arXiv subjects

Feng Ma

Publications and source records attributed to Feng Ma.

At least 19 recordsLinked to original sources

arXiv2026

Inverse Low-Dimensional Manifold Reconstruction Framework for Spatiotemporal Reconstruction of Compressible Physical Fields

Compressible physical fields are widely present in the real physical world, but current artificial intelligence lacks an understanding mechanism for the non-differentiable features in compressible physical fields. Addressing the limitations of existing deep learning architectures in handling global non-differentiable features, we propose the Inverse Low-Dimensional Manifold reconstruction framework (ILDM). This framework couples the Non-differentiable Approximation Function (NAF) for capturing non-differentiable features in compressible flows with the Smooth Fluid Reconstruction (SFR) module tailored for smooth fluid regions. Extensive evaluations across 1D and 2D benchmarks, including Riemann problems and double Mach reflection, demonstrate that ILDM significantly outperforms cPINN and R-adaptive DeepONet. Specifically, ILDM achieves superior localization of non-differentiable interfaces and maintains robust super-resolution performance even with low-resolution inputs, establishing a physically consistent and scalable paradigm for data-driven fluid dynamics.

physics.comp-ph↗

arXiv2025

Enhancing the Geometric Problem-Solving Ability of Multimodal LLMs via Symbolic-Neural Integration

Recent advances in Multimodal Large Language Models (MLLMs) have achieved remarkable progress in general domains and demonstrated promise in multimodal mathematical reasoning. However, applying MLLMs to geometry problem solving (GPS) remains challenging due to lack of accurate step-by-step solution data and severe hallucinations during reasoning. In this paper, we propose GeoGen, a pipeline that can automatically generates step-wise reasoning paths for geometry diagrams. By leveraging the precise symbolic reasoning, \textbf{GeoGen} produces large-scale, high-quality question-answer pairs. To further enhance the logical reasoning ability of MLLMs, we train \textbf{GeoLogic}, a Large Language Model (LLM) using synthetic data generated by GeoGen. Serving as a bridge between natural language and symbolic systems, GeoLogic enables symbolic tools to help verifying MLLM outputs, making the reasoning process more rigorous and alleviating hallucinations. Experimental results show that our approach consistently improves the performance of MLLMs, achieving remarkable results on benchmarks for geometric reasoning tasks. This improvement stems from our integration of the strengths of LLMs and symbolic systems, which enables a more reliable and interpretable approach for the GPS task. Codes are available at https://github.com/ycpNotFound/GeoGen.

cs.CL↗

arXiv2025

MMC: Iterative Refinement of VLM Reasoning via MCTS-based Multimodal Critique

Visual language models (VLMs) have demonstrated strong performance across diverse multimodal reasoning tasks but still face challenges such as hallucinations, resulting in incorrect reasoning outcomes. Inspired by recent research on external feedback mechanisms in large language models (LLMs), we propose a multimodal actor-critic framework to enhance VLM reasoning capabilities. Specifically, the actor model generates step-by-step reasoning paths based on image and text inputs, while the critic model evaluates these reasoning paths and provides corrective feedback. The actor model iteratively refines its reasoning based on the feedback until the reasoning outcome is deemed satisfactory by the critic model. To reduce reliance on costly manual annotations, we introduce an automated method for constructing multimodal critique datasets. By leveraging Monte Carlo Tree Search (MCTS), we systematically guide the actor model to explore diverse reasoning paths. To obtain critique data for correcting erroneous reasoning steps, we prompt an annotator model to compare pairs of reasoning paths diverging from a shared ancestor node - one leading to a correct conclusion and the other to an incorrect one. This approach enables us to construct the MMC (MCTS-based Multimodal Critique) dataset, upon which we further develop a comprehensive training and inference pipeline. Extensive experiments conducted on several public benchmark datasets and mainstream VLMs demonstrate that our approach significantly improves the performance of VLM on complex multimodal reasoning tasks, underscoring its effectiveness and wide applicability.

cs.MM↗

arXiv2025

PRM-BAS: Enhancing Multimodal Reasoning through PRM-guided Beam Annealing Search

Recent work increasingly focuses on improving the reasoning capabilities of Multimodal Large Language Models (MLLMs). Among existing methods, Process Reward Models (PRMs) stand out for offering dense, step-wise supervision to guide intermediate reasoning. However, how to effectively integrate PRMs into search strategies remains an open question. In this paper, we introduce PRM-BAS (PRM-Guided Beam Annealing Search), a lightweight approach for PRM-guided reasoning that dynamically adjusts beam size -- starting with a broader search space and gradually narrowing it as contextual information accumulates, thereby balancing performance and efficiency. We further propose a unified framework for data construction and PRM training. Specifically, we construct the PRM-BAS-300k dataset by selecting 300k questions from existing datasets and performing rollouts at each step to estimate the probability of reaching a correct final answer. The PRM is then trained using a combination of value loss for absolute action quality and rank loss for relative action quality. Extensive experiments on challenging multimodal reasoning benchmarks demonstrate that PRM-BAS significantly improves reasoning performance while maintaining low computational cost. Moreover, it generalizes well across different model scales and architectures, showcasing strong robustness and plug-and-play capability.

cs.MM↗

arXiv2025

Navigation-GPT: A Robust and Adaptive Framework Utilizing Large Language Models for Navigation Applications

Existing navigation decision support systems often perform poorly when handling non-predefined navigation scenarios. Leveraging the generalization capabilities of large language model (LLM) in handling unknown scenarios, this research proposes a dual-core framework for LLM applications to address this issue. Firstly, through ReAct-based prompt engineering, a larger LLM core decomposes intricate navigation tasks into manageable sub-tasks, which autonomously invoke corresponding external tools to gather relevant information, using this feedback to mitigate the risk of LLM hallucinations. Subsequently, a fine-tuned and compact LLM core, acting like a first-mate is designed to process such information and unstructured external data, then to generates context-aware recommendations, ultimately delivering lookout insights and navigation hints that adhere to the International Regulations for Preventing Collisions at Sea (COLREGs) and other rules. Extensive experiments demonstrate the proposed framework not only excels in traditional ship collision avoidance tasks but also adapts effectively to unstructured, non-predefined, and unpredictable scenarios. A comparative analysis with DeepSeek-R1, GPT-4o and other SOTA models highlights the efficacy and rationality of the proposed framework. This research bridges the gap between conventional navigation systems and LLMs, offering a framework to enhance safety and operational efficiency across diverse navigation applications.

cs.AI↗

arXiv2024

The USTC-NERCSLIP Systems for the CHiME-8 NOTSOFAR-1 Challenge

This technical report outlines our submission system for the CHiME-8 NOTSOFAR-1 Challenge. The primary difficulty of this challenge is the dataset recorded across various conference rooms, which captures real-world complexities such as high overlap rates, background noises, a variable number of speakers, and natural conversation styles. To address these issues, we optimized the system in several aspects: For front-end speech signal processing, we introduced a data-driven joint training method for diarization and separation (JDS) to enhance audio quality. Additionally, we also integrated traditional guided source separation (GSS) for multi-channel track to provide complementary information for the JDS. For back-end speech recognition, we enhanced Whisper with WavLM, ConvNeXt, and Transformer innovations, applying multi-task training and Noise KLD augmentation, to significantly advance ASR robustness and accuracy. Our system attained a Time-Constrained minimum Permutation Word Error Rate (tcpWER) of 14.265% and 22.989% on the CHiME-8 NOTSOFAR-1 Dev-set-2 multi-channel and single-channel tracks, respectively.

eess.AS↗

arXiv2024

AutoFirm: Automatically Identifying Reused Libraries inside IoT Firmware at Large-Scale

The Internet of Things (IoT) has become indispensable to our daily lives and work. Unfortunately, developers often reuse software libraries in the IoT firmware, leading to a major security concern. If vulnerabilities or insecure versions of these libraries go unpatched, a massive number of IoT devices can be impacted. In this paper, we propose the AutoFirm, an automated tool for detecting reused libraries in IoT firmware at a large scale. Specifically, AutoFirm leverages the syntax information (library name and version) to determine whether IoT firmware reuses the libraries. We conduct a large-scale empirical study of reused libraries of IoT firmware, investigating more than 6,900+ firmware and 2,700+ distinct vulnerabilities affecting 11,300+ vulnerable versions from 349 open-source software libraries. Leveraging this diverse information set, we conduct a qualitative assessment of vulnerable library versions to understand security gaps and the misplaced trust of libraries in IoT firmware. Our research reveals that: manufacturers neglected to update outdated libraries for IoT firmware in 67.3\% of cases; on average, outdated libraries persisted for over 1.34 years prior to remediation; vulnerabilities of software libraries have posed server threats to widespread IoT devices.

cs.CR↗

arXiv2024

Quality-aware Masked Diffusion Transformer for Enhanced Music Generation

Text-to-music (TTM) generation, which converts textual descriptions into audio, opens up innovative avenues for multimedia creation. Achieving high quality and diversity in this process demands extensive, high-quality data, which are often scarce in available datasets. Most open-source datasets frequently suffer from issues like low-quality waveforms and low text-audio consistency, hindering the advancement of music generation models. To address these challenges, we propose a novel quality-aware training paradigm for generating high-quality, high-musicality music from large-scale, quality-imbalanced datasets. Additionally, by leveraging unique properties in the latent space of musical signals, we adapt and implement a masked diffusion transformer (MDT) model for the TTM task, showcasing its capacity for quality control and enhanced musicality. Furthermore, we introduce a three-stage caption refinement approach to address low-quality captions' issue. Experiments show state-of-the-art (SOTA) performance on benchmark datasets including MusicCaps and the Song-Describer Dataset with both objective and subjective metrics. Demo audio samples are available at https://qa-mdt.github.io/, code and pretrained checkpoints are open-sourced at https://github.com/ivcylc/OpenMusic.

cs.SD↗

arXiv2022

A rank-two relaxed parallel splitting version of the augmented Lagrangian method with step size in (0,2) for separable convex programming

The augmented Lagrangian method (ALM) is classic for canonical convex programming problems with linear constraints, and it finds many applications in various scientific computing areas. A major advantage of the ALM is that the step for updating the dual variable can be further relaxed with a step size in $(0,2)$, and this advantage can easily lead to numerical acceleration for the ALM. When a separable convex programming problem is discussed and a corresponding splitting version of the classic ALM is considered, convergence may not be guaranteed and thus it is seemingly impossible that a step size in $(0,2)$ can be carried on to the relaxation step for updating the dual variable. We show that for a parallel splitting version of the ALM, a step size in $(0,2)$ can be maintained for further relaxing both the primal and dual variables if the relaxation step is simply corrected by a rank-two matrix. Hence, a rank-two relaxed parallel splitting version of the ALM with a step size in $(0,2)$ is proposed for separable convex programming problems. We validate that the new algorithm can numerically outperform existing algorithms of the same kind significantly by testing some applications.

math.OC↗

arXiv2022

Fluid Simulation System Based on Graph Neural Network

Traditional computational fluid dynamics calculates the physical information of the flow field by solving partial differential equations, which takes a long time to calculate and consumes a lot of computational resources. We build a fluid simulation simulator based on the graph neural network architecture. The simulator has fast computing speed and low consumption of computing resources. We regard the computational domain as a structural graph, and the computational nodes in the structural graph determine neighbor nodes through adaptive sampling. Building deep learning architectures with attention graph neural networks. The fluid simulation simulator is trained according to the simulation results of the flow field around the cylinder with different Reynolds numbers. The trained fluid simulation simulator not only has a very high accuracy for the prediction of the flow field in the training set, but also can extrapolate the flow field outside the training set. Compared to traditional CFD solvers, the fluid simulation simulator achieves a speedup of 2-3 orders of magnitude. The fluid simulation simulator provides new ideas for the rapid optimization and design of fluid mechanics models and the real-time control of intelligent fluid mechanisms.

physics.flu-dyn↗

arXiv2021

A generalized primal-dual algorithm with improved convergence condition for saddle point problems

We generalize the well-known primal-dual algorithm proposed by Chambolle and Pock for saddle point problems, and improve the condition for ensuring its convergence. The improved convergence-guaranteeing condition is effective for the generic setting, and it is shown to be optimal. It also allows us to discern larger step sizes for the resulting subproblems, and thus provides a simple and universal way to improve numerical performance of the original primal-dual algorithm. In addition, we present a structure-exploring heuristic to further relax the convergence-guaranteeing condition for some specific saddle point problems, which could yield much larger step sizes and hence significantly better performance. Effectiveness of this heuristic is numerically illustrated by the classic assignment problem.

math.OC↗

arXiv2021

A Four-Stage Data Augmentation Approach to ResNet-Conformer Based Acoustic Modeling for Sound Event Localization and Detection

In this paper, we propose a novel four-stage data augmentation approach to ResNet-Conformer based acoustic modeling for sound event localization and detection (SELD). First, we explore two spatial augmentation techniques, namely audio channel swapping (ACS) and multi-channel simulation (MCS), to deal with data sparsity in SELD. ACS and MDS focus on augmenting the limited training data with expanding direction of arrival (DOA) representations such that the acoustic models trained with the augmented data are robust to localization variations of acoustic sources. Next, time-domain mixing (TDM) and time-frequency masking (TFM) are also investigated to deal with overlapping sound events and data diversity. Finally, ACS, MCS, TDM and TFM are combined in a step-by-step manner to form an effective four-stage data augmentation scheme. Tested on the Detection and Classification of Acoustic Scenes and Events (DCASE) 2020 data set, our proposed augmentation approach greatly improves the system performance, ranking our submitted system in the first place in the SELD task of the DCASE 2020 Challenge. Furthermore, we employ a ResNet-Conformer architecture to model both global and local context dependencies of an audio sequence and win the first place in the DCASE 2022 SELD evaluations.

cs.SD↗

arXiv2020

A Revisit of Chen-Teboulle's Proximal-based Decomposition Method

In this paper, we show that Chen-Teboulle's proximal-based decomposition method can be interpreted as a proximal augmented Lagrangian method. More precisely, it coincides with a linearized augmented Lagrangian method. We then proposed three generalized methods based on this interpretation. By invoking recent work (He et al., IMA J. Numer. Anal., 32 (2020), pp. 227--245), we show that the step size condition of Chen-Teboulle's method can be relaxed without adding any further assumptions. Our analysis offers a new insight into this proximal-based decomposition method.

math.OC↗

arXiv2005

Shot Noise Suppression in Avalanche Photodiodes

We identify a new shot noise suppression mechanism in a thin (~100 nm) heterostructure avalanche photodiode. In the low-gain regime the shot noise is suppressed due to temporal correlations within amplified current pulses. We demonstrate in a Monte Carlo simulation that the effective excess noise factors can be <1, and reconcile the apparent conflict between theory and experiments. This shot noise suppression mechanism is independent of known mechanisms such as Coulomb interaction, or reflection at heterojunction interfaces.

physics.comp-ph↗

arXiv2002

Discovery of Newly Formed Broad Absorption Lines in a Radio-Loud Quasar

We report a serendipitous discovery of broad absorption lines that were newly formed in the spectrum of the high-redshift, luminous radio-loud quasar TEX 1726 + 344, in a time interval of only 12 years. This is the first quasar showing a transition from narrow absorption lines to broad absorption lines. It also becomes one of the few radio-loud broad absorption line quasars. The gas cloud responsible for these broad absorption lines is derived to have parameters coinciding with those of the remnant of a tidally disrupted star.

astro-ph↗

arXiv2002

Spin--Down Power in Astrophysics

While the accretion power in astrophysics has been studied in many astronomical environments, the ``spin--down power'' is often neglected. In this essay I demonstrate that the spin--down power alone may drive a rotating system from sub-critical condition to critical condition with a small but finite probability. In the case of an isolated spinning-down neutron star, the star may undergo a quark--hadron phase transition in its center and become observable as a soft gamma repeater or a cosmological gamma--ray burst. For a spinning--down white dwarf, its Chandrasekhar mass limit will decrease and may reach the stellar mass, then the star explodes to a type Ia supernova. Gravitational wave detectors may be able to test these models.

astro-ph↗

arXiv2001

Hidden Blazars and Emission Line Variability of High Redshift Quasars

We have carried out a survey to search for hidden blazars in a sample of $z{\sim}2$ radio-loud quasars. The idea is based on our prediction that we should be able to see large CIV line variability not associated with observed continuum variations or most other emission lines in every radio-loud quasar. Here we report the initial results including the discovery of large CIV line variations in two quasars.

astro-ph↗

arXiv2001

Discovery of Hidden Blazars

A blazar is believed to exist in every radio-loud quasar. This is expected in a unified scheme where the differences in both optical and radio observations of radio-loud quasars are the result of different viewing angles. We have predicted that blazars may be detected using emission line ratio variations caused by variable illumination of gas clouds in the broad emission line region. In a spectroscopic search of 62 quasars at a redshift of about 2, we have discovered large (>20%) variations of the emission line ratios, CIV/CIII] or CIV/Ly$α$, when compared with historical data taken over 10 years ago. This result is consistent with our prediction, and thus supports the unification scheme for radio-loud quasars.

astro-ph↗