arXiv subjects

Wen Ji

Publications and source records attributed to Wen Ji.

At least 19 recordsLinked to original sources

arXiv2026

DAT: Dual-Aware Adaptive Transmission for Efficient Multimodal LLM Inference in Edge-Cloud Systems

Multimodal large language models (MLLMs) have shown strong capability in semantic understanding and visual reasoning, yet their use on continuous video streams in bandwidth-constrained edge-cloud systems incurs prohibitive computation and communication overhead and hinders low-latency alerting and effective visual evidence delivery. To address this challenge, we propose DAT to achieve high-quality semantic generation, low-latency event alerting, and effective visual evidence supplementation. To reduce unnecessary deep reasoning costs, we propose a collaborative small-large model cascade. A lightweight edge-side small model acts as a gating module to filter non-target-event frames and perform object detection, triggering MLLM inference only for suspicious frames. Building on this, we introduce an efficient fine-tuning strategy with visual guidance and semantic prompting, which improves structured event understanding, object detection, and output consistency. To ensure low-latency semantic alerting and effective visual evidence supplementation under bandwidth constraints, we further devise a semantics and bandwidth-aware multi-stream adaptive transmission optimization method. Experimental results show that DAT achieves 98.83% recognition accuracy and 100% output consistency. Under severe congestion, it reduces weighted semantic alert delay by up to 77.5% and delivers 98.33% of visual evidence within 0.5 s, demonstrating the effectiveness of jointly optimizing cascade inference and elastic transmission.

cs.MM↗

arXiv2026

UrbanAlign: Post-hoc Semantic Calibration for VLM-Human Preference Alignment

Vision-language models (VLMs) can describe urban scenes in rich detail, yet consistently fail to produce reliable human preference labels in domain-specific tasks such as safety assessment and aesthetic evaluation. The standard fix, fine-tuning or RLHF, requires large-scale annotations and model retraining. We ask a different question: can a frozen VLM be aligned with human preferences without modifying any weights? Our key insight is that VLMs are strong concept extractors but poor decision calibrators. We propose a three-stage post-hoc pipeline that exploits this asymmetry: (i) interpretable evaluation dimensions are automatically mined from consensus exemplars; (ii) an Observer-Debater-Judge chain extracts robust concept scores from the frozen VLM; and (iii) locally-weighted ridge regression on a hybrid manifold calibrates these scores to human ratings. Applied as UrbanAlign on Place Pulse 2.0, the framework reaches 72.2% accuracy (kappa=0.45) across six perception categories, outperforming all baselines by +11.0 pp and zero-shot VLM by +15.5 pp, with full interpretability and zero weight modification.

cs.CV↗

arXiv2026

ThinkDrive: Chain-of-Thought Guided Progressive Reinforcement Learning Fine-Tuning for Autonomous Driving

With the rapid advancement of large language models (LLMs) technologies, their application in the domain of autonomous driving has become increasingly widespread. However, existing methods suffer from unstructured reasoning, poor generalization, and misalignment with human driving intent. While Chain-of-Thought (CoT) reasoning enhances decision transparency, conventional supervised fine-tuning (SFT) fails to fully exploit its potential, and reinforcement learning (RL) approaches face instability and suboptimal reasoning depth. We propose ThinkDrive, a CoT guided progressive RL fine-tuning framework for autonomous driving that synergizes explicit reasoning with difficulty-aware adaptive policy optimization. Our method employs a two-stage training strategy. First, we perform SFT using CoT explanations. Then, we apply progressive RL with a difficulty-aware adaptive policy optimizer that dynamically adjusts learning intensity based on sample complexity. We evaluate our approach on a public dataset. The results show that ThinkDrive outperforms strong RL baselines by 1.45%, 1.95%, and 1.01% on exam, easy-exam, and accuracy, respectively. Moreover, a 2B-parameter model trained with our method surpasses the much larger GPT-4o by 3.28% on the exam metric.

cs.AI↗

arXiv2026

AIVD: Adaptive Edge-Cloud Collaboration for Accurate and Efficient Industrial Visual Detection

Multimodal large language models (MLLMs) demonstrate exceptional capabilities in semantic understanding and visual reasoning, yet they still face challenges in precise object localization and resource-constrained edge-cloud deployment. To address this, this paper proposes the AIVD framework, which achieves unified precise localization and high-quality semantic generation through the collaboration between lightweight edge detectors and cloud-based MLLMs. To enhance the cloud MLLM's robustness against edge cropped-box noise and scenario variations, we design an efficient fine-tuning strategy with visual-semantic collaborative augmentation, significantly improving classification accuracy and semantic consistency. Furthermore, to maintain high throughput and low latency across heterogeneous edge devices and dynamic network conditions, we propose a heterogeneous resource-aware dynamic scheduling algorithm. Experimental results demonstrate that AIVD substantially reduces resource consumption while improving MLLM classification performance and semantic generation quality. The proposed scheduling strategy also achieves higher throughput and lower latency across diverse scenarios.

cs.CV↗

arXiv2025

Adaptive Guidance Semantically Enhanced via Multimodal LLM for Edge-Cloud Object Detection

Traditional object detection methods face performance degradation challenges in complex scenarios such as low-light conditions and heavy occlusions due to a lack of high-level semantic understanding. To address this, this paper proposes an adaptive guidance-based semantic enhancement edge-cloud collaborative object detection method leveraging Multimodal Large Language Models (MLLM), achieving an effective balance between accuracy and efficiency. Specifically, the method first employs instruction fine-tuning to enable the MLLM to generate structured scene descriptions. It then designs an adaptive mapping mechanism that dynamically converts semantic information into parameter adjustment signals for edge detectors, achieving real-time semantic enhancement. Within an edge-cloud collaborative inference framework, the system automatically selects between invoking cloud-based semantic guidance or directly outputting edge detection results based on confidence scores. Experiments demonstrate that the proposed method effectively enhances detection accuracy and efficiency in complex scenes. Specifically, it can reduce latency by over 79% and computational cost by 70% in low-light and highly occluded scenes while maintaining accuracy.

cs.CV↗

arXiv2025

MoA-Off: Adaptive Heterogeneous Modality-Aware Offloading with Edge-Cloud Collaboration for Efficient Multimodal LLM Inference

Multimodal large language models (MLLMs) enable powerful cross-modal inference but impose significant computational and latency burdens, posing severe challenges for deployment in resource-constrained environments. In this paper, we propose MoA-Off, an adaptive heterogeneous modality-aware offloading framework with edge-cloud collaboration for efficient MLLM inference. MoA-Off introduces a lightweight heterogeneous modality-aware module that estimates the complexity of heterogeneous inputs through multi-dimensional feature analysis. Then, an adaptive edge-cloud collaborative offloading strategy is proposed that dynamically schedules workloads between edge and cloud based on modality-aware complexity scores and real-time system states. The experimental results demonstrate that MoA-Off can achieve over 30% reduction in latency and 30%-65% decrease in resource overhead while maintaining competitive accuracy compared to traditional approaches.

cs.DC↗

arXiv2025

EC2MoE: Adaptive End-Cloud Pipeline Collaboration Enabling Scalable Mixture-of-Experts Inference

The Mixture-of-Experts (MoE) paradigm has emerged as a promising solution to scale up model capacity while maintaining inference efficiency. However, deploying MoE models across heterogeneous end-cloud environments poses new challenges in expert scheduling, communication overhead, and resource heterogeneity. In this paper, we propose EC2MoE, an adaptive framework for scalable MoE inference via end-cloud pipeline collaboration. First, we design a hardware-aware lightweight group gate network that enhances expert selection and computational efficiency. By incorporating a hardware-aware local expert selection mechanism, the system adaptively filters candidate experts based on real-time device profiles. A lightweight group gate module then integrates local and global gating outputs to achieve high-quality expert routing with minimal overhead. Second, we develop a pipeline optimization mechanism based on endcloud collaboration to accelerate MoE inference. This includes an encoder-decoder structure based on low-rank compression, which reduces transmission and computation costs. And a route-aware heuristic pipeline scheduling algorithm that dynamically allocates inference stages across devices according to workload and network topology. Extensive experiments show that EC2MoE can increase throughput by 2.2x to 5.1x and reduce end-to-end latency by 53% to 67% while maintaining high accuracy compared to state-of-the-art methods. It also maintains good scalability under dynamic load and network environments.

cs.DC↗

arXiv2025

Designing optimal subsidy schemes and recycling plans for sustainable treatment of construction and demolition waste

More than 10 billion tons of construction and demolition waste (CW) are generated globally each year, exerting a significant impact on the environment. In the CW recycling process, the government and the carrier are the two primary stakeholders. The carrier is responsible for transporting CW from production sites to backfill sites or processing facilities, with a primary focus on transport efficiency and revenue. Meanwhile, the government aims to minimize pollution from the recycling system, which is influenced by transport modes, shipment distances, and the processing methods used for CW. This paper develops a bi-objective, bi-level optimization model to address these challenges. The upper-level model is a linear programming model that optimizes the government's subsidy scheme, while the lower-level model is a minimum-cost flow model that optimizes the carrier's recycling plan. A hybrid heuristic solution method is proposed to tackle the problem's complexity. A case study in Chengdu, China, demonstrates the computational efficiency of the model and its small solution gap. With an optimized subsidy scheme and recycling plan, pollution can be reduced by over 29.29% through a relatively small investment in subsidies.

math.OC↗

arXiv2025

CDIO: Cross-Domain Inference Optimization with Resource Preference Prediction for Edge-Cloud Collaboration

Currently, massive video tasks are processed by edge-cloud collaboration. However, the diversity of task requirements and the dynamics of resources pose great challenges to efficient inference, resulting in many wasted resources. In this paper, we present CDIO, a cross-domain inference optimization framework designed for edge-cloud collaboration. For diverse input tasks, CDIO can predict resource preference types by analyzing spatial complexity and processing requirements of the task. Subsequently, a cross-domain collaborative optimization algorithm is employed to guide resource allocation in the edge-cloud system. By ensuring that each task is matched with the ideal servers, the edge-cloud system can achieve higher efficiency inference. The evaluation results on public datasets demonstrate that CDIO can effectively meet the accuracy and delay requirements for task processing. Compared to state-of-the-art edge-cloud solutions, CDIO achieves a computing and bandwidth consumption reduction of 20%-40%. And it can reduce energy consumption by more than 40%.

cs.MM↗

arXiv2024

Integrating optimal ridesharing matching into multimodal traffic model: Implications for policy and sustainable transport system

Integrating ridesharing matching explicitly into multimodal traffic models is crucial for accurately assessing the impacts of multimodal transport (MT) on urban economic and environmental aspects. This paper integrates an optimal ridesharing matching method into a path-based deterministic day-to-day traffic assignment framework, considers match cancellations, and captures the interactions between various modes on the road. The model incorporates five traffic modes (solo driving, ridesharing as a driver, ridesharing as a passenger, bus travel, and metro travel) and two groups of travelers based on their ownership status. Its steady state is determined through numerical experiments. The sensitivity analyses reveal that the MT system's performance varies with changes in ownership, bus fare, and ridesharing fare, demonstrating diverse impacts on mode split, travel cost, and emissions across different groups, road links, and regions. Our findings suggest that vehicle restrictions and pricing strategies have both benefits and drawbacks in managing MT system, emphasizing the need for careful consideration of trade-offs and social equity implications in policy-making and implementation. This study not only enhances the theoretical understanding of MT system but also provides valuable support for urban transportation policy-making aimed at achieving efficient, sustainable, and socially equitable transport systems.

physics.soc-ph↗

arXiv2024

Enhancing the sensing power of bike-sharing system for urban environment

The development of smart cities requires innovative sensing solutions for efficient and low-cost urban environment monitoring. Bike-sharing systems, with their wide coverage, flexible mobility, and dense urban distribution, present a promising platform for pervasive sensing. At a relative early stage, research on bike-based sensing focuses on the application of data collected via passive sensing, without consideration of the optimization of data collection through sensor deployment or vehicle scheduling. To address this gap, this study integrates a binomial probability model with a mixed-integer linear programming model to optimize sensor allocation across bike stands. Additionally, an active scheduling strategy guides user bike selection to enhance the efficacy of data collection. A case study in Manhattan validates the proposed strategy, showing that equipping sensors on just 1\% of the bikes covers approximately 70\% of road segments in a day, highlighting the significant potential of bike-sharing systems for urban sensing.

math.OC↗

arXiv2024

Fine-gained air quality inference based on low-quality sensing data using self-supervised learning

Fine-grained air quality (AQ) mapping is made possible by the proliferation of cheap AQ micro-stations (MSs). However, their measurements are often inaccurate and sensitive to local disturbances, in contrast to standardized stations (SSs) that provide accurate readings but fall short in number. To simultaneously address the issues of low data quality (MSs) and high label sparsity (SSs), a multi-task spatio-temporal network (MTSTN) is proposed, which employs self-supervised learning to utilize massive unlabeled data, aided by seasonal and trend decomposition of MS data offering reliable information as features. The MTSTN is applied to infer NO$_2$, O$_3$ and PM$_{2.5}$ concentrations in a 250 km$^2$ area in Chengdu, China, at a resolution of 500m$\times$500m$\times$1hr. Data from 55 SSs and 323 MSs were used, along with meteorological, traffic, geographic and timestamp data as features. The MTSTN excels in accuracy compared to several benchmarks, and its performance is greatly enhanced by utilizing low-quality MS data. A series of ablation and pressure tests demonstrate the results' robustness and interpretability, showcasing the MTSTN's practical value for accurate and affordable AQ inference.

cs.LG↗

arXiv2024

Unsteady aerodynamic prediction using limited samples based on transfer learning

In this study, a method for predicting unsteady aerodynamic forces under different initial conditions using a limited number of samples based on transfer learning is proposed, aiming to avoid the need for large-scale high-fidelity aerodynamic simulations. First, a large number of training samples are acquired through high-fidelity simulation under the initial condition for the baseline, followed by the establishment of a pre-trained network as the source model using a long short-term memory (LSTM) network. When unsteady aerodynamic forces are predicted under the new initial conditions, a limited number of training samples are collected by high-fidelity simulations. Then, the parameters of the source model are transferred to the new prediction model, which is further fine-tuned and trained with limited samples. The new prediction model can be used to predict the unsteady aerodynamic forces of the entire process under the new initial conditions. The proposed method is validated by predicting the aerodynamic forces of free flight of a high-spinning projectile with a large extension of initial angular velocity and pitch angle. The results indicatethat the proposed method can predict unsteady aerodynamic forces under different initial conditions using 1/3 of the sample size of the source model. Compared with direct modeling using the LSTM networks, the proposed method shows improved accuracy and efficiency.

physics.flu-dyn↗

arXiv2024

PerLLM: Personalized Inference Scheduling with Edge-Cloud Collaboration for Diverse LLM Services

With the rapid growth in the number of large language model (LLM) users, it is difficult for bandwidth-constrained cloud servers to simultaneously process massive LLM services in real-time. Recently, edge-cloud infrastructures have been used to improve the processing efficiency of large-scale LLM services. However, the diversity of task requirements and the dynamics of resources pose great challenges to inference scheduling, leading to the wastage of many resources. In this paper, we present PerLLM, a personalized inference scheduling framework with edge-cloud collaboration designed for diverse LLM services. For the complexity of multiple constraints and the decision-making process of edge-cloud collaboration, we integrate the upper confidence bound algorithm based on the constraint satisfaction mechanism in PerLLM. For diverse LLM services, PerLLM can optimize service scheduling and resource allocation solutions within the edge-cloud infrastructure to meet processing time requirements while minimizing energy costs. Experimental results from different model deployments show that PerLLM can effectively meet the processing time requirements of personalized services. Compared to other methods, PerLLM achieves 2.2x, 2.1x, and 1.6x throughput and reduces the energy cost by more than 50%.

cs.DC↗

arXiv2024

Exploring the sensing power of mixed vehicle fleets

Vehicle-based mobile sensing, also known as drive-by sensing, efficiently surveys urban environments at low costs by leveraging the mobility of urban vehicles. While recent studies have focused on drive-by sensing for fleets of a single type, our work explores the sensing power and cost-effectiveness of a mixed fleet that consists of vehicles with distinct and complementary mobility patterns. We formulate the drive-by sensing coverage (DSC) problem, proposing a method to quantify sensing utility and an optimization procedure that determines fleet composition, sensor allocation, and vehicle routing for a given budget. Our air quality sensing case study in Longquanyi District (Chengdu, China) demonstrates that using a mixed fleet enhances sensing utilities and achieves close approximations to the target sensing distribution at a lower cost. Generalizing these insights to two additional real-world networks, our regression analysis uncovers key factors influencing the sensing power of mixed fleets. This research provides quantitative and managerial insights into drive-by sensing, showcasing a positive externality of urban transport activities.

math.OC↗

arXiv2024

Route planning of mobile sensing fleets for repeatable visits

Vehicle-based mobile sensing is an emerging data collection paradigm that leverages vehicle mobilities to scan a city at low costs. Certain urban sensing scenarios require dedicated vehicles for highly targeted monitoring, such as volatile organic compounds (VOCs, a type of air pollutant) sensing, road surface monitoring, and accident site investigation. A hallmark of these scenarios is that the points of interest (POIs) need to be repeatedly visited by a set of agents, whose routes should provide sufficient sensing coverage with coordinated overlap at certain important POIs. For these applications, this paper presents the open team orienteering problem with repeatable visits (OTOP-RV). The adaptive large neighborhood search (ALNS) algorithm is tailored to solve the OTOP-RV considering specific features of the problem. Test results on randomly generated datasets show that: (1) For small cases, the ALNS matches Gurobi in terms of optimality but with shorter computational times; (2) For large cases, the ALNS significantly outperforms the greedy algorithm (by 9.7% to 25.4%), and a heuristic based on sequential orienteering problems (by 6%). Finally, a real-world case study of VOCs sensing is presented, which highlights the unique applicability of the OTOP-RV to such specific sensing tasks, as well as the effectiveness of the proposed algorithms in optimizing the sensing utilities.

math.OC↗

arXiv2024

The Share-a-Ride Problem with mixed ride-hailing and logistic vehicles

This study explores the potential of using ride-hailing vehicles (RVs) for integrated passenger and freight transport based on shared mobility. In this crowd-sourced mode, ride-hailing platforms can profit from parcel delivery services, and logistics companies can reduce operational costs by utilizing the capacities of RVs. The Share-a-Ride problem with ride-hailing and logistic vehicles (SARP-RL) determines the number of logistic vehicles (LVs) and the assignment of passenger/parcel requests to RVs and LVs, aiming at maximizing the total RV profits and minimizing logistic costs. An exact solution framework is proposed by (1) generating a feasible trip that serves a given set of requests at maximal profits; (2) generating all feasible trips for the entire set of passenger and parcel requests via an efficient enumeration method; and (3) finding all Pareto-optimal solutions of the bi-objective problem via an $\varepsilon$-constraint method. Not only is the proposed method exact, it also converts the NP-hard problem to a simple vehicle-trip matching problem. More importantly, the total computational time can be compressed to an arbitrary degree via straightforward parallelization. A case study of the Manhattan network demonstrates the solution characteristics of SARP-RL. The results indicate that: (i) Coordinating RV and LV operations to serve passenger and parcel requests (SARP-RL) can simultaneously reduce logistic costs and increase RV profits. (ii) Key factors influencing the performance of SARP-RL include the RV fleet size, spatial distribution of parcel requests, passenger/parcel request ratio, and unit price of transport service, which are quantitatively analyzed to offer managerial insights for real-world implementation.

math.OC↗

arXiv2023

A survey of urban drive-by sensing: An optimization perspective

Pervasive and mobile sensing is an integral part of smart transport and smart city applications. Vehicle-based mobile sensing, or drive-by sensing (DS), is gaining popularity in both academic research and field practice. The DS paradigm has an inherent transport component, as the spatial-temporal distribution of the sensors are closely related to the mobility patterns of their hosts, which may include third-party (e.g. taxis, buses) or for-hire (e.g. unmanned aerial vehicles and dedicated vehicles) vehicles. It is therefore essential to understand, assess and optimize the sensing power of vehicle fleets under a wide range of urban sensing scenarios. To this end, this paper offers an optimization-oriented summary of recent literature by presenting a four-step discussion, namely (1) quantifying the sensing quality (objective); (2) assessing the sensing power of various fleets (strategic); (3) sensor deployment (strategic/tactical); and (4) vehicle maneuvers (tactical/operational). By compiling research findings and practical insights in this way, this review article not only highlights the optimization aspect of drive-by sensing, but also serves as a practical guide for configuring and deploying vehicle-based urban sensing systems.

cs.NI↗