Multimodal large language models (MLLMs) have shown remarkable performance across various visual understanding tasks. However, most existing MLLMs still lack image detail perception, limiting their effectiveness in tasks that require detailed visual information. In this paper, we introduce Percept-DriveLM, a novel MLLM designed to tackle the fine-grained perception challenges in autonomous driving tasks. At the core of our model is the Visual Fusion Module, which integrates several innovative components: a dynamic resolution mechanism that combines both high and low resolution features, and an RoI conditional mechanism to incorporate object/region-level features identified by offline detectors, further refining the model's fine-grained perception abilities. Trained in a two-stage process, our model demonstrates exceptional performance, outperforming existing MLLMs with comparable parameter sizes and excelling in both autonomous driving perception and general vision-language tasks. The effectiveness of our approach is validated through extensive empirical studies. Code will be available at https://github.com/DebuggerSunfz/PerceptDriveLM.

In audio-visual joint analysis, generalized zero-shot learning (GZSL) aims to recognize unseen categories by aligning semantic information across modalities. However, significant challenges arise from temporal and semantic discrepancies between audio and video modalities. Traditional methods typically depend on static semantic embeddings, thereby overlooking the dynamic nature of these modalities and often resulting in modality imbalance. We propose the Dual-Factor Compensatory Clustering Network (DFCNet), an end-to-end framework for dynamic fusion and optimized alignment of heterogeneous modal information to address these limitations. DFCNet employs a multi-branch architecture, integrating a parallel multi-layer perceptron (MLP) for semantic modeling and a Bidirectional Long Short-Term Memory (BiLSTM) network for capturing temporal consistency. The Compensatory Fusion Block (CFB) employs tensor decomposition to facilitate cross-modal coupling, where the low-rank representation decomposition aligns intra-modal feature distributions. Additionally, we introduce the Dual-Factor Clustering Multi-objective Optimization Framework (DCMOF), which ensures gradient equilibrium and adaptively adjusts modality contribution weights to strengthen robust cross-modal alignment. Designed as a pluggable module, DFCNet can be seamlessly integrated into existing base models. Experimental results demonstrate that our framework significantly improves the performance of the Audio-Visual Cross-Modal Alignment (AVCA) model across multiple GZSL datasets. Ablation studies further validate the critical role of CFB in cross-modal alignment and highlight the significance of DCMOF in optimizing modality coordination. The code is available at https://github.com/ATKEROM/DFCNet.

Bandwidth resources are crucial for the internal communication and task execution of multi-UAV systems. However, traditional allocation models for multi-UAV system are inefficient and result in low utility of UAVs. To overcome this challenge, this paper proposes a novel bandwidth-oriented task allocation model that accounts for the supply-demand relationship between UAVs and tasks for bandwidth resources. To solve this model, this paper presents a Double Auction-based Task Allocation Mechanism, which can achieve distributed allocation of tasks and bandwidth among multi-UAV. The main steps of this mechanism are as follows: (1) each UAV privately submits its bandwidth demand and bids for each task; (2) each task determines the bandwidth price and initial allocation based on the bids, and announces them to each UAV; (3) each UAV selects the task that maximizes its utility, and obtains the final allocation. Furthermore, this paper proves that this mechanism satisfies desirable economic properties such as individual rationality, budget balance, and incentive compatibility. Finally, experimental results show that the proposed mechanism exhibits excellent performance in terms of UAV and task utility, as well as the rate of bandwidth allocation.

Intelligent vehicle driving systems aim to control the driving behavior of a vehicle in real time without human intervention by perceiving and monitoring the surrounding environment. Describing images of traffic scenes automatically, which is one of the key problems of intelligent vehicle driving technology, has drawn attention since its inception. In recent years, a variety of automatic image description technologies have been proposed, among which the attention-based encoder-decoder framework achieved good results. In this paper we will discuss the fusing of a variety of information from multiple aspects of the images of traffic scenes. First, we will introduce visual attention, text attention and image topics attention which generates the weighted visual features, the attentive text information and the global image topics information respectively. We will then propose an adaptive two-stage merging network based on an encoder-decoder framework, which can fully integrate the three kinds of information in two stages, while automatically calculating the proportions of the information at each time step. Numerous experiments conducted on COCO2014 and Flickr30K datasets have demonstrated the effectiveness and advantages of the proposed method.

Abstract The flexibility and high reliability of cloud computing have generated a tremendous amount of economic benefits. With the growing demand for cloud resources, the task about how to making users' demand adaptive to the supply of cloud resources during peak and nonpeak hours becomes extremely challenging. However, most of the existing mechanisms mainly focus on the allocation of cloud resources, and ignore the balance between supply and demand of cloud resources. To improve the situation, this paper presents a combinatorial auction model for the cloud resource allocation and consumption shifting problem, and then puts forward a truthful auction mechanism with two different price curve functions, that is, staircase and continuous price curve. The proposed mechanism can effectively allocate and shift cloud resources to alleviate the pressure of cloud infrastructure during peak hours. Both the theoretical analysis and simulation results show that the proposed mechanism guarantees desired properties including individual rationality, budget balance, truthfulness and computational efficiency. In addition, extensive simulation results also show that continuous price curve outperforms stair case price curve in terms of system efficiency and balance efficiency.

As technology evolves and electricity demand rises, more and more research focus on the efficient electricity allocation mechanisms so as to make consumer demand adaptive to the supply of electricity at all times. In this paper, we formulate the problem of electricity allocation as a novel combinatorial auction model, and then put forward a directly applicable mechanisms. It is proven that the proposed mechanism is equipped with some useful economic properties and computational traceability. Our works offer potential avenues for the stduy about efficient electricity allocation methods in smart grid.

In the model of partnership formation, every agent is in hot pursuit of maximizing his own utility. Doft Talman and Zaifu Yang have revealed the necessary condition of the existence of partnership equilibrium, however, the methods of how to compute the equilibrium are not given. The paper presents a method to compute the partnership utility equilibrium based on socially optimal assignment. Firstly, the partnership groups with social optimality are exhausted in the all possible cooperative pairs, and then under the condition of the existence of equilibrium, an algorithm is given to get the Equilibrium Utility Vector so as to approach the equilibrium status after multiple iterations. This solution to computing partnership payoff equilibrium can be obtained in O(n2) time. At last, to verify the method given by this paper, an example of municipal water supply planning is given to illustrate the correctness and effectiveness.

This paper first reviews covert timing channels with network packet frequencies as information carriers. Then, based on the study of communication and statistical models, it proposes a method to detect an enhanced covert timing channel and its use of carrier frequencies. With the help of MATLAB for simulation, several experiments have been conducted for the verification of the proposed method.