arXiv subjects

Liang Zheng

Publications and source records attributed to Liang Zheng.

At least 37 records · Page 2Linked to original sources

arXiv2026

Studying Energy-Energy Correlators in pp Collisions at the LHC with a Jet-Free Event-Topology Method

We present a jet-free approach for measuring energy-energy correlators (EEC) in proton-proton (pp) collisions at the Large Hadron Collider (LHC), employing an event-topology method that does not rely on explicit jet reconstruction. Using the leading charged hadron as a reference axis, the azimuthal plane is divided into Toward and Transverse regions, enabling a robust background subtraction and extending EEC measurements into the low $p_T$ regime where conventional jet-based approaches become unreliable. The method is validated through comparisons with conventional jet reconstruction results. We systematically explore the dependence of the EEC on the leading-particle transverse momentum and parton flavor. The observed scaling between the EEC peak position and the hard scale suggests that this topology-based EEC captures effectively the transition between perturbative and non-perturbative QCD regimes. Distinct differences are found between quark- and gluon-initiated events, reflecting their different color charges and radiation patterns. Extending the analysis to heavy flavor, EECs triggered by leading charm mesons exhibit a suppressed magnitude and a peak shifted toward larger angular separations relative to inclusive charged-particle triggers, providing a direct manifestation of the dead-cone effect. This jet-free EEC framework offers a simple and experimentally robust tool for studying the scale and flavor dependence of the QCD dynamics, with promising applications to proton-nucleus and heavy-ion collisions at the LHC.

hep-ph↗

arXiv2026

Flexible Geometric Guidance for Probabilistic Human Pose Estimation with Diffusion Models

3D human pose estimation from 2D images is a challenging problem due to depth ambiguity and occlusion. Because of these challenges the task is underdetermined, where there exists multiple -- possibly infinite -- poses that are plausible given the image. Despite this, many prior works assume the existence of a deterministic mapping and estimate a single pose given an image. Furthermore, methods based on machine learning require a large amount of paired 2D-3D data to train and suffer from generalization issues to unseen scenarios. To address both of these issues, we propose a framework for pose estimation using diffusion models, which enables sampling from a probability distribution over plausible poses which are consistent with a 2D image. Our approach falls under the guidance framework for conditional generation, and guides samples from an unconditional diffusion model, trained only on 3D data, using the gradients of the heatmaps from a 2D keypoint detector. We evaluate our method on the Human 3.6M dataset under best-of-$m$ multiple hypothesis evaluation, showing state-of-the-art performance among methods which do not require paired 2D-3D data for training. We additionally evaluate the generalization ability using the MPI-INF-3DHP and 3DPW datasets and demonstrate competitive performance. Finally, we demonstrate the flexibility of our framework by using it for novel tasks including pose generation and pose completion, without the need to train bespoke conditional models. We make code available at https://github.com/fsnelgar/diffusion_pose .

cs.CV↗

arXiv2026

Investigating the transverse-momentum- and pseudorapidity-dependent flow vector decorrelation in p--Pb collisions with a Multi-Phase Transport model

The event-by-event fluctuations in the initial energy density of the nuclear collisions lead to the decorrelation of second order flow vector, as known as its transverse-momentum ($p_{\mathrm{T}}$) and pseudorapidity ($η$) dependence as observed in high-energy heavy-ion collisions. Existing measurements at the CERN Large Hadron Collider shown that these decorrelations are also observed in small collision systems. In this work, a systematic study of the transverse-momentum- and pseudorapidity-dependent flow vector decorrelation is performed in p--Pb collisions at the 5.02 TeV with A Multi-Phase Transport (AMPT) model using different tunings of the initial conditions, partonic and hadronic interactions. It is found that the string-melting version of the AMPT model provides a reasonable description of the measured flow vector decorrelation as a function of $p_{\mathrm{T}}$ and $η$. We demonstrate that the hadronic scatterings do not have significant impact on decorrelation in p--Pb collisions for different centrality selections, while both initial conditions and partonic interactions influence the magnitude of the decorrelations. In addition, we found that the subtraction of the nonflow, especially the long-range jet correlation, is crucial for the accurate extraction of the flow vector decorrelation in small collision systems. The comparison of data and model presented in this paper provide further insights in understanding the fluctuations of the flow vector with $p_{\mathrm{T}}$ and $η$ in small collision systems and has referential value for future measurements.

hep-ph↗

arXiv2026

Open charm production and $Λ_{c}^{+}/D^{0}$ ratio in pp and Au+Au collisions at the RHIC

We study open charm hadrons production in pp and Au+Au collisions at $\sqrt{s_{\mathrm{NN}}} = 200$~GeV using an improved a multi-phase transport (AMPT) model. Specifically, we show the transverse-momentum spectra and nuclear modification factors $R_{\mathrm{AA}}$ of $D^{0}$ mesons and $Λ_{c}^{+}$ baryons, as well as the $Λ_{c}^{+}/D^{0}$ ratio in pp and Au+Au collisions. The results obtained from the AMPT model simulations are compared with the STAR experimental data and found to be consistent. We further investigate the $Λ_{c}^{+}/D^{0}$ ratio by evaluating contributions from coalescence, fragmentation, and the combined coalescence+fragmentation mechanisms, and we find that fragmentation alone underestimates the pronounced enhancement in Au+Au relative to pp at low and intermediate $p_{\mathrm{T}}$, whereas the coalescence+fragmentation mechanism reproduces the observed trend significantly better. These results indicate that coalescence plays a key role in charm baryon productions and helps constrain the relative importance of different hadronization mechanisms in the ultra-relativistic nuclear collisions.

hep-ph↗

arXiv2026

Mosaic: Unlocking Long-Context Inference for Diffusion LLMs via Global Memory Planning and Dynamic Peak Taming

Diffusion-based large language models (dLLMs) have emerged as a promising paradigm, utilizing simultaneous denoising to enable global planning and iterative refinement. While these capabilities are particularly advantageous for long-context generation, deploying such models faces a prohibitive memory capacity barrier stemming from severe system inefficiencies. We identify that existing inference systems are ill-suited for this paradigm: unlike autoregressive models constrained by the cumulative KV-cache, dLLMs are bottlenecked by transient activations recomputed at every step. Furthermore, general-purpose memory reuse mechanisms lack the global visibility to adapt to dLLMs' dynamic memory peaks, which toggle between logits and FFNs. To address these mismatches, we propose Mosaic, a memory-efficient inference system that shifts from local, static management to a global, dynamic paradigm. Mosaic integrates a mask-only logits kernel to eliminate redundancy, a lazy chunking optimizer driven by an online heuristic search to adaptively mitigate dynamic peaks, and a global memory manager to resolve fragmentation via virtual addressing. Extensive evaluations demonstrate that Mosaic achieves an average 2.71$\times$ reduction in the memory peak-to-average ratio and increases the maximum inference sequence length supportable on identical hardware by 15.89-32.98$\times$. This scalability is achieved without compromising accuracy and speed, and in fact reducing latency by 4.12%-23.26%.

cs.LG↗

arXiv2026

InpaintHuman: Reconstructing Occluded Humans with Multi-Scale UV Mapping and Identity-Preserving Diffusion Inpainting

Reconstructing complete and animatable 3D human avatars from monocular videos remains challenging, particularly under severe occlusions. While 3D Gaussian Splatting has enabled photorealistic human rendering, existing methods struggle with incomplete observations, often producing corrupted geometry and temporal inconsistencies. We present InpaintHuman, a novel method for generating high-fidelity, complete, and animatable avatars from occluded monocular videos. Our approach introduces two key innovations: (i) a multi-scale UV-parameterized representation with hierarchical coarse-to-fine feature interpolation, enabling robust reconstruction of occluded regions while preserving geometric details; and (ii) an identity-preserving diffusion inpainting module that integrates textual inversion with semantic-conditioned guidance for subject-specific, temporally coherent completion. Unlike SDS-based methods, our approach employs direct pixel-level supervision to ensure identity fidelity. Experiments on synthetic benchmarks (PeopleSnapshot, ZJU-MoCap) and real-world scenarios (OcMotion) demonstrate competitive performance with consistent improvements in reconstruction quality across diverse poses and viewpoints.

cs.CV↗

arXiv2026

JavisGPT: A Unified Multi-modal LLM for Sounding-Video Comprehension and Generation

This paper presents JavisGPT, the first unified multimodal large language model (MLLM) for joint audio-video (JAV) comprehension and generation. JavisGPT has a concise encoder-LLM-decoder architecture, which has a SyncFusion module for spatio-temporal audio-video fusion and synchrony-aware learnable queries to bridge a pretrained JAV-DiT generator. This design enables temporally coherent video-audio understanding and generation from multimodal instructions. We design an effective three-stage training pipeline consisting of multimodal pretraining, audio-video fine-tuning, and large-scale instruction-tuning, to progressively build multimodal comprehension and generation from existing vision-language models. For instruction tuning, we construct JavisInst-Omni, a high-quality instruction dataset with over 200K GPT-4o-curated audio-video-text dialogues that cover diverse and multi-level comprehension and generation scenarios. On JAV comprehension and generation benchmarks, our experiments show that JavisGPT outperforms existing MLLMs, particularly in complex and temporally synchronized settings.

cs.CV↗

arXiv2025

What matters for Representation Alignment: Global Information or Spatial Structure?

Representation alignment (REPA) guides generative training by distilling representations from a strong, pretrained vision encoder to intermediate diffusion features. We investigate a fundamental question: what aspect of the target representation matters for generation, its \textit{global} \revision{semantic} information (e.g., measured by ImageNet-1K accuracy) or its spatial structure (i.e. pairwise cosine similarity between patch tokens)? Prevalent wisdom holds that stronger global semantic performance leads to better generation as a target representation. To study this, we first perform a large-scale empirical analysis across 27 different vision encoders and different model scales. The results are surprising; spatial structure, rather than global performance, drives the generation performance of a target representation. To further study this, we introduce two straightforward modifications, which specifically accentuate the transfer of \emph{spatial} information. We replace the standard MLP projection layer in REPA with a simple convolution layer and introduce a spatial normalization layer for the external representation. Surprisingly, our simple method (implemented in $<$4 lines of code), termed iREPA, consistently improves convergence speed of REPA, across a diverse set of vision encoders, model sizes, and training variants (such as REPA, REPA-E, Meanflow, JiT etc). %, etc. Our work motivates revisiting the fundamental working mechanism of representational alignment and how it can be leveraged for improved training of generative models. The code and project page are available at https://end2end-diffusion.github.io/irepa

cs.CV↗

arXiv2025

Investigating jet induced identified hadron productions from the relative transverse activity classifier in pp collision at LHC

In this work, we systematically investigate the jet associated identified hadron productions of pions, kaons, and protons based on the event topological separation method in proton proton (pp) collisions at $\sqrt{s}=13$ TeV employing the AMPT model with PYTHIA8 initial conditions. Enabling relative transverse event activity classifier $R_T$, we analyze the transverse momentum ($p_T$) spectra and particle ratios in the jet aligned toward region and the underlying event (UE) contributions sensitive transverse region varying with $R_T$. The results indicate that the AMPT model, incorporating both partonic and hadronic final-state interactions, provides a satisfactory description to the experimental data of the $p_T$ differential yields, particle ratios and average transverse momentum of identified particles in both the toward and transverse regions across different event activity classes. By subtracting the UE contribution from the toward region using the transverse region hadron yield, we introduce the in-jet hadron productions to analyze the modifications to jet itself. We find that the in-jet baryon to meson ratios reveal a sensitive dependence on $R_T$ driven by final state interactions, leading to a unique crossing pattern at intermediate $p_T$. This behavior can be regarded as a novel signature for probing jet medium interactions and energy loss effects in small collision systems.

hep-ph↗

arXiv2025

SimFlow: Simplified and End-to-End Training of Latent Normalizing Flows

Normalizing Flows (NFs) learn invertible mappings between the data and a Gaussian distribution. Prior works usually suffer from two limitations. First, they add random noise to training samples or VAE latents as data augmentation, introducing complex pipelines including extra noising and denoising steps. Second, they use a pretrained and frozen VAE encoder, resulting in suboptimal reconstruction and generation quality. In this paper, we find that the two issues can be solved in a very simple way: just fixing the variance (which would otherwise be predicted by the VAE encoder) to a constant (e.g., 0.5). On the one hand, this method allows the encoder to output a broader distribution of tokens and the decoder to learn to reconstruct clean images from the augmented token distribution, avoiding additional noise or denoising design. On the other hand, fixed variance simplifies the VAE evidence lower bound, making it stable to train an NF with a VAE jointly. On the ImageNet $256 \times 256$ generation task, our model SimFlow obtains a gFID score of 2.15, outperforming the state-of-the-art method STARFlow (gFID 2.40). Moreover, SimFlow can be seamlessly integrated with the end-to-end representation alignment (REPA-E) method and achieves an improved gFID of 1.91, setting a new state of the art among NFs.

cs.CV↗

arXiv2025

Extracting the kinetic freeze-out properties of high energy pp collisions at the LHC with event shape classifiers

Event shape measurements are crucial for understanding the underlying event and multiple-parton interactions (MPIs) in high energy proton-proton (pp) collisions. In this paper, the Tsallis Blast-Wave model with independent non-extensive parameters for mesons and baryons, was applied to analyze transverse momentum spectra of charged pions, kaons, and protons in pp collision events at $\sqrt{s}=13$ TeV classified by event shape estimators relative transverse event activity, unweighted transverse spherocity, and flattenicity. Our analysis reveals consistent trends in the kinetic freeze-out temperature and non-extensive parameter across different collision systems and event shape classes. The use of diverse event-shape observables in pp collisions has significantly expanded the accessible freeze-out parameter space, allowing for a more comprehensive exploration of its boundaries. Among these event shape classifiers, flattenicity emerges as a unique observable for disentangling hard process contributions from additive MPI effects, allowing the isolation of collective motion effects encoded by the radial flow velocity. Through the analysis of the interplay between event-shape measurements and kinetic freeze-out properties, we gain deeper insights into the mechanisms responsible for flow-like signatures in pp collisions.

hep-ph↗

arXiv2025

REPA-E: Unlocking VAE for End-to-End Tuning with Latent Diffusion Transformers

In this paper we tackle a fundamental question: "Can we train latent diffusion models together with the variational auto-encoder (VAE) tokenizer in an end-to-end manner?" Traditional deep-learning wisdom dictates that end-to-end training is often preferable when possible. However, for latent diffusion transformers, it is observed that end-to-end training both VAE and diffusion-model using standard diffusion-loss is ineffective, even causing a degradation in final performance. We show that while diffusion loss is ineffective, end-to-end training can be unlocked through the representation-alignment (REPA) loss -- allowing both VAE and diffusion model to be jointly tuned during the training process. Despite its simplicity, the proposed training recipe (REPA-E) shows remarkable performance; speeding up diffusion model training by over 17x and 45x over REPA and vanilla training recipes, respectively. Interestingly, we observe that end-to-end tuning with REPA-E also improves the VAE itself; leading to improved latent space structure and downstream generation performance. In terms of final performance, our approach sets a new state-of-the-art; achieving FID of 1.12 and 1.69 with and without classifier-free guidance on ImageNet 256 x 256. Code is available at https://end2end-diffusion.github.io.

cs.CV↗

arXiv2025

Confidence and Dispersity as Signals: Unsupervised Model Evaluation and Ranking

Assessing model generalization under distribution shift is essential for real-world deployment, particularly when labeled test data is unavailable. This paper presents a unified and practical framework for unsupervised model evaluation and ranking in two common deployment settings: (1) estimating the accuracy of a fixed model on multiple unlabeled test sets (dataset-centric evaluation), and (2) ranking a set of candidate models on a single unlabeled test set (model-centric evaluation). We demonstrate that two intrinsic properties of model predictions, namely confidence (which reflects prediction certainty) and dispersity (which captures the diversity of predicted classes), together provide strong and complementary signals for generalization. We systematically benchmark a set of confidence-based, dispersity-based, and hybrid metrics across a wide range of model architectures, datasets, and distribution shift types. Our results show that hybrid metrics consistently outperform single-aspect metrics on both dataset-centric and model-centric evaluation settings. In particular, the nuclear norm of the prediction matrix provides robust and accurate performance across tasks, including real-world datasets, and maintains reliability under moderate class imbalance. These findings offer a practical and generalizable basis for unsupervised model assessment in deployment scenarios.

cs.LG↗

arXiv2025

Assessing background effects in search of the chiral vortical effect in relativistic heavy-ion collisions

The search for the Chiral Vortical Effect (CVE) in relativistic heavy-ion collisions is carried out by measuring azimuthal correlators for baryon pairs such as $Λ$ and protons. Experimental results from the ALICE collaboration show significant separations in these observables, however, the interpretation remains unclear. It is believed that background contributions from baryon production mechanisms may play an important role. Using three phenomenological models, the Blast Wave, AMPT, and AVFD+UrQMD, we systematically investigate the background effects in Pb--Pb collisions at \snn = 5.02 TeV. We demonstrate that local baryon conservation, as well as hadronic annihilation processes, can significantly influence the correlators. The feed-down contribution from secondary protons is also estimated. Our study provides a foundation for disentangling background mechanisms and further facilitates the search for the CVE.

nucl-th↗

arXiv2025

Prototype-Aware Multimodal Alignment for Open-Vocabulary Visual Grounding

Visual Grounding (VG) aims to utilize given natural language queries to locate specific target objects within images. While current transformer-based approaches demonstrate strong localization performance in standard scene (i.e, scenarios without any novel objects), they exhibit notable limitations in open-vocabulary scene (i.e, both familiar and novel object categories during testing). These limitations primarily stem from three key factors: (1) imperfect alignment between visual and linguistic modalities, (2) insufficient cross-modal feature fusion, and (3) ineffective utilization of semantic prototype information. To overcome these challenges, we present Prototype-Aware Multimodal Learning (PAML), an innovative framework that systematically addresses these issues through several key components: First, we leverage ALBEF to establish robust cross-modal alignment during initial feature encoding. Subsequently, our Visual Discriminative Feature Encoder selectively enhances salient object representations while suppressing irrelevant visual context. The framework then incorporates a novel prototype discovering and inheriting mechanism that extracts and aggregates multi-neighbor semantic prototypes to facilitate open-vocabulary recognition. These enriched features undergo comprehensive multimodal integration through our Multi-stage Decoder before final bounding box regression. Extensive experiments across five benchmark datasets validate our approach, showing competitive performance in standard scene while achieving state-of-the-art results in open-vocabulary scene. Our code is available at https://github.com/plankXie/PAML.

cs.CV↗

arXiv2025

Effective Training Data Synthesis for Improving MLLM Chart Understanding

Being able to effectively read scientific plots, or chart understanding, is a central part toward building effective agents for science. However, existing multimodal large language models (MLLMs), especially open-source ones, are still falling behind with a typical success rate of 30%-50% on challenging benchmarks. Previous studies on fine-tuning MLLMs with synthetic charts are often restricted by their inadequate similarity to the real charts, which could compromise model training and performance on complex real-world charts. In this study, we show that modularizing chart generation and diversifying visual details improves chart understanding capabilities. In particular, we design a five-step data synthesis pipeline, where we separate data and function creation for single plot generation, condition the generation of later subplots on earlier ones for multi-subplot figures, visually diversify the generated figures, filter out low quality data, and finally generate the question-answer (QA) pairs with GPT-4o. This approach allows us to streamline the generation of fine-tuning datasets and introduce the effective chart dataset (ECD), which contains 10k+ chart images and 300k+ QA pairs, covering 25 topics and featuring 250+ chart type combinations with high visual complexity. We show that ECD consistently improves the performance of various MLLMs on a range of real-world and synthetic test sets. Code, data and models are available at: https://github.com/yuweiyang-anu/ECD.

cs.CV↗

arXiv2025

Vec2Face+ for Face Dataset Generation

When synthesizing identities as face recognition training data, it is generally believed that large inter-class separability and intra-class attribute variation are essential for synthesizing a quality dataset. % This belief is generally correct, and this is what we aim for. However, when increasing intra-class variation, existing methods overlook the necessity of maintaining intra-class identity consistency. % To address this and generate high-quality face training data, we propose Vec2Face+, a generative model that creates images directly from image features and allows for continuous and easy control of face identities and attributes. Using Vec2Face+, we obtain datasets with proper inter-class separability and intra-class variation and identity consistency using three strategies: 1) we sample vectors sufficiently different from others to generate well-separated identities; 2) we propose an AttrOP algorithm for increasing general attribute variations; 3) we propose LoRA-based pose control for generating images with profile head poses, which is more efficient and identity-preserving than AttrOP. % Our system generates VFace10K, a synthetic face dataset with 10K identities, which allows an FR model to achieve state-of-the-art accuracy on seven real-world test sets. Scaling the size to 4M and 12M images, the corresponding VFace100K and VFace300K datasets yield higher accuracy than the real-world training dataset, CASIA-WebFace, on five real-world test sets. This is the first time a synthetic dataset beats the CASIA-WebFace in average accuracy. In addition, we find that only 1 out of 11 synthetic datasets outperforms random guessing (\emph{i.e., 50\%}) in twin verification and that models trained with synthetic identities are more biased than those trained with real identities. Both are important aspects for future investigation. Code is available at https://github.com/HaiyuWu/Vec2Face_plus

cs.CV↗

arXiv2025

Can We Predict Performance of Large Models across Vision-Language Tasks?

Evaluating large vision-language models (LVLMs) is very expensive, due to high computational cost and the wide variety of tasks. The good news is that if we already have some observed performance scores, we may be able to infer unknown ones. In this study, we propose a new framework for predicting unknown performance scores based on observed ones from other LVLMs or tasks. We first formulate the performance prediction as a matrix completion task. Specifically, we construct a sparse performance matrix $\boldsymbol{R}$, where each entry $R_{mn}$ represents the performance score of the $m$-th model on the $n$-th dataset. By applying probabilistic matrix factorization (PMF) with Markov chain Monte Carlo (MCMC), we can complete the performance matrix, i.e., predict unknown scores. Additionally, we estimate the uncertainty of performance prediction based on MCMC. Practitioners can evaluate their models on untested tasks with higher uncertainty first, which quickly reduces the prediction errors. We further introduce several improvements to enhance PMF for scenarios with sparse observed performance scores. Our experiments demonstrate the accuracy of PMF in predicting unknown scores, the reliability of uncertainty estimates in ordering evaluations, and the effectiveness of our enhancements for handling sparse data.

cs.CV↗