arXiv subjects

Zhengping Ji

Publications and source records attributed to Zhengping Ji.

At least 19 recordsLinked to original sources

arXiv2026

BodyCam-VQA: Enhanced Body-Worn Camera Video Captioning via Multimodal Reasoning and Probe Question Generation

Police body-worn camera (BWC) footage has emerged as a critical aspect of law enforcement that ensures legal transparency, officer accountability, and the protection of civil rights. However, effectively processing this data remains a significant challenge due to its multimodal video format. BWC videos, in many cases, comprise chaotic scenes with low visual quality, rapid movement/interactions, and high-noise audio that make visual understanding a challenge for even SOTA multimodal models. Current Vision-Language Models (VLMs) frequently overlook critical forensic details, such as the presence of valuable evidence or the latent nuances of suspect-officer interactions, which are vital for fair legal outcomes and civilian/officer safety. To address these limitations, we propose an Adaptive Visual Question Answering (VQA) framework engineered for high-stakes law enforcement. Our framework employs a structured reasoning approach to extract fine-grained visual evidence that traditional captioning systems fail to capture. We experiment with multiple question generation models, including foundation models and fine-tuned open-weight models, to observe performance variation among question generation model implementations. Our results demonstrate that this VQA-driven architecture provides a more reliable, objective, and detailed record of enforcement events, ultimately serving as a powerful tool to protect both law enforcement officers and the public through AI-assisted forensic clarity.

cs.CV↗

arXiv2026

Mean-covariance turnpikes in Wasserstein distributionally robust linear-quadratic control

We study long-horizon Wasserstein-penalized minimax control for discrete-time stochastic linear systems with empirical disturbance data, in which adversarial disturbance distributions induce time-varying mean and covariance dynamics, making standard turnpike arguments not directly applicable. For possibly uncentered data, we characterize the generally nonzero mean reference through a reduced convex-concave Hamiltonian saddle problem. We prove horizon-uniform, two-sided exponential turnpike estimates for the mean state, adjoint, control, worst-case disturbance mean, and closed-loop covariance, showing that they spend the majority of time near static references when the horizon is long. We further construct a hybrid policy combining time-independent affine feedback with finite-horizon steering over a terminal layer, proving that its worst-case cost gap decays exponentially with the terminal-layer length uniformly in the horizon, which helps reducing the computation cost for long-horizon robust controls. Numerical examples illustrate the estimates and their dependence on the Wasserstein penalty.

math.OC↗

arXiv2026

Reachability and asymptotics of Gaussian Transformer dynamics

We formulate data propagation through the Transformer, the machine learning architecture powering large language models, as a nonlinear control system on the space of probability measures. For the mean-field Transformer model with self-attention and affine feed-forward layers, we prove that Gaussian distributions remain exactly Gaussian along the induced flow. This invariance reduces the infinite-dimensional measure dynamics to a finite-dimensional bilinear control system governing the evolution of the mean and covariance, reformulates the expressive capacity of Transformers as a reachability problem for prescribed Gaussian moments, and reveals a novel connection with Riccati-type equations from classical filtering and control. For time-varying controls, we prove exact finite-time reachability of any target Gaussian distribution whose covariance matrix has the same rank as the initial one, this rank constraint being an intrinsic invariant of the dynamics. For time-invariant parameters, we derive explicit spectral conditions leading either to asymptotic stability toward positive-definite equilibria or to finite-time blow-up of the covariance. Numerical experiments complement the theory by showing that practical Transformers with Gaussian inputs remain close to moment-matched Gaussian distributions through early and intermediate layers, while Transformers with prescribed attention matrices reproduce the predicted covariance regimes: bounded evolution in stabilizing configurations and blow-up in destabilizing ones.

cs.LG↗

arXiv2026

How Should Video LLMs Output Time? An Analysis of Efficient Temporal Grounding Paradigms

While Multimodal Large Language Models (MLLMs) have advanced Video Temporal Grounding (VTG), existing methods often couple output paradigms with different backbones, datasets, and training protocols. This makes it challenging to isolate the specific impact of the output design. Additionally, as VTG systems are increasingly considered for resource-constrained edge deployment, the trade-off between output formulation and system-level efficiency requires systematic investigation. In this paper, we present a controlled empirical study comparing three dominant VTG output paradigms: Text Numeral Generation, Temporal Token Generation, and Continuous Temporal Decoding. We evaluate these paradigms across identical compact VLMs (SmolVLM2, FastVLM, and Molmo2) using consistent datasets and LoRA fine-tuning protocols. Evaluations on Charades-STA, QVHighlights, and YouCook2 measure both localization accuracy and system efficiency, including inference latency, training throughput, and parameter overhead. Our results demonstrate that the choice of output formulation significantly affects both grounding accuracy and computational cost, independent of model scale. Specifically, the continuous distribution paradigm consistently achieves the most favorable efficiency-accuracy trade-off on the Pareto frontier, delivering robust localization with minimal latency overhead. These findings provide objective empirical guidelines for designing efficient, deployment-ready VTG systems.

cs.CV↗

arXiv2026

LiME: Lightweight Mixture of Experts for Efficient Multimodal Multi-task Learning

MoE-PEFT methods combine Mixture of Experts with parameter-efficient fine-tuning for multi-task adaptation, but require separate adapters per expert causing trainable parameters to scale linearly with expert count and limiting applicability to adapter-based architectures. We propose LiME (Lightweight Mixture of Experts), which achieves expert specialization through lightweight modulation rather than adapter replication. Instead of separate adapters, LiME uses a single shared PEFT module and modulates its output with lightweight expert vectors, reducing expert parameters while generalizing to any PEFT method. Notably, LiME introduces zero-parameter routing by leveraging existing frozen and adapted representations eliminating learned router parameters typically required per layer. Theoretically, we prove that (i) more experts preserve more task-relevant information and (ii) modulation approximates full expert-specific PEFT with bounded error. LiME further incorporates n-gram windowed routing and adaptive expert selection (Auto Top-K) based on routing confidence. Experiments on MMT-47, a multimodal multi-task benchmark with 47 tasks spanning text, image, and video, demonstrate that LiME achieves competitive or superior performance while using up to 4x fewer trainable parameters and up to 29% faster training compared to corresponding MoE-PEFT baselines.

cs.LG↗

arXiv2025

Hadamard-L\'{e}vy theorems for maps taking values in a finite dimensional space

We propose global surjectivity theorems of differentiable maps based on second order conditions. Using the homotopy continuation method, we demonstrate that, for a $C^2$ differentiable map from a Hilbert space to a finite-dimensional Euclidean space, when its second-order differential has uniform upper and lower bounds, it has a global path-lifting property in the presence of singularities. This is then applied to the nonlinear motion planning problem, establishing in some cases the well-posedness of the continuation method despite critical values of the endpoint maps.

math.CA↗

arXiv2025

Auto-Drafting Police Reports from Noisy ASR Outputs: A Trust-Centered LLM Approach

Achieving a delicate balance between fostering trust in law enforcement and protecting the rights of both officers and civilians continues to emerge as a pressing research and product challenge in the world today. In the pursuit of fairness and transparency, this study presents an innovative AI-driven system designed to generate police report drafts from complex, noisy, and multi-role dialogue data. Our approach intelligently extracts key elements of law enforcement interactions and includes them in the draft, producing structured narratives that are not only high in quality but also reinforce accountability and procedural clarity. This framework holds the potential to transform the reporting process, ensuring greater oversight, consistency, and fairness in future policing practices. A demonstration video of our system can be accessed at https://drive.google.com/file/d/1kBrsGGR8e3B5xPSblrchRGj-Y-kpCHNO/view?usp=sharing

cs.CL↗

arXiv2025

DHP Benchmark: Are LLMs Good NLG Evaluators?

Large Language Models (LLMs) are increasingly serving as evaluators in Natural Language Generation (NLG) tasks; this is often referred to as ``LLM-as-a-judge'' paradigm. However, the capabilities of LLMs in evaluating NLG quality remain underexplored. Current studies depend on human assessments and simple metrics that fail to capture the discernment of LLMs across diverse NLG tasks. To address this gap, we propose the Discernment of Hierarchical Perturbation (DHP) benchmarking framework, which provides quantitative discernment scores for LLMs. This framework leverages hierarchically perturbed text data and statistical tests to systematically measure the NLG evaluation capabilities of LLMs. We re-established six evaluation datasets for this benchmark, covering four NLG tasks: Summarization, Story Completion, Question Answering, and Translation. Our comprehensive benchmarking of five major LLM families provides critical insight into their strengths and limitations as NLG evaluators. Our dataset is available at https://huggingface.co/datasets/YCWANGVINCE/DHP_Benchmark.

cs.CL↗

arXiv2024

On Control Networks Over Finite Lattices

The modeling and control of networks over finite lattices are studied via the algebraic state space approach. Using the semi-tensor product of matrices, we obtain the algebraic state space representation of the dynamics of (control) networks over finite lattices. Basic properties concerning networks over sublattices and product lattices are investigated, which shows the application of the analysis of lattice structure in the model reduction and control design of networks. Then algorithms are developed to recover the lattice structure from the structural matrix of a network over a lattice, and to construct comparability graphs over a finite set to verify whether a multiple-valued logical network is defined over a lattice. Finally, numerical examples are presented to illustrate the results.

math.OC↗

arXiv2024

EVLM: Self-Reflective Multimodal Reasoning for Cross-Dimensional Visual Editing

Editing complex visual content from ambiguous or partially specified instructions remains a core challenge in vision-language modeling. Existing models can contextualize content but often fail to infer the underlying intent within a reference image or scene, leading to inconsistent or misaligned edits. We introduce the Editing Vision-Language Model (EVLM), a system that interprets ambiguous instructions in conjunction with reference visuals to produce precise, context-aware editing prompts. EVLM's key innovation is a reflective reasoning framework that translates subjective user intent into structured, actionable outputs by aligning with human-rated rationales through Reflection-Aware KL-Divergence Target Optimization (RKTO). By combining Chain-of-Thought (CoT) reasoning with RKTO alignment, EVLM captures fine-grained editing preferences without relying on binary supervision. Trained on a dataset of 30,000 CoT examples with human-annotated rationale quality, EVLM achieves substantial gains in alignment with human intent. Experiments across image, video, 3D, and 4D editing tasks show that EVLM generates coherent and high-quality instructions, providing a scalable foundation for multimodal editing and reasoning.

cs.CV↗

arXiv2024

t-Product and t-STP of Cubic Matrices With Application to Hyper-Networked Systems

Motivated by the study of dynamic control systems, this paper proposes novel algebraic operations on cubic matrices to construct both linear and nonlinear controlled dynamics. The standard t-product of cubic matrices imposes strict dimensional constraints; to resolve this, we first introduce the dimension-keeping semi-tensor product (DK-STP), which generalizes the matrix product for arbitrary dimensions. However, the DK-STP yields decoupled subsystem dynamics because it fails to capture interactions across subsystems corresponding to frontal slices. To overcome this limitation, we propose the t-semi-tensor product (t-STP), an integration of the t-product and the DK-STP that allows for coupled subsystems and greater modeling flexibility. We systematically study the algebraic structures derived from the t-STP over cubic matrices, including groups, rings, modules, and Lie groups. Finally, we obtain t-STP-based dynamic control systems over cubic matrices and demonstrate the utility of this framework by applying it to a hyper-networked evolutionary game modeling supply chain interactions.

math.RA↗

arXiv2024

FairLENS: Assessing Fairness in Law Enforcement Speech Recognition

Automatic speech recognition (ASR) techniques have become powerful tools, enhancing efficiency in law enforcement scenarios. To ensure fairness for demographic groups in different acoustic environments, ASR engines must be tested across a variety of speakers in realistic settings. However, describing the fairness discrepancies between models with confidence remains a challenge. Meanwhile, most public ASR datasets are insufficient to perform a satisfying fairness evaluation. To address the limitations, we built FairLENS - a systematic fairness evaluation framework. We propose a novel and adaptable evaluation method to examine the fairness disparity between different models. We also collected a fairness evaluation dataset covering multiple scenarios and demographic dimensions. Leveraging this framework, we conducted fairness assessments on 1 open-source and 11 commercially available state-of-the-art ASR models. Our results reveal that certain models exhibit more biases than others, serving as a fairness guideline for users to make informed choices when selecting ASR models for a given real-world scenario. We further explored model biases towards specific demographic groups and observed that shifts in the acoustic domain can lead to the emergence of new biases.

eess.AS↗

arXiv2024

Observer-Based Realization of Control Systems

A novel model reduction framework for large-scale complex systems is proposed by introducing function-type dynamic control systems via the dimension-keeping semi-tensor product (DK-STP) of matrices. Utilizing bridge matrices, the DK-STP facilitates the construction of an approximate observer-based realization (OR) of a linear control system in the form of a function-type control system, where the functions serve as observers. A necessary and sufficient condition is established for the OR-system to admit exact observer dynamics. When an exact OR-system does not exist, an extended OR-system is developed by incorporating the original system's observers into its state. Furthermore, a minimal feedback extended OR-system is constructed, and its relationship to Kalman's minimal realization is analyzed. Finally, the proposed approach is extended to nonlinear control-affine systems.

math.OC↗

arXiv2024

Analysis of Discrete-Time Switched Linear Systems under Logic Dynamic Switchings

The control properties of discrete-time switched linear systems (SLS) with switching signals generated by logical dynamic systems are studied using the semi-tensor product (STP) approach. With the algebraic state space representation (ASSR), the linear modes and the logical generators are aggregated as a hybrid system, leading to the criteria of reachability, controllability, observability, and reconstructibility of the SLSs. Algorithms for checking these properties are given. Then, two kinds of realization problems concerning whether the logical dynamic systems can generate the desired switching signals are investigated, and necessary and sufficient conditions for the realisability of the required switching signals are given with respect to the cases of fixed operating time switching and finite reference signal switching.

eess.SY↗

arXiv2023

On Universal Eigenvalues and Eigenvectors of Hypermatrices

A generalized eigenvector of a hypermatrix, called the universal (U-) eigenvector, is proposed, which extended the notion of diagonal (D-) eigenvectors in the literature. Using the semi-tensor product, the homogeneous U-eigenequation can be converted into a general eigenequation of matrix $(A-\lambda B)x=0$. A general technique for solving this equation is proposed, which leads to two kinds of eigenvalues: essential and quasi eigenvalues. The technique to convert nonhomogeneous eigenequation to homogeneous ones is also revealed. Then a hypervector decomposing method, called the monic decomposition algorithm (MDA), is developed. Using the MDA, the U-eigenproblem (including the D-eigenproblem) can be converted into general matrix eigenproblems. Some examples are presented, demonstrating the geometric meaning and potential applications of the U-eigenvalue/eigenvector.

math.NA↗

arXiv2023

Global Controllability Criteria and Motion Planning of Regular Affine Systems With Drifts

In this article, we give a condition for the global controllability of affine nonlinear control systems with drifts on Euclidean spaces. Under regularity assumptions, the condition is necessary and sufficient in the codimension-1 and codimension-2 cases, and holds for systems of higher codimensions under mild restrictions. We then investigate motion planning problems for codimension-1 affine systems, and give proof of the global existence of the lift to control curves for certain drifted systems using the homotopy continuation method.

math.OC↗

arXiv2023

Contracted Product of Hypermatrices via STP of Matrices

An equivalent definition of hypermatrices is introduced. The matrix expression of hypermatrices is proposed. Using permutation matrices, the conversion of different matrix expressions is revealed. The various contracted products of hypermatrices are realized by semi-tensor products (STP) of matrices via matrix expressions of hypermatrices.

math.NA↗

arXiv2023

Transition System Representation of Boolean Control Networks

First, the topological structure of a transition system is studied. Then, two types of transition system (TS) representations of Boolean networks (BNs) and Boolean control networks (BCNs) are investigated. The first kind of representation is state-based, which converts a BCN into a TS with either distinct control or non-distinct control. The second representation is output-based, which is also called the simulation of the original BCN. Some applications are also studied.

eess.SY↗