arXiv subjects

Ziyan Chen

Publications and source records attributed to Ziyan Chen.

At least 37 records · Page 2Linked to original sources

arXiv2025

SimpleGVR: A Simple Baseline for Latent-Cascaded Video Super-Resolution

Latent diffusion models have emerged as a leading paradigm for efficient video generation. However, as user expectations shift toward higher-resolution outputs, relying solely on latent computation becomes inadequate. A promising approach involves decoupling the process into two stages: semantic content generation and detail synthesis. The former employs a computationally intensive base model at lower resolutions, while the latter leverages a lightweight cascaded video super-resolution (VSR) model to achieve high-resolution output. In this work, we focus on studying key design principles for latter cascaded VSR models, which are underexplored currently. First, we propose two degradation strategies to generate training pairs that better mimic the output characteristics of the base model, ensuring alignment between the VSR model and its upstream generator. Second, we provide critical insights into VSR model behavior through systematic analysis of (1) timestep sampling strategies, (2) noise augmentation effects on low-resolution (LR) inputs. These findings directly inform our architectural and training innovations. Finally, we introduce interleaving temporal unit and sparse local attention to achieve efficient training and inference, drastically reducing computational overhead. Extensive experiments demonstrate the superiority of our framework over existing methods, with ablation studies confirming the efficacy of each design choice. Our work establishes a simple yet effective baseline for cascaded video super-resolution generation, offering practical insights to guide future advancements in efficient cascaded synthesis systems.

cs.CV↗

arXiv2025

Imitate Optimal Policy: Prevail and Induce Action Collapse in Policy Gradient

Policy gradient (PG) methods in reinforcement learning frequently utilize deep neural networks (DNNs) to learn a shared backbone of feature representations used to compute likelihoods in an action selection layer. Numerous studies have been conducted on the convergence and global optima of policy networks, but few have analyzed representational structures of those underlying networks. While training an optimal policy DNN, we observed that under certain constraints, a gentle structure resembling neural collapse, which we refer to as Action Collapse (AC), emerges. This suggests that 1) the state-action activations (i.e. last-layer features) sharing the same optimal actions collapse towards those optimal actions respective mean activations; 2) the variability of activations sharing the same optimal actions converges to zero; 3) the weights of action selection layer and the mean activations collapse to a simplex equiangular tight frame (ETF). Our early work showed those aforementioned constraints to be necessary for these observations. Since the collapsed ETF of optimal policy DNNs maximally separates the pair-wise angles of all actions in the state-action space, we naturally raise a question: can we learn an optimal policy using an ETF structure as a (fixed) target configuration in the action selection layer? Our analytical proof shows that learning activations with a fixed ETF as action selection layer naturally leads to the AC. We thus propose the Action Collapse Policy Gradient (ACPG) method, which accordingly affixes a synthetic ETF as our action selection layer. ACPG induces the policy DNN to produce such an ideal configuration in the action selection layer while remaining optimal. Our experiments across various OpenAI Gym environments demonstrate that our technique can be integrated into any discrete PG methods and lead to favorable reward improvements more quickly and robustly.

cs.LG↗

arXiv2025

Unveiling the landscape of Mottness and its proximity to superconductivity in 4Hb-TaS$_2$

Mott physics is at the root of a plethora of many-body quantum phenomena in quantum materials. Recently, the stacked or twisted structures of van der Waals (vdW) materials have emerged as a unique platform for realizing exotic correlated states in the vicinity of the Mott transition. However, the definitive feature of Mottness and how it rules the low-energy electronic state remain elusive and experimentally inaccessible in many interesting regimes. Here, we quantitatively describe a filling-controlled Mott state and its interplay with superconductivity by scanning tunnelling spectroscopy in a vdW bulk heterostructure, 4Hb-TaS$_2$, that interleaves strongly correlated 1T-TaS$_2$ layers with superconducting 1H-Ta$_2$ layers. The fine tunability of electron doping induced by interlayer charge transfer allows us to continuously track the spectral function with unsurpassed energy resolution from a depleted narrow band (0.2 electrons per site) toward a Mott transition at half filling. The gradually emerging Mott-Hubbard bands, followed by the sharpening and vanishing of the central quasiparticle peak as predicted in the Brinkman-Rice scenario, unambiguously demonstrate the Mott physics at play. Importantly, the renormalization of the low-energy electrons acts destructively on the superconducting pairing potential, leaving behind nonsuperconducting, paramagnetic puddles at the nanoscale. Our results reveal a seminal system near the border of the Mott criterion that enables us to illustrate the predictive power of the Hubbard model, and set such heterostructures as promising ground for realizing new correlated states in the heavily doped Mott regime.

cond-mat.supr-con↗

arXiv2025

MAGE: Multimodal Alignment and Generation Enhancement via Bridging Visual and Semantic Spaces

In the latest advancements in multimodal learning, effectively addressing the spatial and semantic losses of visual data after encoding remains a critical challenge. This is because the performance of large multimodal models is positively correlated with the coupling between visual encoders and large language models. Existing approaches often face issues such as vector gaps or semantic disparities, resulting in information loss during the propagation process. To address these issues, we propose MAGE (Multimodal Alignment and Generation Enhancement), a novel framework that bridges the semantic spaces of vision and text through an innovative alignment mechanism. By introducing the Intelligent Alignment Network (IAN), MAGE achieves dimensional and semantic alignment. To reduce the gap between synonymous heterogeneous data, we employ a training strategy that combines cross-entropy and mean squared error, significantly enhancing the alignment effect. Moreover, to enhance MAGE's "Any-to-Any" capability, we developed a fine-tuning dataset for multimodal tool-calling instructions to expand the model's output capability boundaries. Finally, our proposed multimodal large model architecture, MAGE, achieved significantly better performance compared to similar works across various evaluation benchmarks, including MME, MMBench, and SEED. Complete code and appendix are available at: https://github.com/GTCOM-NLP/MAGE.

cs.CV↗

arXiv2025

TurboFill: Adapting Few-step Text-to-image Model for Fast Image Inpainting

This paper introduces TurboFill, a fast image inpainting model that enhances a few-step text-to-image diffusion model with an inpainting adapter for high-quality and efficient inpainting. While standard diffusion models generate high-quality results, they incur high computational costs. We overcome this by training an inpainting adapter on a few-step distilled text-to-image model, DMD2, using a novel 3-step adversarial training scheme to ensure realistic, structurally consistent, and visually harmonious inpainted regions. To evaluate TurboFill, we propose two benchmarks: DilationBench, which tests performance across mask sizes, and HumanBench, based on human feedback for complex prompts. Experiments show that TurboFill outperforms both multi-step BrushNet and few-step inpainting methods, setting a new benchmark for high-performance inpainting tasks. Our project page: https://liangbinxie.github.io/projects/TurboFill/

cs.CV↗

arXiv2025

DiffVSR: Revealing an Effective Recipe for Taming Robust Video Super-Resolution Against Complex Degradations

Diffusion models have demonstrated exceptional capabilities in image restoration, yet their application to video super-resolution (VSR) faces significant challenges in balancing fidelity with temporal consistency. Our evaluation reveals a critical gap: existing approaches consistently fail on severely degraded videos--precisely where diffusion models' generative capabilities are most needed. We identify that existing diffusion-based VSR methods struggle primarily because they face an overwhelming learning burden: simultaneously modeling complex degradation distributions, content representations, and temporal relationships with limited high-quality training data. To address this fundamental challenge, we present DiffVSR, featuring a Progressive Learning Strategy (PLS) that systematically decomposes this learning burden through staged training, enabling superior performance on complex degradations. Our framework additionally incorporates an Interweaved Latent Transition (ILT) technique that maintains competitive temporal consistency without additional training overhead. Experiments demonstrate that our approach excels in scenarios where competing methods struggle, particularly on severely degraded videos. Our work reveals that addressing the learning strategy, rather than focusing solely on architectural complexity, is the critical path toward robust real-world video super-resolution with diffusion models.

cs.CV↗

arXiv2025

Hyperspectral image reconstruction by deep learning with super-Rayleigh speckles

Ghost imaging via sparsity constraints (GISC) spectral camera modulates the three-dimensional (3D) hyperspectral image into a two-dimensional (2D) compressive image with speckles in a single shot. It obtains a 3D hyperspectral image (HSI) by reconstruction algorithms. The rapid development of deep learning has provided a new method for 3D HSI reconstruction. Moreover, the imaging performance of the GISC spectral camera can be improved by optimizing the speckle modulation. In this paper, we propose an end-to-end GISCnet with super-Rayleigh speckle modulation to improve the imaging quality of the GISC spectral camera. The structure of GISCnet is very simple but effective, and we can easily adjust the network structure parameters to improve the image reconstruction quality. Relative to Rayleigh speckles, our super-Rayleigh speckles modulation exhibits a wealth of detail in reconstructing 3D HSIs. After evaluating 648 3D HSIs, it was found that the average peak signal-to-noise ratio increased from 27 dB to 31 dB. Overall, the proposed GISCnet with super-Rayleigh speckle modulation can effectively improve the imaging quality of the GISC spectral camera by taking advantage of both optimized super-Rayleigh modulation and deep-learning image reconstruction, inspiring joint optimization of light-field modulation and image reconstruction to improve ghost imaging performance.

eess.IV↗

arXiv2025

Automatically Detecting Heterogeneous Bugs in High-Performance Computing Scientific Software

Scientific advancements rely on high-performance computing (HPC) applications that model real-world phenomena through simulations. These applications process vast amounts of data on specialized accelerators (eg., GPUs) using special libraries. Heterogeneous bugs occur in these applications when managing data movement across different platforms, such as CPUs and GPUs, leading to divergent behavior when using heterogeneous platforms compared to using only CPUs. Existing software testing techniques often fail to detect such bugs because either they do not account for platform-specific characteristics or target specific platforms. To address this problem, we present HeteroBugDetect, an automated approach to detect platform-dependent heterogeneous bugs in HPC scientific applications. HeteroBugDetect combines natural-language processing, off-target testing, custom fuzzing, and differential testing to provide an end-to-end solution for detecting platform-specific bugs in scientific applications. We evaluate HeteroBugDetect on LAMMPS, a molecular dynamics simulator, where it detected multiple heterogeneous bugs, enhancing its reliability across diverse HPC environments.

cs.SE↗

arXiv2024

Electronic structure and superconducting properties of LaNiO$_2$

Motivated by recent photoemission measurements on the La$_{0.8}$Sr$_{0.2}$NiO$_2$, we carry out a systematic study of the infinite-layer nickelate using both dynamical mean-field theory and density matrix embedding theory. The renormalized electronic structure and Fermi surface of correlated La$_{0.8}$Sr$_{0.2}$NiO$_2$ are studied in an effective two-band model through the dynamical mean-field calculation. We find the correlation effects reflect mainly on the Ni $d$ band, which is consistent with the experimental findings. We further study the ground state including magnetism and superconductivity through the density matrix embedding theory. Within the experimental doping range and rigid-band approximation, we show that the $d$-wave superconductivity is the lowest energy state, while the static magnetism is absent except very close to zero doping. These findings provide a new understanding of infinite-layer nickelate superconductivity.

cond-mat.str-el↗

arXiv2024

Towards Real-world Video Face Restoration: A New Benchmark

Blind face restoration (BFR) on images has significantly progressed over the last several years, while real-world video face restoration (VFR), which is more challenging for more complex face motions such as moving gaze directions and facial orientations involved, remains unsolved. Typical BFR methods are evaluated on privately synthesized datasets or self-collected real-world low-quality face images, which are limited in their coverage of real-world video frames. In this work, we introduced new real-world datasets named FOS with a taxonomy of "Full, Occluded, and Side" faces from mainly video frames to study the applicability of current methods on videos. Compared with existing test datasets, FOS datasets cover more diverse degradations and involve face samples from more complex scenarios, which helps to revisit current face restoration approaches more comprehensively. Given the established datasets, we benchmarked both the state-of-the-art BFR methods and the video super resolution (VSR) methods to comprehensively study current approaches, identifying their potential and limitations in VFR tasks. In addition, we studied the effectiveness of the commonly used image quality assessment (IQA) metrics and face IQA (FIQA) metrics by leveraging a subjective user study. With extensive experimental results and detailed analysis provided, we gained insights from the successes and failures of both current BFR and VSR methods. These results also pose challenges to current face restoration approaches, which we hope stimulate future advances in VFR research.

cs.CV↗

arXiv2024

DiffBIR: Towards Blind Image Restoration with Generative Diffusion Prior

We present DiffBIR, a general restoration pipeline that could handle different blind image restoration tasks in a unified framework. DiffBIR decouples blind image restoration problem into two stages: 1) degradation removal: removing image-independent content; 2) information regeneration: generating the lost image content. Each stage is developed independently but they work seamlessly in a cascaded manner. In the first stage, we use restoration modules to remove degradations and obtain high-fidelity restored results. For the second stage, we propose IRControlNet that leverages the generative ability of latent diffusion models to generate realistic details. Specifically, IRControlNet is trained based on specially produced condition images without distracting noisy content for stable generation performance. Moreover, we design a region-adaptive restoration guidance that can modify the denoising process during inference without model re-training, allowing users to balance realness and fidelity through a tunable guidance scale. Extensive experiments have demonstrated DiffBIR's superiority over state-of-the-art approaches for blind image super-resolution, blind face restoration and blind image denoising tasks on both synthetic and real-world datasets. The code is available at https://github.com/XPixelGroup/DiffBIR.

cs.CV↗

arXiv2023

Proximal Gradient Descent Unfolding Dense-spatial Spectral-attention Transformer for Compressive Spectral Imaging

The Coded Aperture Snapshot Spectral Compressive Imaging (CASSI) system modulates three-dimensional hyperspectral images into two-dimensional compressed images in a single exposure. Subsequently, three-dimensional hyperspectral images (HSI) can be reconstructed from the two-dimensional compressed measurements using reconstruction algorithms. Among these methods, deep unfolding techniques have demonstrated excellent performance, with RDLUF-MixS^2 achieving the best reconstruction results. However, RDLUF-MixS^2 requires extensive training time, taking approximately 14 days to train RDLUF-MixS^2-9stg on a single RTX 3090 GPU, making it computationally expensive. Furthermore, RDLUF-MixS^2 performs poorly on real data, resulting in significant artifacts in the reconstructed images. In this study, we introduce the Dense-spatial Spectral-attention Transformer (DST) into the Proximal Gradient Descent Unfolding Framework (PGDUF), creating a novel approach called Proximal Gradient Descent Unfolding Dense-spatial Spectral-attention Transformer (PGDUDST). Compared to RDLUF-MixS^2, PGDUDST not only surpasses the network reconstruction performance limit of RDLUF-MixS^2 but also achieves faster convergence. PGDUDST requires only 58% of the training time of RDLUF-MixS^2-9stg to achieve comparable reconstruction results. Additionally, PGDUDST significantly alleviates the artifact issues caused by RDLUF-MixS^2 in real experimental data, demonstrating superior performance and producing clearer reconstructed images.

eess.SP↗

arXiv2023

Discovery of a Single-Band Mott Insulator in a van der Waals Flat-Band Compound

The Mott insulator provides an excellent foundation for exploring a wide range of strongly correlated physical phenomena, such as high-temperature superconductivity, quantum spin liquid, and colossal magnetoresistance. A Mott insulator with the simplest degree of freedom is an ideal and highly desirable system for studying the fundamental physics of Mottness. In this study, we have unambiguously identified such an anticipated Mott insulator in a van der Waals layered compound Nb3Cl8. In the high-temperature phase, where interlayer coupling is negligible, density functional theory calculations for the monolayer of Nb3Cl8 suggest a half-filled flat band at the Fermi level, whereas angle-resolved photoemission spectroscopy experiments observe a large gap. This observation is perfectly reproduced by dynamical mean-field theory calculations considering strong electron correlations, indicating a correlation-driven Mott insulator state. Since this half-filled band derived from a single 2a1 orbital is isolated from all other bands, the monolayer of Nb3Cl8 is an ideal realization of the celebrated single-band Hubbard model. Upon decreasing the temperature, the bulk system undergoes a phase transition, where structural changes significantly enhance the interlayer coupling. This results in a bonding-antibonding splitting in the Hubbard bands, while the Mott gap remains dominant. Our discovery provides a simple and seminal model system for investigating Mott physics and other emerging correlated states.

cond-mat.str-el↗

arXiv2023

Phase retrieval based on shaped incoherent sources

The current ghost imaging phase reconstruction schemes require either complex optical systems, Fourier transform steps, or iterative algorithms, which may increase the difficulty of system design, cause phase retrieval error or take too much time. To address this problem, we propose a five-step phase-shifting method in which no complex optical systems, Fourier transform steps, or iterative algorithms are needed. With five designed incoherent sources, one can obtain five different corresponding ghost imaging patterns, then the phase information of the object can be calculated from those five speckle patterns. The applicability of this theoretical proposal is demonstrated via numerical simulations with two kinds of complicated objects, and the results illustrate the phase information of the complicated object can be reconstructed successfully and quantitatively.

physics.optics↗

arXiv2022

T-SEA: Transfer-based Self-Ensemble Attack on Object Detection

Compared to query-based black-box attacks, transfer-based black-box attacks do not require any information of the attacked models, which ensures their secrecy. However, most existing transfer-based approaches rely on ensembling multiple models to boost the attack transferability, which is time- and resource-intensive, not to mention the difficulty of obtaining diverse models on the same task. To address this limitation, in this work, we focus on the single-model transfer-based black-box attack on object detection, utilizing only one model to achieve a high-transferability adversarial attack on multiple black-box detectors. Specifically, we first make observations on the patch optimization process of the existing method and propose an enhanced attack framework by slightly adjusting its training strategies. Then, we analogize patch optimization with regular model optimization, proposing a series of self-ensemble approaches on the input data, the attacked model, and the adversarial patch to efficiently make use of the limited information and prevent the patch from overfitting. The experimental results show that the proposed framework can be applied with multiple classical base attack methods (e.g., PGD and MIM) to greatly improve the black-box transferability of the well-optimized patch on multiple mainstream detectors, meanwhile boosting white-box performance. Our code is available at https://github.com/VDIGPKU/T-SEA.

cs.CV↗

arXiv2022

Hyperspectral image reconstruction for spectral camera based on ghost imaging via sparsity constraints using V-DUnet

Spectral camera based on ghost imaging via sparsity constraints (GISC spectral camera) obtains three-dimensional (3D) hyperspectral information with two-dimensional (2D) compressive measurements in a single shot, which has attracted much attention in recent years. However, its imaging quality and real-time performance of reconstruction still need to be further improved. Recently, deep learning has shown great potential in improving the reconstruction quality and reconstruction speed for computational imaging. When applying deep learning into GISC spectral camera, there are several challenges need to be solved: 1) how to deal with the large amount of 3D hyperspectral data, 2) how to reduce the influence caused by the uncertainty of the random reference measurements, 3) how to improve the reconstructed image quality as far as possible. In this paper, we present an end-to-end V-DUnet for the reconstruction of 3D hyperspectral data in GISC spectral camera. To reduce the influence caused by the uncertainty of the measurement matrix and enhance the reconstructed image quality, both differential ghost imaging results and the detected measurements are sent into the network's inputs. Compared with compressive sensing algorithm, such as PICHCS and TwIST, it not only significantly improves the imaging quality with high noise immunity, but also speeds up the reconstruction time by more than two orders of magnitude.

eess.IV↗

arXiv2022

FV-UPatches: Enhancing Universality in Finger Vein Recognition

Many deep learning-based models have been introduced in finger vein recognition in recent years. These solutions, however, suffer from data dependency and are difficult to achieve model generalization. To address this problem, we are inspired by the idea of domain adaptation and propose a universal learning-based framework, which achieves generalization while training with limited data. To reduce differences between data distributions, a compressed U-Net is introduced as a domain mapper to map the raw region of interest image onto a target domain. The concentrated target domain is a unified feature space for the subsequent matching, in which a local descriptor model SOSNet is employed to embed patches into descriptors measuring the similarity of matching pairs. In the proposed framework, the domain mapper is an approximation to a specific extraction function thus the training is only a one-time effort with limited data. Moreover, the local descriptor model can be trained to be representative enough based on a public dataset of non-finger-vein images. The whole pipeline enables the framework to be well generalized, making it possible to enhance universality and helps to reduce costs of data collection, tuning and retraining. The comparable experimental results to state-of-the-art (SOTA) performance in five public datasets prove the effectiveness of the proposed framework. Furthermore, the framework shows application potential in other vein-based biometric recognition as well.

cs.CV↗

arXiv2020

SRQA: Synthetic Reader for Factoid Question Answering

The question answering system can answer questions from various fields and forms with deep neural networks, but it still lacks effective ways when facing multiple evidences. We introduce a new model called SRQA, which means Synthetic Reader for Factoid Question Answering. This model enhances the question answering system in the multi-document scenario from three aspects: model structure, optimization goal, and training method, corresponding to Multilayer Attention (MA), Cross Evidence (CE), and Adversarial Training (AT) respectively. First, we propose a multilayer attention network to obtain a better representation of the evidences. The multilayer attention mechanism conducts interaction between the question and the passage within each layer, making the token representation of evidences in each layer takes the requirement of the question into account. Second, we design a cross evidence strategy to choose the answer span within more evidences. We improve the optimization goal, considering all the answers' locations in multiple evidences as training targets, which leads the model to reason among multiple evidences. Third, adversarial training is employed to high-level variables besides the word embedding in our model. A new normalization method is also proposed for adversarial perturbations so that we can jointly add perturbations to several target variables. As an effective regularization method, adversarial training enhances the model's ability to process noisy data. Combining these three strategies, we enhance the contextual representation and locating ability of our model, which could synthetically extract the answer span from several evidences. We perform SRQA on the WebQA dataset, and experiments show that our model outperforms the state-of-the-art models (the best fuzzy score of our model is up to 78.56%, with an improvement of about 2%).

cs.CL↗