25$ mK emission features were driven largely by the assumption of efficient early star-formation and X-ray heating, our new fiducial model peaks in absorption at $ν\sim 110$ MHz at a depth of $\sim -160$ mK and has a negligible emission component. As a result, a strong emission signal would provide convincing evidence that HMXBs are not the only drivers of cosmic reheating. Shallow absorption troughs should accompany strong heating scenarios, but could also be caused by a low escape fraction of Lyman-Werner photons. Generating signals with troughs at $ν\lesssim 95$ MHz requires a floor in the star-formation efficiency in halos below $\sim 10^{9} M_{\odot}$, which is equivalent to steepening the faint-end of the galaxy LF. These findings demonstrate that the global 21-cm signal is a powerful complement to current and future galaxy surveys and efforts to better understand the interstellar medium in high-$z$ galaxies.

0 citations0 reviewsNo code linkedOpen access
preprint2016arXiv

Topological quasi-one-dimensional state of interacting spinless electrons

By decreasing the transversal confinement potential in interacting one-dimensional spinless electrons and populating the second energetically lowest sub-band, for not too strong interactions system transitions into a quasi-one-dimensional state with dominant superconducting correlations and one gapless mode. By combining effective field theory approach and numerical density matrix renormalization group simulations we show that this quasi-one-dimensional state is a topological state that hosts zero-energy edge modes. We also study the single-particle correlations across the interface between this quasi-one-dimensional and single-channel states.

0 citations0 reviewsNo code linkedOpen access
preprint2015arXiv

Fidelity at Berezinskii-Kosterlitz-Thouless quantum phase transitions

We clarify the long-standing controversy concerning the behavior of the ground state fidelity in the vicinity of a quantum phase transition of the Berezinskii-Kosterlitz-Thouless type in one-dimensional systems. Contrary to the prediction based on the Gaussian approximation of the Luttinger liquid approach, it is shown that the fidelity susceptibility does not diverge at the transition, but has a cusp-like peak $χ_c- χ(λ)\sim \sqrt{|λ_c-λ|} $, where $λ$ is a parameter driving the transition, and $χ_c$ is the peak value at the transition point $λ=λ_c$. Numerical claims of the logarithmic divergence of fidelity susceptibility with the system size (or temperature) are explained by logarithmic corrections due to marginal operators, which is supported by numerical calculations for large systems.

0 citations0 reviewsNo code linkedOpen access
preprint2015arXiv

Ground-state phases of rung-alternated spin-1/2 Heisenberg ladder

The ground-state phase diagram of Heisenberg spin-1/2 system on a two-leg ladder with rung alternation is studied by combining analytical approaches with numerical simulations. For the case of ferromagnetic leg exchanges a unique ferrimagnetic ground state emerges, whereas for the case of antiferromagnetic leg exchanges several different ground states are stabilized depending on the ratio between exchanges along legs and rungs. For the more general case of a honeycomb-ladder model for the case of ferromagnetic leg exchanges besides usual rung-singlet and saturated ferromagnetic states we obtain a ferrimagnetic Luttinger liquid phase with both linear and quadratic low energy dispersions and ground state magnetization continuously changing with system parameters. For the case of antiferromagnetic exchanges along legs, different dimerized states including states with additional topological order are suggested to be realized.

0 citations0 reviewsNo code linkedOpen access
preprint2015arXiv

JUNO Conceptual Design Report

The Jiangmen Underground Neutrino Observatory (JUNO) is proposed to determine the neutrino mass hierarchy using an underground liquid scintillator detector. It is located 53 km away from both Yangjiang and Taishan Nuclear Power Plants in Guangdong, China. The experimental hall, spanning more than 50 meters, is under a granite mountain of over 700 m overburden. Within six years of running, the detection of reactor antineutrinos can resolve the neutrino mass hierarchy at a confidence level of 3-4$σ$, and determine neutrino oscillation parameters $\sin^2θ_{12}$, $Δm^2_{21}$, and $|Δm^2_{ee}|$ to an accuracy of better than 1%. The JUNO detector can be also used to study terrestrial and extra-terrestrial neutrinos and new physics beyond the Standard Model. The central detector contains 20,000 tons liquid scintillator with an acrylic sphere of 35 m in diameter. $\sim$17,000 508-mm diameter PMTs with high quantum efficiency provide $\sim$75% optical coverage. The current choice of the liquid scintillator is: linear alkyl benzene (LAB) as the solvent, plus PPO as the scintillation fluor and a wavelength-shifter (Bis-MSB). The number of detected photoelectrons per MeV is larger than 1,100 and the energy resolution is expected to be 3% at 1 MeV. The calibration system is designed to deploy multiple sources to cover the entire energy range of reactor antineutrinos, and to achieve a full-volume position coverage inside the detector. The veto system is used for muon detection, muon induced background study and reduction. It consists of a Water Cherenkov detector and a Top Tracker system. The readout system, the detector control system and the offline system insure efficient and stable data acquisition and processing.

0 citations0 reviewsNo code linkedOpen access
preprint2014arXiv

Density-Dependent Synthetic Gauge Fields Using Periodically Modulated Interactions

We show that density-dependent synthetic gauge fields may be engineered by combining periodically modu- lated interactions and Raman-assisted hopping in spin-dependent optical lattices. These fields lead to a density- dependent shift of the momentum distribution, may induce superfluid-to-Mott insulator transitions, and strongly modify correlations in the superfluid regime. We show that the interplay between the created gauge field and the broken sublattice symmetry results, as well, in an intriguing behavior at vanishing interactions, characterized by the appearance of a fractional Mott insulator.

0 citations0 reviewsNo code linkedOpen access
preprint2014arXiv

Topological order-by-disorder in orbitally degenerate dipolar bosons in zig-zag lattice

Spinor bosons offer conceptually simple picture of macroscopic quantum behavior of topological order-by-disorder: Paramagnetic state of two-component dipolar bosons in orbitally degenerate zig-zag lattice is unstable against infinitezimal quantum fluctuations of orbitals towards developing non-local hidden order. Adjacent to the topological state locally correlated exact ground state with spontaneously quadrupoled lattice constant is realized for the broad parameter regime.

0 citations0 reviewsNo code linkedOpen access
preprint2013arXiv

Ferromagnetic spin-orbital liquid of dipolar fermions in zigzag lattices

Two-component dipolar fermions in zigzag optical lattices allow for the engineering of spin-orbital models. We show that dipolar lattice fermions permit the exploration of a regime typically unavailable in solid-state compounds that is characterized by a novel spin-liquid phase with a finite magnetization and spontaneously broken SU(2) symmetry. This peculiar spin liquid may be understood as a Luttinger liquid of composite particles consisting of bound states of spin waves and orbital domain walls moving in an unsaturated ferromagnetic background. In addition, we show that the system exhibits a boundary phase transitions involving non-local entanglement of edge spins.

0 citations0 reviewsNo code linkedOpen access
preprint2013arXiv

Spin-orbit coupled fermions in ladder-like optical lattices at half-filling

We study the ground-state phase diagram of two-component fermions loaded in a ladder-like lattice at half filling in the presence of spin-orbit coupling. For repulsive fermions with unidirectional spin-orbit coupling along the legs we identify a Néel state which is separated from rung-singlet and ferromagnetic states by Ising phase transition lines. These lines cross for maximal spin-orbit coupling and a direct Gaussian phase transition between rung-singlet and ferro phases is realized. For the case of Rashba-like spin-orbit coupling, besides the rung singlet phases two distinct striped ferromagnetic phases are formed. In case of attractive fermions with spin-orbit coupling at half-filling for decoupled chains we identify a dimerized state that separates a singlet superconductor and a ferromagnetic states.

0 citations0 reviewsNo code linkedOpen access
preprint2012arXiv

Exploring spin-orbital models with dipolar fermions in zig-zag optical lattices

Ultra-cold dipolar spinor fermions in zig-zag type optical lattices can mimic spin-orbital models relevant in solid-state systems, as transition-metal oxides with partially filled d-levels, with the interesting advantage of reviving the quantum nature of orbital fluctuations. We discuss two different physical systems in which these models may be simulated, showing that the interplay between lattice geometry and spin-orbital quantum dynamics produces a wealth of novel quantum phases.

0 citations0 reviewsNo code linkedOpen access
preprint2010arXiv

Influence of N*-resonances on hyperon production in the channel pp->K+ Lambda p at 2.95, 3.20 and 3.30 GeV/c beam momentum

Hyperon production in the threshold region was studied in the reaction pp -> K+Lp using the time-of-flight spectrometer COSY-TOF. Exclusive data, covering the full phase-space, were taken at the three different beam momenta of p_beam=2.95, 3.20 and 3.30 GeV/c, corresponding to excess energies of epsilon=204, 285 and 316 MeV, respectively. Total cross-sections were deduced for the three beam momenta to be 23.9+/-0.8 +/-2.0 ub, 28.4+/-1.3 +/-2.2 ub and 35.0+/-1.3 +/-3.0 ub. Differential observables including Dalitz plots were obtained. The analysis of the Dalitz plots reveals a strong influence of the N(1650)-resonance at p_beam=2.95 GeV/c, whereas for the higher momenta an increasing relative contribution of the N(1710)- and/or of the N(1720)-resonance was observed. In addition, the pL-final-state interaction turned out to have a significant influence on the Dalitz plot distribution.

0 citations0 reviewsNo code linkedOpen access