arXiv subjects

Xiao Fu

Publications and source records attributed to Xiao Fu.

At least 37 records · Page 2Linked to original sources

arXiv2025

SemanticGen: Video Generation in Semantic Space

State-of-the-art video generative models typically learn the distribution of video latents in the VAE space and map them to pixels using a VAE decoder. While this approach can generate high-quality videos, it suffers from slow convergence and is computationally expensive when generating long videos. In this paper, we introduce SemanticGen, a novel solution to address these limitations by generating videos in the semantic space. Our main insight is that, due to the inherent redundancy in videos, the generation process should begin in a compact, high-level semantic space for global planning, followed by the addition of high-frequency details, rather than directly modeling a vast set of low-level video tokens using bi-directional attention. SemanticGen adopts a two-stage generation process. In the first stage, a diffusion model generates compact semantic video features, which define the global layout of the video. In the second stage, another diffusion model generates VAE latents conditioned on these semantic features to produce the final output. We observe that generation in the semantic space leads to faster convergence compared to the VAE latent space. Our method is also effective and computationally efficient when extended to long video generation. Extensive experiments demonstrate that SemanticGen produces high-quality videos and outperforms state-of-the-art approaches and strong baselines.

cs.CV↗

arXiv2025

Rethinking Coupled Tensor Analysis for Hyperspectral Super-Resolution: Recoverable Modeling Under Endmember Variability

This work revisits the hyperspectral super-resolution (HSR) problem, i.e., fusing a pair of spatially co-registered hyperspectral (HSI) and multispectral (MSI) images to recover a super-resolution image (SRI) that enhances the spatial resolution of the HSI. Coupled tensor decomposition (CTD)-based methods have gained traction in this domain, offering recoverability guarantees under various assumptions. Existing models such as canonical polyadic decomposition (CPD) and Tucker decomposition provide strong expressive power but lack physical interpretability. The block-term decomposition model with rank-$(L_r, L_r, 1)$ terms (the LL1 model) yields interpretable factors under the linear mixture model (LMM) of spectral images, but LMM assumptions are often violated in practice -- primarily due to nonlinear effects such as endmember variability (EV). To address this, we propose modeling spectral images using a more flexible block-term tensor decomposition with rank-$(L_r, M_r, N_r)$ terms (the LMN model). This modeling choice retains interpretability, subsumes CPD, Tucker, and LL1 as special cases, and robustly accounts for non-ideal effects such as EV, offering a balanced tradeoff between expressiveness and interpretability for HSR. Importantly, under the LMN model for HSI and MSI, recoverability of the SRI can still be established under proper conditions -- providing strong theoretical support. Extensive experiments on synthetic and real datasets further validate the effectiveness and robustness of the proposed method compared with existing CTD-based approaches.

eess.IV↗

arXiv2025

Immunometabolic Gatekeeping: Reconciling Peto's & the T-cell Infiltration Prognostic Paradox

Classical models of cancer focus on tumour-intrinsic genetic aberrations and immune dynamics and often overlook how the metabolic environment of healthy tissues shapes tumour development and immune efficacy. Here, we propose that tissue-intrinsic metabolic intensity and waste-handling capacity act as an upstream gatekeeper of anti-tumour immunity, determining whether immune infiltration translates into effective immune function and safeguards the tissue from tumourigenesis. Across human cancers, tumours arising in high-metabolism tissues - like kidney, brain, and eye - tend to show high T cell infiltration but poor prognosis, suggesting pre-existing metabolic environments prior to malignant transformation may undermine immune function. This pattern is mirrored across species: large mammals with lower mass-specific metabolic rates (e.g., elephants, whales) accumulate fewer metabolic byproducts and show lower cancer incidence (Peto's paradox), while long-lived small mammals like bats and naked mole-rats resist tumourigenesis via suppressed glycolysis or altered hypoxia responses leading to lower metabolic rates and/or byproduct accumulation. Through integrative synthesis spanning human single-cell expression data and cross-species comparisons, we outline a framework of 'immunometabolic gatekeeping', where tissues with high metabolic rate and poor waste clearance foster immune-exhausting niches even before transformation. This unifying framework reconciles multiple paradoxes in cancer biology: Peto's paradox, T cell infiltration non-prognosticity, tissue tropisms, and sex-based inequalities, and suggests new principles for identifying high-risk patients and metabolic-immune combination strategies for prevention and treatment. By shifting focus from tumour-intrinsic mutations to host-tissue metabolism, this work offers a novel, integrative lens on cancer vulnerability and immune failure.

q-bio.TO↗

arXiv2025

Diversified Flow Matching with Translation Identifiability

Diversified distribution matching (DDM) finds a unified translation function mapping a diverse collection of conditional source distributions to their target counterparts. DDM was proposed to resolve content misalignment issues in unpaired domain translation, achieving translation identifiability. However, DDM has only been implemented using GANs due to its constraints on the translation function. GANs are often unstable to train and do not provide the transport trajectory information -- yet such trajectories are useful in applications such as single-cell evolution analysis and robot route planning. This work introduces diversified flow matching (DFM), an ODE-based framework for DDM. Adapting flow matching (FM) to enforce a unified translation function as in DDM is challenging, as FM learns the translation function's velocity rather than the translation function itself. A custom bilevel optimization-based training loss, a nonlinear interpolant, and a structural reformulation are proposed to address these challenges, offering a tangible implementation. To our knowledge, DFM is the first ODE-based approach guaranteeing translation identifiability. Experiments on synthetic and real-world datasets validate the proposed method.

cs.LG↗

arXiv2025

Diverse Influence Component Analysis: A Geometric Approach to Nonlinear Mixture Identifiability

Latent component identification from unknown nonlinear mixtures is a foundational challenge in machine learning, with applications in tasks such as disentangled representation learning and causal inference. Prior work in nonlinear independent component analysis (nICA) has shown that auxiliary signals -- such as weak supervision -- can support identifiability of conditionally independent latent components. More recent approaches explore structural assumptions, e.g., sparsity in the Jacobian of the mixing function, to relax such requirements. In this work, we introduce Diverse Influence Component Analysis (DICA), a framework that exploits the convex geometry of the mixing function's Jacobian. We propose a Jacobian Volume Maximization (J-VolMax) criterion, which enables latent component identification by encouraging diversity in their influence on the observed variables. Under reasonable conditions, this approach achieves identifiability without relying on auxiliary information, latent component independence, or Jacobian sparsity assumptions. These results extend the scope of identifiability analysis and offer a complementary perspective to existing methods.

cs.LG↗

arXiv2025

Downlink MIMO Channel Estimation from Bits: Recoverability and Algorithm

In frequency division duplex (FDD) massive MIMO systems, a major challenge lies in acquiring the downlink channel state information}\ (CSI) at the base station (BS) from limited feedback sent by the user equipment (UE). To tackle this fundamental task, our contribution is twofold: First, a simple feedback framework is proposed, where a compression and Gaussian dithering-based quantization strategy is adopted at the UE side, and then a maximum likelihood estimator (MLE) is formulated at the BS side. Recoverability of the MIMO channel under the widely used double directional model is established. Specifically, analyses are presented for two compression schemes -- showing one being more overhead-economical and the other computationally lighter at the UE side. Second, to realize the MLE, an alternating direction method of multipliers (ADMM) algorithm is proposed. The algorithm is carefully designed to integrate a sophisticated harmonic retrieval (HR) solver as subroutine, which turns out to be the key of effectively tackling this hard MLE problem.Extensive numerical experiments are conducted to validate the efficacy of our approach.

eess.SP↗

arXiv2025

Towards Identifiable Unsupervised Domain Translation: A Diversified Distribution Matching Approach

Unsupervised domain translation (UDT) aims to find functions that convert samples from one domain (e.g., sketches) to another domain (e.g., photos) without changing the high-level semantic meaning (also referred to as ``content''). The translation functions are often sought by probability distribution matching of the transformed source domain and target domain. CycleGAN stands as arguably the most representative approach among this line of work. However, it was noticed in the literature that CycleGAN and variants could fail to identify the desired translation functions and produce content-misaligned translations. This limitation arises due to the presence of multiple translation functions -- referred to as ``measure-preserving automorphism" (MPA) -- in the solution space of the learning criteria. Despite awareness of such identifiability issues, solutions have remained elusive. This study delves into the core identifiability inquiry and introduces an MPA elimination theory. Our analysis shows that MPA is unlikely to exist, if multiple pairs of diverse cross-domain conditional distributions are matched by the learning function. Our theory leads to a UDT learner using distribution matching over auxiliary variable-induced subsets of the domains -- other than over the entire data domains as in the classical approaches. The proposed framework is the first to rigorously establish translation identifiability under reasonable UDT settings, to our best knowledge. Experiments corroborate with our theoretical claims.

cs.LG↗

arXiv2025

Domain-Factored Untrained Deep Prior for Spectrum Cartography

Spectrum cartography (SC) focuses on estimating the radio power propagation map of multiple emitters across space and frequency using limited sensor measurements. Recent advances in SC have shown that leveraging learned deep generative models (DGMs) as structural constraints yields state-of-the-art performance. By harnessing the expressive power of neural networks, these structural "priors" capture intricate patterns in radio maps. However, training DGMs requires substantial data, which is not always available, and distribution shifts between training and testing data can further degrade performance. To address these challenges, this work proposes using untrained neural networks (UNNs) for SC. UNNs, commonly applied in vision tasks to represent complex data without training, encode structural information of data in neural architectures. In our approach, a custom-designed UNN represents radio maps under a spatio-spectral domain factorization model, leveraging physical characteristics to reduce sample complexity of SC. Experiments show that the method achieves performance comparable to learned DGM-based SC, without requiring training data.

eess.SP↗

arXiv2025

Accelerating Amorphous Alloy Discovery: Data-Driven Property Prediction via General-Purpose Machine Learning Interatomic Potential

While traditional trial-and-error methods for designing amorphous alloys are costly and inefficient, machine learning approaches based solely on composition lack critical atomic structural information. Machine learning interatomic potentials, trained on data from first-principles calculations, offer a powerful alternative by efficiently approximating the complex three-dimensional potential energy surface with near-DFT accuracy. In this work, we develop a general-purpose machine learning interatomic potential for amorphous alloys by using a dataset comprising 20400 configurations across representative binary and ternary amorphous alloys systems. The model demonstrates excellent predictive performance on an independent test set, with a mean absolute error of 5.06 meV/atom for energy and 128.51 meV/Å for forces. Through extensive validation, the model is shown to reliably capture the trends in macroscopic property variations such as density, Young's modulus and glass transition temperature across both the original training systems and the compositionally modified systems derived from them. It can be directly applied to composition-property mapping of amorphous alloys. Furthermore, the developed interatomic potential enables access to the atomic structures of amorphous alloys, allowing for microscopic analysis and interpretation of experimental results, particularly those deviating from empirical trends.This work breaks the long-standing computational bottleneck in amorphous alloys research by developing the first general-purpose machine learning interatomic potential for amorphous alloy systems. The resulting framework provides a robust foundation for data-driven design and high-throughput composition screening in a field previously constrained by traditional simulation limitations.

cond-mat.mtrl-sci↗

arXiv2025

PREF: Reference-Free Evaluation of Personalised Text Generation in LLMs

Personalised text generation is essential for user-centric information systems, yet most evaluation methods overlook the individuality of users. We introduce \textbf{PREF}, a \textbf{P}ersonalised \textbf{R}eference-free \textbf{E}valuation \textbf{F}ramework that jointly measures general output quality and user-specific alignment without requiring gold personalised references. PREF operates in a three-step pipeline: (1) a coverage stage uses a large language model (LLM) to generate a comprehensive, query-specific guideline covering universal criteria such as factuality, coherence, and completeness; (2) a preference stage re-ranks and selectively augments these factors using the target user's profile, stated or inferred preferences, and context, producing a personalised evaluation rubric; and (3) a scoring stage applies an LLM judge to rate candidate answers against this rubric, ensuring baseline adequacy while capturing subjective priorities. This separation of coverage from preference improves robustness, transparency, and reusability, and allows smaller models to approximate the personalised quality of larger ones. Experiments on the PrefEval benchmark, including implicit preference-following tasks, show that PREF achieves higher accuracy, better calibration, and closer alignment with human judgments than strong baselines. By enabling scalable, interpretable, and user-aligned evaluation, PREF lays the groundwork for more reliable assessment and development of personalised language generation systems.

cs.CL↗

arXiv2025

ReCamMaster: Camera-Controlled Generative Rendering from A Single Video

Camera control has been actively studied in text or image conditioned video generation tasks. However, altering camera trajectories of a given video remains under-explored, despite its importance in the field of video creation. It is non-trivial due to the extra constraints of maintaining multiple-frame appearance and dynamic synchronization. To address this, we present ReCamMaster, a camera-controlled generative video re-rendering framework that reproduces the dynamic scene of an input video at novel camera trajectories. The core innovation lies in harnessing the generative capabilities of pre-trained text-to-video models through a simple yet powerful video conditioning mechanism--its capability is often overlooked in current research. To overcome the scarcity of qualified training data, we construct a comprehensive multi-camera synchronized video dataset using Unreal Engine 5, which is carefully curated to follow real-world filming characteristics, covering diverse scenes and camera movements. It helps the model generalize to in-the-wild videos. Lastly, we further improve the robustness to diverse inputs through a meticulously designed training strategy. Extensive experiments show that our method substantially outperforms existing state-of-the-art approaches. Our method also finds promising applications in video stabilization, super-resolution, and outpainting. Our code and dataset are publicly available at: https://github.com/KwaiVGI/ReCamMaster.

cs.CV↗

arXiv2025

3DTrajMaster: Mastering 3D Trajectory for Multi-Entity Motion in Video Generation

This paper aims to manipulate multi-entity 3D motions in video generation. Previous methods on controllable video generation primarily leverage 2D control signals to manipulate object motions and have achieved remarkable synthesis results. However, 2D control signals are inherently limited in expressing the 3D nature of object motions. To overcome this problem, we introduce 3DTrajMaster, a robust controller that regulates multi-entity dynamics in 3D space, given user-desired 6DoF pose (location and rotation) sequences of entities. At the core of our approach is a plug-and-play 3D-motion grounded object injector that fuses multiple input entities with their respective 3D trajectories through a gated self-attention mechanism. In addition, we exploit an injector architecture to preserve the video diffusion prior, which is crucial for generalization ability. To mitigate video quality degradation, we introduce a domain adaptor during training and employ an annealed sampling strategy during inference. To address the lack of suitable training data, we construct a 360-Motion Dataset, which first correlates collected 3D human and animal assets with GPT-generated trajectory and then captures their motion with 12 evenly-surround cameras on diverse 3D UE platforms. Extensive experiments show that 3DTrajMaster sets a new state-of-the-art in both accuracy and generalization for controlling multi-entity 3D motions. Project page: http://fuxiao0719.github.io/projects/3dtrajmaster

cs.CV↗

arXiv2025

High-Throughput NEB for Li-Ion Conductor Discovery via Fine-Tuned CHGNet Potential

Solid-state electrolytes are essential in the development of all-solid-state batteries. While density functional theory (DFT)-based nudged elastic band (NEB) and ab initio molecular dynamics (AIMD) methods provide fundamental insights on lithium-ion migration barriers and ionic conductivity, their computational costs make large-scale materials exploration challenging. In this study, we developed a high-throughput NEB computational framework integrated with the fine-tuned universal machine learning interatomic potentials (uMLIPs), enabling accelerated prediction of migration barriers based on transition state theory for the efficient discovery of fast-ion conductors. This framework automates the construction of initial/final states and migration paths, mitigating the inaccurate barriers prediction in pretrained potentials due to the insufficient training data on high-energy states. We employed the fine-tuned CHGNet model into NEB/MD calculations and the dual CHGNet-NEB/MD achieves a balance between computational speed and accuracy, as validated in NASICON-type Li$_{1+x}$Al$_x$Ti$_{2-x}$(PO$_4$)$_3$ (LATP) structures. Through high-throughput screening, we identified orthorhombic Pnma-group structures (LiMgPO$_4$, LiTiPO$_5$, etc.) which can serve as promising frameworks for fast ion conductors. Their aliovalent-doped variants, Li$_{0.5}$Mg$_{0.5}$Al$_{0.5}$PO$_4$ and Li$_{0.5}$TiPO$_{4.5}$F$_{0.5}$, were predicted to possess low activation energies, as well as high ionic conductivity of 0.19 mS/cm and 0.024 mS/cm, respectively.

cond-mat.mtrl-sci↗

arXiv2025

Learning From Crowdsourced Noisy Labels: A Signal Processing Perspective

One of the primary catalysts fueling advances in artificial intelligence (AI) and machine learning (ML) is the availability of massive, curated datasets. A commonly used technique to curate such massive datasets is crowdsourcing, where data are dispatched to multiple annotators. The annotator-produced labels are then fused to serve downstream learning and inference tasks. This annotation process often creates noisy labels due to various reasons, such as the limited expertise, or unreliability of annotators, among others. Therefore, a core objective in crowdsourcing is to develop methods that effectively mitigate the negative impact of such label noise on learning tasks. This feature article introduces advances in learning from noisy crowdsourced labels. The focus is on key crowdsourcing models and their methodological treatments, from classical statistical models to recent deep learning-based approaches, emphasizing analytical insights and algorithmic developments. In particular, this article reviews the connections between signal processing (SP) theory and methods, such as identifiability of tensor and nonnegative matrix factorization, and novel, principled solutions of longstanding challenges in crowdsourcing -- showing how SP perspectives drive the advancements of this field. Furthermore, this article touches upon emerging topics that are critical for developing cutting-edge AI/ML systems, such as crowdsourcing in reinforcement learning with human feedback (RLHF) and direct preference optimization (DPO) that are key techniques for fine-tuning large language models (LLMs).

eess.SP↗

arXiv2025

Radio Map Estimation via Latent Domain Plug-and-Play Denoising

Radio map estimation (RME), also known as spectrum cartography, aims to reconstruct the strength of radio interference across different domains (e.g., space and frequency) from sparsely sampled measurements. To tackle this typical inverse problem, state-of-the-art RME methods rely on handcrafted or data-driven structural information of radio maps. However, the former often struggles to model complex radio frequency (RF) environments and the latter requires excessive training -- making it hard to quickly adapt to in situ sensing tasks. This work presents a spatio-spectral RME approach based on plug-and-play (PnP) denoising, a technique from computational imaging. The idea is to leverage the observation that the denoising operations of signals like natural images and radio maps are similar -- despite the nontrivial differences of the signals themselves. Hence, sophisticated denoisers designed for or learned from natural images can be directly employed to assist RME, avoiding using radio map data for training. Unlike conventional PnP methods that operate directly in the data domain, the proposed method exploits the underlying physical structure of radio maps and proposes an ADMM algorithm that denoises in a latent domain. This design significantly improves computational efficiency and enhances noise robustness. Theoretical aspects, e.g., recoverability of the complete radio map and convergence of the ADMM algorithm are analyzed. Synthetic and real data experiments are conducted to demonstrate the effectiveness of our approach.

eess.SP↗

arXiv2025

PanopticNeRF-360: Panoramic 3D-to-2D Label Transfer in Urban Scenes

Training perception systems for self-driving cars requires substantial 2D annotations that are labor-intensive to manual label. While existing datasets provide rich annotations on pre-recorded sequences, they fall short in labeling rarely encountered viewpoints, potentially hampering the generalization ability for perception models. In this paper, we present PanopticNeRF-360, a novel approach that combines coarse 3D annotations with noisy 2D semantic cues to generate high-quality panoptic labels and images from any viewpoint. Our key insight lies in exploiting the complementarity of 3D and 2D priors to mutually enhance geometry and semantics. Specifically, we propose to leverage coarse 3D bounding primitives and noisy 2D semantic and instance predictions to guide geometry optimization, by encouraging predicted labels to match panoptic pseudo ground truth. Simultaneously, the improved geometry assists in filtering 3D&2D annotation noise by fusing semantics in 3D space via a learned semantic field. To further enhance appearance, we combine MLP and hash grids to yield hybrid scene features, striking a balance between high-frequency appearance and contiguous semantics. Our experiments demonstrate PanopticNeRF-360's state-of-the-art performance over label transfer methods on the challenging urban scenes of the KITTI-360 dataset. Moreover, PanopticNeRF-360 enables omnidirectional rendering of high-fidelity, multi-view and spatiotemporally consistent appearance, semantic and instance labels. We make our code and data available at https://github.com/fuxiao0719/PanopticNeRF

cs.CV↗

arXiv2025

FieldFormer: Self-supervised Reconstruction of Physical Fields via Tensor Attention Prior

Reconstructing physical field tensors from \textit{in situ} observations, such as radio maps and ocean sound speed fields, is crucial for enabling environment-aware decision making in various applications, e.g., wireless communications and underwater acoustics. Field data reconstruction is often challenging, due to the limited and noisy nature of the observations, necessitating the incorporation of prior information to aid the reconstruction process. Deep neural network-based data-driven structural constraints (e.g., ``deeply learned priors'') have showed promising performance. However, this family of techniques faces challenges such as model mismatches between training and testing phases. This work introduces FieldFormer, a self-supervised neural prior learned solely from the limited {\it in situ} observations without the need of offline training. Specifically, the proposed framework starts with modeling the fields of interest using the tensor Tucker model of a high multilinear rank, which ensures a universal approximation property for all fields. In the sequel, an attention mechanism is incorporated to learn the sparsity pattern that underlies the core tensor in order to reduce the solution space. In this way, a ``complexity-adaptive'' neural representation, grounded in the Tucker decomposition, is obtained that can flexibly represent various types of fields. A theoretical analysis is provided to support the recoverability of the proposed design. Moreover, extensive experiments, using various physical field tensors, demonstrate the superiority of the proposed approach compared to state-of-the-art baselines.

eess.SP↗

arXiv2025

High-Entropy Solid Electrolytes Discovery: A Dual-Stage Machine Learning Framework Bridging Atomic Configurations and Ionic Transport Properties

The rapid development of computational materials science powered by machine learning (ML) is gradually leading to solutions to several previously intractable scientific problems. One of the most prominent is machine learning interatomic potentials (MLIPs), which expedites the study of dynamical methods for large-scale systems. However, a promising field, high-entropy (HE) solid-state electrolytes (SEs) remain constrained by trial-and-error paradigms, lacking systematic computational strategies to address their huge and high-dimensional composition space. In this work, we establish a dual-stage ML framework that combines fine-tuned MLIPs with interpretable feature-property mapping to accelerate the high-entropy SEs discovery. Using Li$_3$Zr$_2$Si$_2$PO$_{12}$ (LZSP) as a prototype, the fine-tuned CHGNet-based relaxation provides atomic structure for each configuration, the structure features - mean squared displacement (SF-MSD) model predicts the ionic transport properties and identifies critical descriptors. The theoretical studies indicate that the framework can satisfy the multiple requirements including computational efficiency, generalization reliability and prediction accuracy. One of the most promising element combinations in the quinary HE-LZSP space containing 4575 compositions is identified with a high ionic conductivity of 4.53 mS/cm as an application example. The framework contains generalizability and extensibility to other SE families.

cond-mat.mtrl-sci↗