arXiv subjects

Feng Liu

Publications and source records attributed to Feng Liu.

At least 253 records · Page 14Linked to original sources

arXiv2025

Progressive Growing of Video Tokenizers for Temporally Compact Latent Spaces

Video tokenizers are essential for latent video diffusion models, converting raw video data into spatiotemporally compressed latent spaces for efficient training. However, extending state-of-the-art video tokenizers to achieve a temporal compression ratio beyond 4x without increasing channel capacity poses significant challenges. In this work, we propose an alternative approach to enhance temporal compression. We find that the reconstruction quality of temporally subsampled videos from a low-compression encoder surpasses that of high-compression encoders applied to original videos. This indicates that high-compression models can leverage representations from lower-compression models. Building on this insight, we develop a bootstrapped high-temporal-compression model that progressively trains high-compression blocks atop well-trained lower-compression models. Our method includes a cross-level feature-mixing module to retain information from the pretrained low-compression model and guide higher-compression blocks to capture the remaining details from the full video sequence. Evaluation of video benchmarks shows that our method significantly improves reconstruction quality while increasing temporal compression compared to directly training the full model. Furthermore, the resulting compact latent space effectively trains a video diffusion model for high-quality video generation with a significantly reduced token budget.

cs.CV↗

arXiv2025

Spatial-Temporal-Spectral Unified Modeling for Remote Sensing Dense Prediction

The proliferation of multi-source remote sensing data has propelled the development of deep learning for dense prediction, yet significant challenges in data and task unification persist. Current deep learning architectures for remote sensing are fundamentally rigid. They are engineered for fixed input-output configurations, restricting their adaptability to the heterogeneous spatial, temporal, and spectral dimensions inherent in real-world data. Furthermore, these models neglect the intrinsic correlations among semantic segmentation, binary change detection, and semantic change detection, necessitating the development of distinct models or task-specific decoders. This paradigm is also constrained to a predefined set of output semantic classes, where any change to the classes requires costly retraining. To overcome these limitations, we introduce the Spatial-Temporal-Spectral Unified Network (STSUN) for unified modeling. STSUN can adapt to input and output data with arbitrary spatial sizes, temporal lengths, and spectral bands by leveraging their metadata for a unified representation. Moreover, STSUN unifies disparate dense prediction tasks within a single architecture by conditioning the model on trainable task embeddings. Similarly, STSUN facilitates flexible prediction across multiple set of semantic categories by integrating trainable category embeddings as metadata. Extensive experiments on multiple datasets with diverse Spatial-Temporal-Spectral configurations in multiple scenarios demonstrate that a single STSUN model effectively adapts to heterogeneous inputs and outputs, unifying various dense prediction tasks and diverse semantic class predictions. The proposed approach consistently achieves state-of-the-art performance, highlighting its robustness and generalizability for complex remote sensing applications.

cs.CV↗

arXiv2025

Phase Competition and Rutile Phase Stabilization of Growing GeO2 Films by MOCVD

Rutile germanium dioxide (r-GeO2) is an ultra-wide bandgap semiconductor with potential for ambipolar doping, making it a promising candidate for next-generation power electronics and optoelectronics. Growth of phase-pure r-GeO2 films by vapor phase techniques like metalorganic chemical vapor deposition (MOCVD) is challenging because of polymorphic competition from amorphous and quartz GeO2. Here, we introduce seed-driven stepwise crystallization (SDSC) as a segmented growth strategy for obtaining r-GeO2 films on r-TiO2 (001) substrate. SDSC divides the growth into repeated cycles of film deposition and cooling-heating ramps, which suppress the non-rutile phases. We discuss the underlying mechanisms of phase selection during SDSC growth. We demonstrate continuous, phase-pure, partially epitaxial r-GeO2 (001) films exhibiting x-ray rocking curves with a FWHM of 597 arcsec. SDSC-based growth provides a generalizable pathway for selective vapor-phase growth of metastable or unstable phases, offering new opportunities for phase-selective thin-film engineering.

cond-mat.mtrl-sci↗

arXiv2025

Ordered phases and superconductivity in two-dimensional electron systems subject to pair spin-orbit interaction

Pair spin-orbit interaction can emerge in strongly-interacting systems characterized by a large spin-orbit coupling. Here we study the role of this interaction in stabilizing ordered and unconventional superconducting phases. We find that, if the system avoids superconductivity, the order realized is a combination of charge-density and spin-vorticity waves. The latter is reminiscent of a loop-current state, albeit in the spin, rather in the charge, channel. If the system becomes superconducting, the order parameter assumes the form of a paired density wave, i.e. pairing occurs at finite momentum. Intriguingly, one of the possible pairings acquires a form analogous to Amperean superconductivity. However, the order parameter here is always a blend of paired density wave and Amperean pairing, rather than being purely one or the other.

cond-mat.supr-con↗

arXiv2025

An Empirical Study on Virtual Reality Software Security Weaknesses

Virtual Reality (VR) has emerged as a transformative technology across industries, yet its security weaknesses, including vulnerabilities, are underinvestigated. This study investigates 334 VR projects hosted on GitHub, examining 1,681 software security weaknesses to understand: what types of weaknesses are prevalent in VR software; when and how weaknesses are introduced; how long they have survived; and how they have been removed. Due to the limited availability of VR software security weaknesses in public databases (e.g., the National Vulnerability Database or NVD), we prepare the first systematic dataset of VR software security weaknesses by introducing a novel framework to collect such weaknesses from GitHub commit data. Our empirical study on the dataset leads to useful insights, including: (i) VR weaknesses are heavily skewed toward user interface weaknesses, followed by resource-related weaknesses; (ii) VR development tools pose higher security risks than VR applications; (iii) VR security weaknesses are often introduced at the VR software birth time.

cs.CR↗

arXiv2025

Action Unit Enhance Dynamic Facial Expression Recognition

Dynamic Facial Expression Recognition(DFER) is a rapidly evolving field of research that focuses on the recognition of time-series facial expressions. While previous research on DFER has concentrated on feature learning from a deep learning perspective, we put forward an AU-enhanced Dynamic Facial Expression Recognition architecture, namely AU-DFER, that incorporates AU-expression knowledge to enhance the effectiveness of deep learning modeling. In particular, the contribution of the Action Units(AUs) to different expressions is quantified, and a weight matrix is designed to incorporate a priori knowledge. Subsequently, the knowledge is integrated with the learning outcomes of a conventional deep learning network through the introduction of AU loss. The design is incorporated into the existing optimal model for dynamic expression recognition for the purpose of validation. Experiments are conducted on three recent mainstream open-source approaches to DFER on the principal datasets in this field. The results demonstrate that the proposed architecture outperforms the state-of-the-art(SOTA) methods without the need for additional arithmetic and generally produces improved results. Furthermore, we investigate the potential of AU loss function redesign to address data label imbalance issues in established dynamic expression datasets. To the best of our knowledge, this is the first attempt to integrate quantified AU-expression knowledge into various DFER models. We also devise strategies to tackle label imbalance, or minor class problems. Our findings suggest that employing a diverse strategy of loss function design can enhance the effectiveness of DFER. This underscores the criticality of addressing data imbalance challenges in mainstream datasets within this domain. The source code is available at https://github.com/Cross-Innovation-Lab/AU-DFER.

cs.CV↗

arXiv2025

SeisMoLLM: Advancing Seismic Monitoring via Cross-modal Transfer with Pre-trained Large Language Model

Recent advances in deep learning have revolutionized seismic monitoring, yet developing a foundation model that performs well across multiple complex tasks remains challenging, particularly when dealing with degraded signals or data scarcity. This work presents SeisMoLLM, the first foundation model that utilizes cross-modal transfer for seismic monitoring, to unleash the power of large-scale pre-training from a large language model without requiring direct pre-training on seismic datasets. Through elaborate waveform tokenization and fine-tuning of pre-trained GPT-2 model, SeisMoLLM achieves state-of-the-art performance on the DiTing and STEAD datasets across five critical tasks: back-azimuth estimation, epicentral distance estimation, magnitude estimation, phase picking, and first-motion polarity classification. It attains 36 best results out of 43 task metrics and 12 top scores out of 16 few-shot generalization metrics, with many relative improvements ranging from 10% to 50%. In addition to its superior performance, SeisMoLLM maintains efficiency comparable to or even better than lightweight models in both training and inference. These findings establish SeisMoLLM as a promising foundation model for practical seismic monitoring and highlight cross-modal transfer as an exciting new direction for earthquake studies, showcasing the potential of advanced deep learning techniques to propel seismology research forward.

cs.LG↗

arXiv2025

On Decision-Dependent Uncertainties in Power Systems with High-Share Renewables

The continuously increasing renewable energy sources (RES) and demand response (DR) are becoming important sources of system flexibility. As a consequence, decision-dependent uncertainties (DDUs), interchangeably referred to as endogenous uncertainties, impose new characteristics to power system dispatch. The DDUs faced by system operators originate from uncertain dispatchable resources such as RES units or DR, while reserve providers encounter DDUs arising from the uncertain reserve deployment. This paper presents a systematic framework for addressing robust dispatch problems with DDUs. The main contributions include i) the robust characterization of DDUs with a dependency decomposition structure; ii) a generic DDU coping mechanism, manifested as the bilateral matching between uncertainty and flexibility; iii) analyses of the influence of DDU incorporation on the convexity/non-convexity of robust dispatch problems; and iv) generic solution algorithms adaptive for DDUs. Under this framework, the inherent distinctions and correlations between DDUs and DIUs are revealed, providing a fundamental theoretical basis for the economic and reliable operation of RES-dominated power systems. Applications in the source and demand sides illustrate the importance of considering DDUs and verify the effectiveness of proposed algorithms for robust dispatch with DDUs.

eess.SY↗

arXiv2025

Adaptive Gate-Aware Mamba Networks for Magnetic Resonance Fingerprinting

Magnetic Resonance Fingerprinting (MRF) enables fast quantitative imaging by matching signal evolutions to a predefined dictionary. However, conventional dictionary matching suffers from exponential growth in computational cost and memory usage as the number of parameters increases, limiting its scalability to multi-parametric mapping. To address this, recent work has explored deep learning-based approaches as alternatives to DM. We propose GAST-Mamba, an end-to-end framework that combines a dual Mamba-based encoder with a Gate-Aware Spatial-Temporal (GAST) processor. Built on structured state-space models, our architecture efficiently captures long-range spatial dependencies with linear complexity. On 5 times accelerated simulated MRF data (200 frames), GAST-Mamba achieved a T1 PSNR of 33.12~dB, outperforming SCQ (31.69~dB). For T2 mapping, it reached a PSNR of 30.62~dB and SSIM of 0.9124. In vivo experiments further demonstrated improved anatomical detail and reduced artifacts. Ablation studies confirmed that each component contributes to performance, with the GAST module being particularly important under strong undersampling. These results demonstrate the effectiveness of GAST-Mamba for accurate and robust reconstruction from highly undersampled MRF acquisitions, offering a scalable alternative to traditional DM-based methods.

eess.IV↗

arXiv2025

Extreme Scenario Characterization for High Renewable Energy Penetrated Power Systems over Long Time Scales

Power systems with high renewable energy penetration are highly influenced by weather conditions, often facing significant challenges such as persistent power shortages and severe power fluctuations over long time scales. This paper addresses the critical need for effective characterization of extreme scenarios under these situations. First, novel risk indices are proposed to quantify the severity of continuous power shortages and substantial power fluctuations over long-term operations. These indices are independent of specific scheduling strategies and incorporate the system's resource regulation capabilities. By employing a filtering-based approach, the proposed indices focus on retaining key characteristics of continuous power shortages and fluctuation events, enabling the identification of extreme scenarios on long time scales. Secondly, an extreme scenario generation method is developed using Gaussian mixture models and sequential Monte Carlo simulation. Especially, this method periodically evaluates the severity of generated scenarios based on the defined risk indices, retaining extreme scenarios while discarding less critical ones. Finally, case studies based on real-world data demonstrate the efficacy of the proposed method. The results confirm that integrating the identified extreme scenarios significantly enhances the system's ability to ensure long-term security and reliability under high renewable energy penetration.

eess.SY↗

arXiv2025

AG-VPReID 2025: Aerial-Ground Video-based Person Re-identification Challenge Results

Person re-identification (ReID) across aerial and ground vantage points has become crucial for large-scale surveillance and public safety applications. Although significant progress has been made in ground-only scenarios, bridging the aerial-ground domain gap remains a formidable challenge due to extreme viewpoint differences, scale variations, and occlusions. Building upon the achievements of the AG-ReID 2023 Challenge, this paper introduces the AG-VPReID 2025 Challenge - the first large-scale video-based competition focused on high-altitude (80-120m) aerial-ground ReID. Constructed on the new AG-VPReID dataset with 3,027 identities, over 13,500 tracklets, and approximately 3.7 million frames captured from UAVs, CCTV, and wearable cameras, the challenge featured four international teams. These teams developed solutions ranging from multi-stream architectures to transformer-based temporal reasoning and physics-informed modeling. The leading approach, X-TFCLIP from UAM, attained 72.28% Rank-1 accuracy in the aerial-to-ground ReID setting and 70.77% in the ground-to-aerial ReID setting, surpassing existing baselines while highlighting the dataset's complexity. For additional details, please refer to the official website at https://agvpreid25.github.io.

cs.CV↗

arXiv2025

Unveiling Causal Reasoning in Large Language Models: Reality or Mirage?

Causal reasoning capability is critical in advancing large language models (LLMs) toward strong artificial intelligence. While versatile LLMs appear to have demonstrated capabilities in understanding contextual causality and providing responses that obey the laws of causality, it remains unclear whether they perform genuine causal reasoning akin to humans. However, current evidence indicates the contrary. Specifically, LLMs are only capable of performing shallow (level-1) causal reasoning, primarily attributed to the causal knowledge embedded in their parameters, but they lack the capacity for genuine human-like (level-2) causal reasoning. To support this hypothesis, methodologically, we delve into the autoregression mechanism of transformer-based LLMs, revealing that it is not inherently causal. Empirically, we introduce a new causal Q&A benchmark called CausalProbe-2024, whose corpora are fresh and nearly unseen for the studied LLMs. The LLMs exhibit a significant performance drop on CausalProbe-2024 compared to earlier benchmarks, indicating the fact that they primarily engage in level-1 causal reasoning. To bridge the gap towards level-2 causal reasoning, we draw inspiration from the fact that human reasoning is usually facilitated by general knowledge and intended goals. We propose G^2-Reasoner, a method that incorporates general knowledge and goal-oriented prompts into LLMs' causal reasoning processes. Experiments demonstrate that G^2-Reasoner significantly enhances LLMs' causal reasoning capability, particularly in fresh and counterfactual contexts. This work sheds light on a new path for LLMs to advance towards genuine causal reasoning, going beyond level-1 and making strides towards level-2.

cs.AI↗

arXiv2025

Exclusive Style Removal for Cross Domain Novel Class Discovery

As a promising field in open-world learning, \textit{Novel Class Discovery} (NCD) is usually a task to cluster unseen novel classes in an unlabeled set based on the prior knowledge of labeled data within the same domain. However, the performance of existing NCD methods could be severely compromised when novel classes are sampled from a different distribution with the labeled ones. In this paper, we explore and establish the solvability of NCD with cross domain setting under the necessary condition that the style information needs to be removed. Based on the theoretical analysis, we introduce an exclusive style removal module for extracting style information that is distinctive from the baseline features, thereby facilitating inference. Moreover, this module is easy to integrate with other NCD methods, acting as a plug-in to improve performance on novel classes with different distributions compared to the labeled set. Additionally, recognizing the non-negligible influence of different backbones and pre-training strategies on the performance of the NCD methods, we build a fair benchmark for future NCD research. Extensive experiments on three common datasets demonstrate the effectiveness of our proposed style removal strategy.

cs.CV↗

arXiv2025

Diffusion Transformer-to-Mamba Distillation for High-Resolution Image Generation

The quadratic computational complexity of self-attention in diffusion transformers (DiT) introduces substantial computational costs in high-resolution image generation. While the linear-complexity Mamba model emerges as a potential alternative, direct Mamba training remains empirically challenging. To address this issue, this paper introduces diffusion transformer-to-mamba distillation (T2MD), forming an efficient training pipeline that facilitates the transition from the self-attention-based transformer to the linear complexity state-space model Mamba. We establish a diffusion self-attention and Mamba hybrid model that simultaneously achieves efficiency and global dependencies. With the proposed layer-level teacher forcing and feature-based knowledge distillation, T2MD alleviates the training difficulty and high cost of a state space model from scratch. Starting from the distilled 512$\times$512 resolution base model, we push the generation towards 2048$\times$2048 images via lightweight adaptation and high-resolution fine-tuning. Experiments demonstrate that our training path leads to low overhead but high-quality text-to-image generation. Importantly, our results also justify the feasibility of using sequential and causal Mamba models for generating non-causal visual output, suggesting the potential for future exploration.

cs.CV↗

arXiv2025

Mixed-Mode In-Memory Computing: Towards High-Performance Logic Processing In A Memristive Crossbar Array

In-memory computing is a promising alternative to traditional computer designs, as it helps overcome performance limits caused by the separation of memory and processing units. However, many current approaches struggle with unreliable device behavior, which affects data accuracy and efficiency. In this work, the authors present a new computing method that combines two types of operations,those based on electrical resistance and those based on voltage, within each memory cell. This design improves reliability and avoids the need for expensive current measurements. A new software tool also helps automate the design process, supporting highly parallel operations in dense two-dimensional memory arrays. The approach balances speed and space, making it practical for advanced computing tasks. Demonstrations include a digital adder and a key part of the encryption module, showing both strong performance and accuracy. This work offers a new direction for reliable and efficient in-memory computing systems with real-world applications.

physics.app-ph↗

arXiv2025

Deep Reparameterization for Full Waveform Inversion: Architecture Benchmarking, Robust Inversion, and Multiphysics Extension

Full waveform inversion (FWI) is a high-resolution subsurface imaging technique, but its effectiveness is limited by challenges such as noise contamination, sparse acquisition, and artifacts from multiparameter coupling. To address these limitations, this study develops a deep reparameterized FWI (DR-FWI) framework, in which subsurface parameters are represented by a deep neural network. Instead of directly optimizing the parameters, DR-FWI optimizes the network weights to reconstruct them, thereby embedding structural priors and facilitating optimization. To provide benchmark guidelines for the design of DR-FWI, we conduct a comparative analysis of three representative architectures (U-Net, CNN, MLP) combined with two initial model embedding strategies: one pretraining the network to generate predefined initial models (pretraining-based), while the other directly adds network outputs to the initial models. Extensive ablation experiments show that combining CNN with pretraining-based initialization significantly enhances inversion accuracy, offering valuable insights into network design. To further understand the mechanism of DR-FWI, spectral bias analysis reveals that the network first captures low-frequency features and gradually reconstructs high-frequency details, enabling an adaptive multi-scale inversion strategy. Notably, the robustness of DR-FWI is validated under various noise levels and sparse acquisition scenarios, where its strong performance with limited shots and receivers demonstrates reduced reliance on dense observational data. Additionally, a backbone-branch structure is proposed to extend DR-FWI to multiparameter inversion, and its efficacy in mitigating cross-parameter interference is validated on a synthetic anomaly model and the Marmousi2 model. These results suggest a promising direction for joint inversion involving multiple parameters or multiphysics.

physics.geo-ph↗

arXiv2025

BCRNet: Enhancing Landmark Detection in Laparoscopic Liver Surgery via Bezier Curve Refinement

Laparoscopic liver surgery, while minimally invasive, poses significant challenges in accurately identifying critical anatomical structures. Augmented reality (AR) systems, integrating MRI/CT with laparoscopic images based on 2D-3D registration, offer a promising solution for enhancing surgical navigation. A vital aspect of the registration progress is the precise detection of curvilinear anatomical landmarks in laparoscopic images. In this paper, we propose BCRNet (Bezier Curve Refinement Net), a novel framework that significantly enhances landmark detection in laparoscopic liver surgery primarily via the Bezier curve refinement strategy. The framework starts with a Multi-modal Feature Extraction (MFE) module designed to robustly capture semantic features. Then we propose Adaptive Curve Proposal Initialization (ACPI) to generate pixel-aligned Bezier curves and confidence scores for reliable initial proposals. Additionally, we design the Hierarchical Curve Refinement (HCR) mechanism to enhance these proposals iteratively through a multi-stage process, capturing fine-grained contextual details from multi-scale pixel-level features for precise Bezier curve adjustment. Extensive evaluations on the L3D and P2ILF datasets demonstrate that BCRNet outperforms state-of-the-art methods, achieving significant performance improvements. Code will be available.

cs.CV↗

arXiv2025

SUSEP-Net: Simulation-Supervised and Contrastive Learning-based Deep Neural Networks for Susceptibility Source Separation

Quantitative susceptibility mapping (QSM) provides a valuable tool for quantifying susceptibility distributions in human brains; however, two types of opposing susceptibility sources (i.e., paramagnetic and diamagnetic), may coexist in a single voxel, and cancel each other out in net QSM images. Susceptibility source separation techniques enable the extraction of sub-voxel information from QSM maps. This study proposes a novel SUSEP-Net for susceptibility source separation by training a dual-branch U-net with a simulation-supervised training strategy. In addition, a contrastive learning framework is included to explicitly impose similarity-based constraints between the branch-specific guidance features in specially-designed encoders and the latent features in the decoders. Comprehensive experiments were carried out on both simulated and in vivo data, including healthy subjects and patients with pathological conditions, to compare SUSEP-Net with three state-of-the-art susceptibility source separation methods (i.e., APART-QSM, \c{hi}-separation, and \c{hi}-sepnet). SUSEP-Net consistently showed improved results compared with the other three methods, with better numerical metrics, improved high-intensity hemorrhage and calcification lesion contrasts, and reduced artifacts in brains with pathological conditions. In addition, experiments on an agarose gel phantom data were conducted to validate the accuracy and the generalization capability of SUSEP-Net.

eess.IV↗