Visual anomaly detection is limited by the lack of sufficient anomaly data. While existing anomaly synthesis methods have made remarkable progress, achieving both realism and diversity in synthesis remains a major obstacle. To address this, we propose AnomalyPainter, a novel framework that breaks the diversity-realism trade-off dilemma through synergizing Vision Language Large Model (VLLM), Latent Diffusion Model (LDM), and our newly introduced texture library Tex-9K. Tex-9K is a professional texture library containing 75 categories and 8792 texture assets crafted for diverse anomaly synthesis. Leveraging VLLM's general knowledge, reasonable anomaly text descriptions are generated for each industrial object and matched with relevant diverse textures from Tex-9K. These textures then guide the LDM via ControlNet to paint on normal images. Furthermore, we introduce Texture-Aware Latent Init to stabilize the natural-image-trained ControlNet for industrial images. Extensive experiments show that AnomalyPainter outperforms existing methods in realism, diversity, and generalization, achieving superior downstream performance.

Large language models increasingly expose reasoning traces, yet their underlying cognitive structure and steps remain difficult to identify and analyze beyond surface-level statistics.We adopt Schoenfeld's Episode Theory as an inductive, intermediate-scale lens and introduce ThinkARM (Anatomy of Reasoning in Models), a scalable framework that explicitly abstracts reasoning traces into functional reasoning steps such as Analysis, Explore, Implement, Verify, etc.When applied to mathematical problem solving by diverse models, this abstraction reveals reproducible thinking dynamics and structural differences between reasoning and non-reasoning models, which are not apparent from token-level views.We further present two diagnostic case studies showing that exploration functions as a critical branching step associated with correctness, and that efficiency-oriented methods selectively suppress evaluative feedback steps rather than uniformly shortening responses.Together, our results demonstrate that episode-level representations make reasoning steps explicit, enabling systematic analysis of how reasoning is structured, stabilized, and altered in modern language models.

Ming Li, Nan Zhang, Chenrui Fan, Hong Jiao, Yanbin Fu, Sydney Peters, Qingshu Xu, Robert Lissitz, Tianyi Zhou. Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing. 2025.

The rapid advancement of large language models (LLMs) has significantly enhanced their reasoning abilities, enabling increasingly complex tasks. However, these capabilities often diminish in smaller, more computationally efficient models like GPT-2. Recent research shows that reasoning distillation can help small models acquire reasoning capabilities, but most existing methods focus primarily on improving teacher-generated reasoning paths. Our observations reveal that small models can generate high-quality reasoning paths during sampling, even without chain-of-thought prompting, though these paths are often latent due to their low probability under standard decoding strategies. To address this, we propose Self-Enhanced Reasoning Training (SERT), which activates and leverages latent reasoning capabilities in small models through self-training on filtered, self-generated reasoning paths under zero-shot conditions. Experiments using OpenAI’s GPT-3.5 as the teacher model and GPT-2 models as the student models demonstrate that SERT enhances the reasoning abilities of small models, improving their performance in reasoning distillation.

Due to the challenges of manually collecting accurate editing data, existing datasets are typically constructed using various automated methods, leading to noisy supervision signals caused by the mismatch between editing instructions and original-edited image pairs. Recent efforts attempt to improve editing models through generating higher-quality edited images, pre-training on recognition tasks, or introducing vision-language models (VLMs) but fail to resolve this fundamental issue. In this paper, we offer a novel solution by constructing more effective editing instructions for given image pairs. This includes rectifying the editing instructions to better align with the original-edited image pairs and using contrastive editing instructions to further enhance their effectiveness. Specifically, we find that editing models exhibit specific generation attributes at different inference steps, independent of the text. Based on these prior attributes, we define a unified guide for VLMs to rectify editing instructions. However, there are some challenging editing scenarios that cannot be resolved solely with rectified instructions. To this end, we further construct contrastive supervision signals with positive and negative instructions and introduce them into the model training using triplet loss, thereby further facilitating supervision effectiveness. Our method does not require the VLM modules or pre-training tasks used in previous work, offering a more direct and efficient way to provide better supervision signals, and providing a novel, simple, and effective solution for instruction-based image editing. Results on multiple benchmarks demonstrate that our method significantly outperforms existing approaches. Compared with previous SOTA SmartEdit, we achieve 9.19% improvements on the Real-Edit benchmark with 30x less training data and 13x smaller model size.

Evaluating large language models (LLMs) is costly: it requires the generation and examination of LLM outputs on a large-scale benchmark of various tasks. This paper investigates how to efficiently reduce the tasks used to benchmark LLMs without affecting the evaluation quality. Our study reveals that task transferability and relevance provide critical information to identify the most representative subset of tasks via optimizing a facility location function. We propose a practically efficient metric for estimating the transferability between two tasks via in-context learning (ICL). By analyzing the pairwise transferability, we can reduce tasks in a modern LLM benchmark (e.g., MMLU or FLAN) to 5\% while inducing only a $<4$\% difference to the evaluation on the original benchmark. Compared to prior works, our method is training-free, gradient-free, and highly efficient requiring ICL only.

Video generation techniques have made remarkable progress, promising to be the foundation of interactive world exploration.However, existing video generation datasets are not well-suited for world exploration training as they suffer from some limitations: limited locations, short duration, static scenes, and a lack of annotations about exploration and the world.In this paper, we introduce Sekai (meaning "world" in Japanese), a high-quality first-person view worldwide video dataset with rich annotations for world exploration. It consists of over 5,000 hours of walking or drone view (FPV and UVA) videos from over 100 countries and regions across 750 cities. We develop an efficient and effective toolbox to collect, pre-process and annotate videos with location, scene, weather, crowd density, captions, and camera trajectories.Comprehensive analyses and experiments demonstrate the dataset’s scale, diversity, annotation quality, and effectiveness for training video generation models.We believe Sekai will benefit the area of video generation and world exploration, and motivate valuable applications.

Color plays an important role in human perception and usually provides critical clues in visual reasoning. However, it is unclear whether and how vision-language models (VLMs) can perceive, understand, and leverage color as humans.This paper introduces ColorBench, an innovative benchmark meticulously crafted to assess the capabilities of VLMs in color understanding, including color perception, reasoning, and robustness. By curating a suite of diverse test scenarios, with grounding in real applications, ColorBench evaluates how these models perceive colors, infer meanings from color-based cues, and maintain consistent performance under varying color transformations. Through an extensive evaluation of 32 VLMs with varying language models and vision encoders, our paper reveals some undiscovered findings: (i) The scaling law (larger models are better) still holds on ColorBench, while the language model plays a more important role than the vision encoder. (ii) However, the performance gaps across models are relatively small, indicating that color understanding has been largely neglected by existing VLMs. (iii) CoT reasoning improves color understanding accuracies and robustness, though they are vision-centric tasks. (iv) Color clues are indeed leveraged by VLMs on ColorBench but they can also mislead models in some tasks.These findings highlight the critical limitations of current VLMs and underscore the need to enhance color comprehension. Our ColorBench can serve as a foundational tool for advancing the study of human-level color understanding of multimodal AI.

To enhance controllability in text-to-image generation, ControlNet introduces image-based control signals, while ControlNet++ improves pixel-level cycle consistency between generated images and the input control signal. To avoid the prohibitive cost of back-propagating through the sampling process, ControlNet++ optimizes only low-noise timesteps (e.g., $t < 200$) using a single-step approximation, which not only ignores the contribution of high-noise timesteps but also introduces additional approximation errors. A straightforward alternative for optimizing controllability across all timesteps is Direct Preference Optimization (DPO), a fine-tuning method that increases model preference for more controllable images ($I^{w}$) over less controllable ones ($I^{l}$). However, due to uncertainty in generative models, it is difficult to ensure that win--lose image pairs differ only in controllability while keeping other factors, such as image quality, fixed. To address this, we propose performing preference learning over control conditions rather than generated images. Specifically, we construct winning and losing control signals, $\mathbf{c}^{w}$ and $\mathbf{c}^{l}$, and train the model to prefer $\mathbf{c}^{w}$. This method, which we term \textit{Condition Preference Optimization} (CPO), eliminates confounding factors and yields a low-variance training objective. Our approach theoretically exhibits lower contrastive loss variance than DPO and empirically achieves superior results. Moreover, CPO requires less computation and storage for dataset curation. Extensive experiments show that CPO significantly improves controllability over the state-of-the-art ControlNet++ across multiple control types: over $10\%$ error rate reduction in segmentation, $70$--$80\%$ in human pose, and consistent $2$--$5\%$ reductions in edge and depth maps. The error rate is defined as the difference between the evaluated controllability and the oracle results. Our project is available \textcolor{blue}{\href{https://zonglinl.github.io/CPO_page}{here}}.

Ming Li, Yong Zhang, Shwai He, Zhitao Li, Hongyu Zhao, Jianzong Wang, Ning Cheng, Tianyi Zhou. Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2024.

Large Language Models (LLMs) have shown significant promise in various applications, including zero-shot and few-shot learning. However, their performance can be hampered by inherent biases. Instead of traditionally sought methods that aim to minimize or correct these biases, this study introduces a novel methodology named "bias-kNN". This approach capitalizes on the biased outputs, harnessing them as primary features for kNN and supplementing with gold labels. Our comprehensive evaluations, spanning diverse domain text classification datasets and different GPT-2 model sizes, indicate the adaptability and efficacy of the "bias-kNN" method. Remarkably, this approach not only outperforms conventional in-context learning in few-shot scenarios but also demonstrates robustness across a spectrum of samples, templates and verbalizers. This study, therefore, presents a unique perspective on harnessing biases, transforming them into assets for enhanced model performance.

Recent work in Video Frame Interpolation (VFI) tries to formulate VFI as a diffusion-based conditional image generation problem, synthesizing the intermediate frame given a random noise and neighboring frames. Due to the relatively high resolution of videos, Latent Diffusion Models (LDMs) are employed to run diffusion models in latent space efficiently. Such a formulation poses a crucial challenge: VFI expects that the output is deterministically equal to the ground truth intermediate frame, but LDMs randomly generate a diverse set of different images when the model runs multiple times. The diversity is due to the large cumulative variance (variance accumulated at each generation step) of generated latent representations in LDMs, making the sampling trajectory random. To address this problem, we propose our unique solution: Frame Interpolation with Consecutive Brownian Bridge Diffusion. Specifically, we propose consecutive Brownian Bridge diffusion that takes a deterministic initial value as input, resulting in a much smaller cumulative variance of generated latent representations. Our experiments suggest that our method can improve together with the improvement of the autoencoder and achieve state-of-the-art performance in VFI, leaving strong potential for further enhancement. Our code is available at https://github.com/ZonglinL/ConsecutiveBrownianBridge.

Ming Li, Yong Zhang, Zhitao Li, Jiuhai Chen, Lichang Chen, Ning Cheng, Jianzong Wang, Tianyi Zhou, Jing Xiao. Proceedings of the 2024 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies (Volume 1: Long Papers). 2024.

Recently, open-vocabulary learning has emerged to accomplish segmentation for arbitrary categories of text-based descriptions, which popularizes the segmentation system to more general-purpose application scenarios. However, existing methods devote to designing specialized architectures or parameters for specific segmentation tasks. These customized design paradigms lead to fragmentation between various segmentation tasks, thus hindering the uniformity of segmentation models. Hence in this paper, we propose FreeSeg, a generic framework to accomplish Unified, Universal and Open-Vocabulary Image Segmentation. FreeSeg optimizes an all-in-one network via one-shot training and employs the same architecture and parameters to handle diverse segmentation tasks seamlessly in the inference procedure. Additionally, adaptive prompt learning facilitates the unified model to capture task-aware and category-sensitive concepts, improving model robustness in multi-task and varied scenarios. Extensive experimental results demonstrate that FreeSeg establishes new state-of-the-art results in performance and generalization on three segmentation tasks, which outperforms the best task-specific architectures by a large margin: 5.5% mIoU on semantic segmentation, 17.6% mAP on instance segmentation, 20.1% PQ on panoptic segmentation for the unseen class on COCO. Project page: https://FreeSeg.github.io.

The Retrieval Question Answering (ReQA) task employs the retrieval-augmented framework, composed of a retriever and generator.The generator formulates the answer based on the documents retrieved by the retriever.Incorporating Large Language Models (LLMs) as generators is beneficial due to their advanced QA capabilities, but they are typically too large to be fine-tuned with budget constraints while some of them are only accessible via APIs.To tackle this issue and further improve ReQA performance, we propose a trainable Pluggable Reward-Driven Contextual Adapter (PRCA), keeping the generator as a black box.Positioned between the retriever and generator in a Pluggable manner, PRCA refines the retrieved information by operating in a tokenautoregressive strategy via maximizing rewards of the reinforcement learning phase.Our experiments validate PRCA's effectiveness in enhancing ReQA performance on three datasets by up to 20% improvement to fit black-box LLMs into existing frameworks, demonstrating its considerable potential in the LLMs era.

The paradigm of large-scale pre-training followed by downstream fine-tuning has been widely employed in various object detection algorithms. In this paper, we reveal discrepancies in data, model, and task between the pre-training and fine-tuning procedure in existing practices, which implicitly limit the detector’s performance, generalization ability, and convergence speed. To this end, we propose AlignDet, a unified pre-training framework that can be adapted to various existing detectors to alleviate the discrepancies. AlignDet decouples the pre-training process into two stages, i.e., image-domain and box-domain pre-training. The image-domain pre-training optimizes the detection backbone to capture holistic visual abstraction, and box-domain pre-training learns instance-level semantics and task-aware concepts to initialize the parts out of the backbone. By incorporating the self-supervised pretrained backbones, we can pre-train all modules for various detectors in an unsupervised paradigm. As depicted in Figure 1, extensive experiments demonstrate that AlignDet can achieve significant improvements across diverse protocols, such as ${\color{Green}\text{detection algorithms}}, {\color{Blue}\text{model backbones}}, {\color{Red}\text{data settings}}$, and ${\color{SkyBlue}\text{training schedules}}$. For example, AlignDet improves FCOS by 5.3 mAP, RetinaNet by 2.1 mAP, Faster R-CNN by 3.3 mAP, and DETR by 2.3 mAP under fewer epochs.

Local visual and long-range contextual features yield two complementary cues for human reading text in natural scene. Existing scene text recognition methods mainly extract local features at a low level and then model long-range dependencies at a high level, this sequential pipeline may be sub-optimal to construct complete and effective representation. Except for high-level features, long-range contextual relation is of importance in low-level features as well since it can help separate different characters based on the intervals between characters and thus enhance the character features. To address this issue, we develop a dual relation module to extract complementary features in a parallel manner for scene text recognition, which consists of a local visual branch and a long-range contextual branch. The local visual branch employs a topological-aware operation to model intra-character characteristic and extract discriminative features of different characters. Meanwhile, the long-range contextual branch utilizes a simple but effective strategy to incorporate inter-character relations into feature maps. Our dual relation module is a plug-and-play block which can be easily incorporated into modern deep architectures. Experimental results demonstrate that our methods achieved top performance on several standard benchmarks. Code and models will become publicly available in the future.

Curved scene text recognition is a challenging task in multimedia society due to large shape and texture variance. Previous methods address this challenge by extracting and rectifying text line with equidistantly sampling, which ignore character level information and lead to distorted characters. To address this issue, this paper proposes a Character-Aware Sampling and Rectification (CASR) module, which rectifies irregular text instance according to the location and orientation information of each individual character. Specifically, CASR regards each character as a basic unit and predicts the character-level attributes for sampling and rectification. Our module not only exploits detailed character information to obtain better rectification of text line, but also employs character-level supervision in training process. In addition, CASR provides a plug-and-play module which can be easily incorporated to existing text recognition pipeline. Extensive experiments on several benchmarks demonstrate that our method obtains more accurate rectified text instances and achieves promising performance. We will release our code and models in the future.

Segmentation-based text detectors are flexible to capture arbitrary-shaped text regions. Due to large geometry variance, it is necessary to construct effective and robust representations to identify text regions with various shapes and scales. In this paper, we focus on designing effective multi-scale contextual features for locating text instances. Specially, we develop a Region Context Module (RCM) to summarize the semantic response and adaptively extract text-region-aware information in a limited local area. To construct complementary multi-scale contextual representations, multiple RCM branches with different scales are employed and integrated via Progressive Fusion Module (PFM). Our proposed RCM and PFM serve as the plug-and-play modules which can be incorporated into existing scene text detection platforms to further boost detection performance. Extensive experiments show that our methods achieve state-of-the-art performances on Total-Text, SCUT-CTW1500 and MSRA-TD500 datasets. The code with models will become publicly available athttps://github.com/wqtwjt1996/RP-Text.

Recently, mobile devices have become increasingly affordable, reliable and pervasive in contemporary society. Meanwhile, cloud computing is playing an increasingly role in people's daily life. Together with these two promising technologies, Mobile Cloud Computing (MCC) has been introduced as a potential technology in future. However, since the mobility of cloud users in MCC, service availability becomes one of the significant challenges which has to be addressed. This paper introduces an efficient social based data forwarding mechanism to improve the service availability of MCC under Ad Hoc network environment. Users in MCC are treated as entity nodes with social characteristics as well as can be grouped into different communities. Thus mobile user can connect to the cloud through neighbouring nodes instead of direct link to clouds. By using this mechanism, the service availability will be improved with the improvement of data forwarding efficiency. We performed simulation of our technique on CloudSim. The empirical results show that the proposed mechanism is promising in enabling the mobile nodes to be efficiently searched.

Cloud computing is experiencing phenomenal growth and there are now many vendors offering their cloud services. In cloud computing, cloud providers cooperate together to offer their computing resource as a utility and software as a service to customers. The demands and the price of cloud service should be negotiated between providers and users based on the Service Level Agreement (SLA). In order to help cloud providers achieving an agreeable price for their services and maximizing the benefits of both cloud providers and clients, this paper proposes a cloud pricing system consisting of hierarchical system, M/M/c queuing model and pricing model. Simulation results verify the efficiency of our proposed system.

Human-associated delay-tolerant networks (HDTNs) are new networks where mobile devices are associated with humans and can be viewed from multiple dimensions including geographic and social aspects. The combination of these different dimensions enables us to comprehend delay-tolerant networks and consequently use this multidimensional information to improve overall network efficiency. Alongside the geographic dimension of the network, which is concerned with geographic topology of routing, social dimensions such as social characters can be used to guide the routing message to improve not only the routing efficiency for individual nodes, but also efficiency for the entire network. We propose a multidimensional routing protocol (M-Dimension) for the human-associated delay-tolerant networks which uses local information derived from multiple dimensions to identify a mobile node more accurately. The importance of each dimension has been measured by the weight function and it is used to calculate the best route. The greedy routing strategy is applied to select an intermediary node to forward message. We compare M-Dimension to the existing benchmark routing protocols via MIT reality Data Set and INFOCOM 2006 Data Set, which are real human-associated mobile network trace files. The results of our simulations show that M-Dimension significantly increases the average success ratio with a competitive end-to-end delay when compared with other multicast DTNs routing protocols.

Online social networks have not only become a point of aggregation and exchange of information, they have so radically rooted into our everyday behaviors that they have become the target of important network attacks. We have seen an increasing trend in Sybil based activity, such as in personification, fake profiling and attempts to maliciously subvert the community stability in order to illegally create benefits for some individuals, such as online voting, and also from more classic informatics assaults using specifically mutated worms. Not only these attacks, in the latest months, we have seen an increase in spam activities on social networks such as Facebook and RenRen, and most importantly, the first attempts at propagating worms within these communities. What differentiates these attacks from normal network attacks, is that compared to anonymous and stealthy activities, or by commonly untrusted emails, social networks regain the ability to propagate within consentient users, who willingly accept to partake. In this paper, we will demonstrate the effects of influential nodes against non-influential nodes through in simulated scenarios and provide an overview and analysis of the outcomes.

Human associated delay-tolerant networks (HDTNs) are new networks where mobile devices are associated with humans and demonstrate social-related communication characteristics. Most of recent works use real social trace file to analyse its social characteristics, however social-related data is sensitive and has concern of privacy issues. In this paper, we propose an anonymous method that anonymize the original data by coding to preserve individual's privacy. The Shannon entropy is applied to the anonymous data to keep rich useful social characteristics for network optimization, e.g. routing optimization. We use an existing MIT reality dataset and Infocom 06 dataset, which are human associated mobile network trace files, to simulate our method. The results of our simulations show that this method can make data anonymously while achieving network optimization.

Two main problems prevent the deployment of content delivery in a wireless sensor network: the address, which is widely used in the Internet as the identifier, is meaningless in wireless network, and the routing efficiency is a big concern in wireless sensor network. This paper presents an embedded multi-level ring (MVR) structure to address those two problems. The MVR uses names rather than addresses to identify sensor nodes. The MVR routes packets on the name identifiers without being aware the location. Some sensor nodes are selected as the backbone nodes and are placed on the different levels of the virtual rings. MVR hashes nodes and contents identifiers, and stores them at the backbone nodes. MVR takes the cross-level routing to improve the routing efficiency. Further, MVR is constructed decentralized and runs on the mobile nodes themselves, requiring no central control. Experiments using ns2 simulator for up to 200 nodes show that the storage and bandwidth requirements of MVR grow slowly with the size of the network. Furthermore, MVR has demonstrated as self-administrating, fault-tolerant, and resilient under the different workloads. We also discuss alternative implementation options, and future work.

Now days, the online social networks (OSN) have gained considerable popularity. More and more people use OSN to share their interests and make friends, also the OSN helps users overcome the geographical barriers. With the development of OSN, there is an important problem users have to face that is trust evaluation. Before user makes friends with a stranger, the user need to consider the following issues: Can a stranger be trusted? How much the stranger can be trusted? How to measure the trust of a stranger? In this paper, we take two factors, Degree and Contact Interval into consideration, which produce a new trust evaluation model (T-OSN). T-OSN is aimed to solve how to evaluate the trust value of an OSN user, also which is more efficient, more reliable and easy to implement. Base on our research, this model can be used in wide range, such as online social network (OSN) trust evaluation, mobile network message forwarding, ad hoc wireless networking, routing message on Internet and peer-to-peer file sharing network. The T-OSN model has following obvious advantages compare to other trust evaluate methods. First of all, it is not base on features of traditional social network, such as, distance and shortest path. We choose the special features of OSN to build up the model, that is including numbers of friends(Degree) and contact frequency(Contact Interval). These species features makes our model more suitable to evaluate OSN users trust value. Second, the formulations of our model are quite simple but effective. That means, to calculate the result by using our formulations will not cost too much resources. Last but not least, our model is easy to implement for an OSN website, because of the features that we used in our model, such as numbers of friends and contact frequency are easy to obtain. To sum up, our model is using a few resources to obtain a valuable trust value that can help OSN users to solve an important security problem, we believe that will be big step for development of OSN.

With internet services to the end users becoming more homogenous, thus providing high bandwidth for all users, multimedia services such as IPTV to the public as a whole will finally become a reality, but even given the more abundant resources, IPTV architecture is far from being highly available due to technical limitations, we aim to provide a meaningful optimization in the P2P distribution model, which is currently based on a random structure bounded by high delays and low performance, by using channel probability, user's habits studies and users' similarity, in order to optimize one of the key aspects of IPTV which is the peers management, which directly reflects on resources and user's Quality of Experience.

The continuous growth of the users pool of Social Networking web sites such as Face book and My Space, and their incessant augmentation of services and capabilities will in the future, meet and compare in contrast with today's Content distribution Networks (CDN) and Peer-to-Peer File sharing applications such as Kazaa and Bit Torrent, but how can these two main streams applications, that already encounter their own security problems cope with the combined issues, trust for Social Networks, content and index poisoning in CDN? We will address the problems of Social Trust and File Sharing with an overlay level of trust model based on social activity and transactions, this can be an answer to enable users to increase the reliability of their online social life and also enhance the content distribution and create a better file sharing example. The aim of this research is to lower the risk of malicious activity on a given Social Network by applying a correlated trust model, to guarantee the validity of someone's identity, privacy and trustfulness in sharing content.

The key nodes in network play the critical role in system recovery and survival. Many traditional key nodes selection algorithms utilize the characters of the physical topology to find the key nodes. But they can hardly succeed in the mobile ad hoc network due to the mobility nature of the network. In this paper we propose a social-aware Kcore selection algorithm to work in the Pocket Switched Network. The social view of the network suggests the social position of the mobile nodes can help to find the key nodes in the Pocket Switched Network. The S-Kcore selection algorithm is designed to exploit the nodes' social features to improve the performance in data communication. Experiments use the NS2 shows S-Kcore selection algorithm workable in the Pocket Switched Network. Furthermore, with the social behavior information, those key nodes are more suitable to represent and improve the whole network's performance.

Most work on multi-biometric fusion is based on static fusion rules which cannot respond to the changes of the environment and the individual users. This paper proposes adaptive multi-biometric fusion, which dynamically adjusts the fusion rules to suit the real-time external conditions. As a typical example, the adaptive fusion of gait and face in video is studied. Two factors that may affect the relationship between gait and face in the fusion are considered, i.e., the view angle and the subject-to-camera distance. Together they determine the way gait and face are fused at an arbitrary time. Experimental results show that the adaptive fusion performs significantly better than not only single biometric traits, but also those widely adopted static fusion rules including SUM, PRODUCT, MIN, and MAX.

Differentiated service network (DiffServ) provides an architecture that is scalable and capable of differentiating applications' quality of service (QoS). However, it could not control its loads under heavy traffic conditions, and it could not achieve assured service for TCP and UDP traffic-mixes if these flows are aggregated into a DiffServ class. This paper investigates these problems across multiple DiffServ domains. The paper suggests a solution that involves an edge-aware resource discovery (RD) feedback loop and fair intelligent admission control (FIAC) scheme for each DiffServ domain. The scheme is characterized by two fundamental features. First, the RD loop infers the availability of network resources, yet for scalability, it does not involve core routers. Second, the admission control module understands incoming traffic's requirements, reconciles with available resources via the RD loop, and admits traffic intelligently according to the FIAC algorithm. Multiple-domain control is obtained by concatenation of the RD loop and FIAC admission for each domain along the path from a source to a destination. Simulation results demonstrate that the enhanced DiffServ solution can admit traffic appropriately in terms of fairness and efficiency.