Search

Though deep neural networks have succeeded in various real applications, the prediction performance is significantly degraded when facing adversarial attacks. In this work, we investigate the alternation of the prediction distribution pattern under adversarial attacks and argue that such alternation is the primary reason for performance drop. To this end, we propose a simple yet effective method by introducing an induced class to attract the adversarial attack and thus protect the original classes' prediction order. Experiments on two real-world datasets demonstrate that the proposed method can maintain the prediction performance for both natural and adversarial examples.

Positive-unlabeled (PU) learning is a learning paradigm when only positive and unlabeled data are available in the training stage. This paradigm is particularly useful for the applications that the negative samples are hard to define or expensive to obtain. We propose a simple yet effective and scalable method to address PU learning with a novel asymmetric loss. The proposed asymmetric loss behaviors differently for the prediction errors of the labeled and unlabeled samples, and thus encourages the identification of the negative examples from the unlabeled set. For the PU learning with SCAR assumption, neither hyper-parameter nor class prior is required to be tuned or known. For the situation with selection bias on the labeled samples, we propose a heuristic method to automatically choose the hyper-parameter according to the class prior on the training data. Compared with previous approaches, our method only requires a slight modification of the conventional cross-entropy loss and is compatible with various deep neural networks in an end-to-end way. Extensive experiments on synthetic and real-world datasets with and without SCAR assumption verify the superior performance of the proposed method.

Gait recognition plays a vital role in human identification since gait is a unique biometric feature that can be perceived at a distance. Although existing gait recognition methods can learn gait features from gait sequences in different ways, the performance of gait recognition suffers from insufficient labeled data, especially in some practical scenarios associated with short gait sequences or various clothing styles. It is unpractical to label the numerous gait data. In this work, we propose a self-supervised gait recognition method, termed SelfGait, which takes advantage of the massive, diverse, unlabeled gait data as a pre-training process to improve the representation abilities of spatiotemporal backbones. Specifically, we employ the horizontal pyramid mapping (HPM) and micro-motion template builder (MTB) as our spatiotemporal backbones to capture the multi-scale spatiotemporal representations. Experiments on CASIA-B and OU-MVLP benchmark gait datasets demonstrate the effectiveness of the proposed SelfGait compared with four state-of-the-art gait recognition methods. The source code has been released at https://github.com/EchoItLiu/SelfGait.

Knowledge distillation, which aims to train model under the supervision from another large model (teacher model) to the original model (student model), has achieved remarkable results in supervised learning. However, there are two major problems with existing knowledge distillation methods. One is the teacher's supervision is sometimes misleading, and the other is the student's prediction is not accurate enough. To address the first issue, instead of learning a combination of both teachers and ground truth, we apply knowledge adjustment to correct teachers' supervision using ground truth. For the second problem, we use the selective classification framework to train the student model. In particular, the deep gambler loss is adopted to predict with reservation by explicitly introducing the ( $m+1$ )-th class. We consider two settings of knowledge distillation: (1) distillation across different network structures (AlexNet, ResNet), and (2) distillation across networks with different depths (ResNet18, ResNet5 0) to evaluate the effectiveness of our method. The experimental results on benchmark datasets (i.e., Fashion-MNIST, SVHN, CIFAR10, CIFAR100) are reported with higher prediction accuracies and lower coverage errors.

Sparse representation and dictionary learning of image patches are well-known methods for single-image super-resolution. However, due to the regularization term of sparse-inducing penalties, the solution is usually biased. In this study, we present a de-biasing framework by adding a de-biasing step after sparse representation. Two de-biasing methods with sign consistency and feature consistency are further proposed under this framework. Using a unified proximal gradient method, we can solve the proposed de-biasing methods efficiently. Experiments on real super-resolution datasets validate the effectiveness and robustness of the proposed de-biasing methods.

Early and accurate diagnosis of Alzheimer's disease is beneficial to both preserve daily functioning and test possible new treatments. However, current diagnosis depends on dozens of factors, including the family member, past medical problems, tests of memory, blood and urine tests, brain scans and even cerebrospinal fluid specimens. Among them, regular features (e.g., blood and urine tests, brain scans) are simple and accurate. Privileged features (e.g., tests of memory, cerebrospinal fluid specimens) are inconvenient and uncomfortable to acquire, and thus incomplete in most cases. In this work, we propose a two-stage learning framework to predict merely using the regular feature with the aid of incomplete privileged data in the training time. In particular, we first complement the missing data of privileged features by exploring the relationship with regular features and labels. The recovered privileged features, regular features, as well as labels, are combined as a new set of privileged features. Then privileged learning via matching logits is applied to boost the diagnosis accuracy. Our experiments and comparison studies with competing techniques on both synthetic data and real benchmarks have corroborated the effectiveness and superiority of the proposed framework for biomedical applications.

Brain tumors, especially high-grade gliomas, are one of the most lethal cancers for humankind today. Early and accurate diagnosis of tumor grading is the key for subsequent therapy and treatment. In the past, conventional computer-aided diagnosis relies on handcrafted features from magnetic resonance images (MRI), which are usually inaccurate and laborious. Recently, deep neural networks have been developed and applied for tumor segmentation and classification. However, most existing methods consider 3D MRI as a series of 2D images and use a simple modality fusion method via feature concatenation. In this paper, we propose an end-to-end 3-dimensional convolutional neural network (3D CNN) with gated multimodal unit (GMU) fusion to integrate the information both in three dimensions and in multiple modalities. Specifically, 3D convolutional kernels are directly applied to the whole MRI images, gathering the abnormalities in sagittal, axial and coronal directions. GMU with hidden states is proposed to fuse the information of multiple MRI modalities in both feature and decision level. Based on these, privilege information extracted by GMU fusion model is utilized to train a novel network called distilled-CNN, which significantly improves the performance of classification using single modality. Empirical studies on BRATS datasets corroborate the effectiveness of the proposed 3D CNN with GMU fusion and distilled-CNN to distinguish benign gliomas and malignant gliomas.

Splitting control is the last defence of power system emergency control against severe disturbance. With system-level and real-time decision mode, active splitting control shows advantages over traditional splitting control. Unfortunately, searching for splitting section is a NP hard problem. In this paper, a novel active splitting strategy search method inspired by modularity-based network partition method is proposed. The weighted graph model of power system is established based on both electrical distance and power flow state. An aggregation splitting strategy search method is designed based on a comprehensive evaluation index. The proposed index collects network modularity, power balance constraints, synchronization separation constraints and island connectivity so that it is able to reflect the reasonability of system partition comprehensively. Simulation in IEEE-68 system shows that the proposed method is able to find a splitting strategy quickly and effectively, and finally control the system collapse.

In this paper, we propose a new open-loop rate/link adaptation algorithm (ARFHT) for the emerging high-speed IEEE 802.11n WLANs. ARFHT extends the legacy rate adaptation algorithms for SISO WLANs to make it applicable in the context of MIMO-based 802.11n WLANs. It adapts the MIMO mode in terms of spatial multiplexing and spatial diversity, the two fundamental characteristics of the 802.11n MIMO PHY. It also modifies the link estimation and probing behavior of legacy SISO algorithms. The combined adaptation to the appropriate MIMO mode as well as the appropriate modulation coding scheme selection achieves high channel utilization. In this paper, we provide the intuition and the design details of the ARFHT algorithm. A comprehensive simulation study using ns-2 will demonstrate that ARFHT achieves excellent throughput performance in most scenarios and is highly responsive to varying link conditions, with minimum overhead.

TCP is currently the dominate congestion control protocol for the Internet. However, as the Internet evolves into a high-speed wired-cum-wireless hybrid network, performance degradation problems of TCP have appeared, such as underutilizing high-speed links, regarding wireless loss as congestion signal, and unfairness among flows with different RTTs. In order to improve the quality of service for such high-speed hybrid networks, we propose a router-assisted congestion control protocol called quick flow control protocol (QFCP). Performance evaluation using network simulator NS-2 shows that QFCP can significantly shorten flow completion time, fairly allocate bandwidth resource, and be robust to non-congestion- related loss.

This paper introduces a reliability conscious admission controller for enterprise level data networks. It builds upon earlier work on adaptive admission controllers designed to optimize utility (often revenue) in a data network. This was generally accomplished by using service level agreements (SLAs) to describe user requirements, and developing techniques for optimally accommodating a subset of all the SLAs requests for admission. Such admission problem can be mapped to a variant of the classic Knapsack problem. The controller introduced here, called the reliable SLA optimizer (R-SLAOpt), extends the previous model to make provisions for improved reliability. For instance, an SLA must be allocated to a new path in the event of a link failure. To accomplish this SLA adaptation, alternate paths are pre-calculated and therefore these paths can be quickly selected and activated. This makes R-SLAOpt a far more appropriate model for time critical applications, such as managing multimedia sessions.

Router-assisted explicit-feedback congestion control protocols have recently been introduced to overcome the inefficiency problem of TCP in high bandwidth-delay product (BDP) wired networks. However, two main challenges are encountered when applying this kind of congestion control to wireless networks. One is how to distinguish the two kinds of packet loss (non-congestion-related loss and congestion-related loss) in lossy wireless networks as well as how to react to them accordingly and properly. The other is bow to probe the unknown bandwidth capacity of a wireless link which is required in calculating router feedback. Through detailed analysis of these challenges, we have proposed some practical and novel enhancements on router-assisted congestion control for wireless environment We have also implemented these enhancements in a router-assisted congestion control protocol called QFCP. Simulation results using ns-2 show that it can fairly allocate wireless bandwidth resource in heterogeneous networks.

Neural network system is a self - learning adaptive system, and it is easy to associate, synthesize and generalize with its properties of fault - tolerance and robustness. Therefore, it is available to process the pattern information, which is hard to describe with language. In consideration of the shortage of Fuzzy Theory and the advantage of vague set, that is Fuzzy membership function has only one single value; it cannot get more reasonable classified and cognizable results. While vague sets' distinguishing feature is having two values which present both of the opposite factors to deal with nonlinearities and uncertain system. In this paper, we define new fuzzy neurons and propose the structure of Vague Neural Network (VNN), and then we expounded the recognition method based on it.

Quick flow control protocol (QFCP) is a new congestion control protocol designed for high bandwidth-delay product networks. QFCP has two good features: ";quick start"; and ";quick convergence";. It allows a sender to start with a high initial sending rate identified by routers along the path, and all running flows converge to the fair-share sending rate quickly based on feedback from routers. Although it needs the assistance of routers, QFCP does not require routers to store any per-flow information or to do any complex per-packet calculation. The rate allocation algorithm is quite simple and only needs to be run periodically by routers. We have implemented QFCP in network simulator (NS). Simulations have been done to address the issues such as fairness, convergence, responsiveness and utilization, as well as flow completion time for Poison-arriving Pareto- distributed-size flows. Performance evaluation of these simulations is presented in this paper. The preliminary results are promising.

Alternative paths may significantly improve routing reliability in an IP network, One application of this technique is to enhance the reliability of the popular open shortest path first (OSPF) routing protocol. In our proposed reliable OSPF (ROSPF) routing protocol, one primary path and two alternate backup paths are deployed for data transmission. As expected, the number of shared links and routers among the three paths dominates the reliability of the routing connection between two routers. The calculation of routing reliability of multiple paths is very important in alternate path-finding algorithms. To solve this problem, we use the Venn diagram model to analyze the overall failure probability of three partially disjoint paths and to understand how the double- and triple-shared links affect the routing reliability. General mathematical formulas to calculate the failure probability are also obtained.