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

Ziyue Li

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

ResearcherUnclaimed source record
Artificial Intelligenceeess.SPhep-phhep-thMachine LearningApplicationsComputer VisionDigital LibrariesInformation Theorymath.IT

Catalog footprint

What is connected

9works
10topics
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

9 published item(s)

preprint2026arXiv

Review of Passenger Flow Modelling Approaches Based on a Bibliometric Analysis

This paper presents a bibliometric analysis of the field of short-term passenger flow forecasting within local public transit, covering 814 publications that span from 1984 to 2024. In addition to common bibliometric analysis tools, a variant of a citation network was developed, and topic modelling was conducted. The analysis reveals that research activity exhibited sporadic patterns prior to 2008, followed by a marked acceleration, characterised by a shift from conventional statistical and machine learning methodologies (e.g., ARIMA, SVM, and basic neural networks) to specialised deep learning architectures. Based on this insight, a connection to more general fields such as machine learning and time series modelling was established. In addition to modelling, spatial, linguistic, and modal biases were identified and findings from existing secondary literature were validated and quantified. This revealed existing gaps, such as constrained data fusion, open (multivariate) data, and underappreciated challenges related to model interpretability, cost-efficiency, and a balance between algorithmic performance and practical deployment considerations. In connection with the superordinate fields, the growth in relevance of foundation models is also noteworthy.

0 citations0 reviewsNo code linkedOpen access
preprint2024arXiv

Online Test-Time Adaptation of Spatial-Temporal Traffic Flow Forecasting

Accurate spatial-temporal traffic flow forecasting is crucial in aiding traffic managers in implementing control measures and assisting drivers in selecting optimal travel routes. Traditional deep-learning based methods for traffic flow forecasting typically rely on historical data to train their models, which are then used to make predictions on future data. However, the performance of the trained model usually degrades due to the temporal drift between the historical and future data. To make the model trained on historical data better adapt to future data in a fully online manner, this paper conducts the first study of the online test-time adaptation techniques for spatial-temporal traffic flow forecasting problems. To this end, we propose an Adaptive Double Correction by Series Decomposition (ADCSD) method, which first decomposes the output of the trained model into seasonal and trend-cyclical parts and then corrects them by two separate modules during the testing phase using the latest observed data entry by entry. In the proposed ADCSD method, instead of fine-tuning the whole trained model during the testing phase, a lite network is attached after the trained model, and only the lite network is fine-tuned in the testing process each time a data entry is observed. Moreover, to satisfy that different time series variables may have different levels of temporal drift, two adaptive vectors are adopted to provide different weights for different time series variables. Extensive experiments on four real-world traffic flow forecasting datasets demonstrate the effectiveness of the proposed ADCSD method. The code is available at https://github.com/Pengxin-Guo/ADCSD.

0 citations0 reviewsNo code linkedOpen access
preprint2022arXiv

Domain Generalization using Pretrained Models without Fine-tuning

Fine-tuning pretrained models is a common practice in domain generalization (DG) tasks. However, fine-tuning is usually computationally expensive due to the ever-growing size of pretrained models. More importantly, it may cause over-fitting on source domain and compromise their generalization ability as shown in recent works. Generally, pretrained models possess some level of generalization ability and can achieve decent performance regarding specific domains and samples. However, the generalization performance of pretrained models could vary significantly over different test domains even samples, which raises challenges for us to best leverage pretrained models in DG tasks. In this paper, we propose a novel domain generalization paradigm to better leverage various pretrained models, named specialized ensemble learning for domain generalization (SEDGE). It first trains a linear label space adapter upon fixed pretrained models, which transforms the outputs of the pretrained model to the label space of the target domain. Then, an ensemble network aware of model specialty is proposed to dynamically dispatch proper pretrained models to predict each test sample. Experimental studies on several benchmarks show that SEDGE achieves significant performance improvements comparing to strong baselines including state-of-the-art method in DG tasks and reduces the trainable parameters by ~99% and the training time by ~99.5%.

0 citations0 reviewsNo code linkedOpen access
preprint2021arXiv

The Twelvefold Way of Non-Sequential Lossless Compression

Many information sources are not just sequences of distinguishable symbols but rather have invariances governed by alternative counting paradigms such as permutations, combinations, and partitions. We consider an entire classification of these invariances called the twelvefold way in enumerative combinatorics and develop a method to characterize lossless compression limits. Explicit computations for all twelve settings are carried out for i.i.d. uniform and Bernoulli distributions. Comparisons among settings provide quantitative insight.

0 citations0 reviewsNo code linkedOpen access
preprint2020arXiv

Long-Short Term Spatiotemporal Tensor Prediction for Passenger Flow Profile

Spatiotemporal data is very common in many applications, such as manufacturing systems and transportation systems. It is typically difficult to be accurately predicted given intrinsic complex spatial and temporal correlations. Most of the existing methods based on various statistical models and regularization terms, fail to preserve innate features in data alongside their complex correlations. In this paper, we focus on a tensor-based prediction and propose several practical techniques to improve prediction. For long-term prediction specifically, we propose the "Tensor Decomposition + 2-Dimensional Auto-Regressive Moving Average (2D-ARMA)" model, and an effective way to update prediction real-time; For short-term prediction, we propose to conduct tensor completion based on tensor clustering to avoid oversimplifying and ensure accuracy. A case study based on the metro passenger flow data is conducted to demonstrate the improved performance.

0 citations0 reviewsNo code linkedOpen access
preprint2015arXiv

Interpolating Helicity Spinors Between the Instant Form and the Light-front Form

We discuss the helicity spinors interpolating between the instant form dynamics (IFD) and the front form dynamics, or the light-front dynamics (LFD), and present the interpolating helicity amplitudes as well as their squares for the scattering of two fermions, and the annihilation of fermion and anti-fermion. We parametrize the interpolation between the two dynamics, IFD and LFD, by an interpolation angle and derive not only the generalized helicity spinors in the $(0,J)\oplus(J,0)$ chiral representation that links naturally the two typical IFD vs. LFD helicity spinors but also the generalized Melosh transformation that relates these generalized helicity spinors to the usual Dirac spinors. Analyzing the directions of the particle momentum and spin with the variation of the interpolation angle, we inspect the whole landscape of the generalized helicity intermediating between the usual Jacob-Wick helicity in the IFD and the light-front helicity in the LFD. Our analysis clarifies the characteristic difference of the helicity amplitudes between the IFD and the LFD. In particular, we find that the behavior of the angle between the momentum direction and the spin direction bifurcates at a critical interpolation angle and the IFD and the LFD separately belong to the two different branches bifurcated at this critical interpolation angle. This finding further clarifies any conceivable confusion in the prevailing notion of the equivalence between the infinite momentum frame and the LFD. The existence of the universal J-curve found in our previous works of scalar field theory and the sQED theory is confirmed in the present work of interpolating helicity amplitudes for the fermion scattering and annihilation processes. In conjunction with the bifurcation of branches, the two boundaries appear in the interpolating helicity amplitudes and interestingly the J-curve persists within these two boundaries.

0 citations0 reviewsNo code linkedOpen access
preprint2015arXiv

Update on the Light-Front Quark Model and Mass Spectrum Calculations for Ground State Pseudoscalar and Vector Mesons

We present an update on the meson mass spectra calculation with the light-front quark model constrained by the variational principle for the QCD-motivated effective Hamiltonian. By smearing out the Dirac delta function in the hyperfine interaction, we were able to avoid the negative infinity that one encounters when using variational principle for the entire Hamiltonian. We obtained a better fitting for the mass spectra of ground state pseudoscalar and vector mesons from $π$ to $Υ$, compared to the previous calculation that handled the hyperfine interaction as a perturbation rather than including it in the parameterization process. Our new parameters are further tested in decay constant calculation. We showed that by taking a larger harmonic oscillator basis in our trial wave function, the decay constants calculated from our model can be improved to agree reasonably well with the experimental data.

0 citations0 reviewsNo code linkedOpen access
preprint2015arXiv

Variational Analysis of Mass Spectra and Decay Constants for Ground State Pseudoscalar and Vector Mesons in Light-Front Quark Model

Using the variational principle, we compute mass spectra and decay constants of ground state pseudoscalar and vector mesons in the light-front quark model (LFQM) with the QCD-motivated effective Hamiltonian including the hyperfine interaction. By smearing out the Dirac delta function in the hyperfine interaction, we avoid the issue of negative infinity in applying the variational principle to the computation of meson mass spectra and provide analytic expressions for the meson mass spectra. Our analysis with the smeared hyperfine interaction indicates that the interaction for the heavy meson sector including the bottom and charm quarks gets more point-like. We also consider the flavor mixing effect in our analysis and determine the mixing angles from the mass spectra of $(ω,ϕ)$ and $(η,η')$. Our variational analysis with the trial wave function including the two lowest order harmonic oscillator basis functions appears to improve the agreement with the data of meson decay constants and the heavy meson mass spectra over the previous computation handling the hyperfine interaction as perturbation.

0 citations0 reviewsNo code linkedOpen access
preprint2014arXiv

The Electromagnetic Gauge Field Interpolation between the Instant Form and the Front Form of the Hamiltonian Dynamics

We present the electromagnetic gauge field interpolation between the instant form and the front form of the relativistic Hamiltonian dynamics and extend our interpolation of the scattering amplitude presented in the simple scalar field theory to the case of the electromagnetic gauge field theory with the scalar fermion fields known as the sQED theory. We find that the Coulomb gauge in the instant form dynamics (IFD) and the light-front gauge in the front form dynamics, or the light-front dynamics (LFD), are naturally linked by the unified general physical gauge that interpolates between these two forms of dynamics and derive the spin-1 polarization vector for the photon that can be generally applicable for any interpolation angle. Corresponding photon propagator for an arbitrary interpolation angle is found and examined in terms of the gauge field polarization and the interpolating time ordering. Using these results, we calculate the lowest-order scattering processes for an arbitrary interpolation angle in sQED. We provide an example of breaking the reflection symmetry under the longitudinal boost, $P^z \leftrightarrow -P^z$, for the time-ordered scattering amplitude in any interpolating dynamics except the LFD and clarify the confusion in the prevailing notion of the equivalence between the infinite momentum frame (IMF) and the LFD. The particular correlation found in our previous analysis of the scattering amplitude in the simple scalar field theory, coined as the J-shaped correlation, between the total momentum of the system and the interpolation angle persists in the present analysis of the sQED scattering amplitude. We discuss the singular behavior of this correlation in conjunction with the zero-mode issue in the LFD.

0 citations0 reviewsNo code linkedOpen access