With the easy availability of low-quality 3D facial data, research on low-quality 3D face recognition (FR) has gained widespread attention. However, most existing methods struggle to strike a balance between accuracy and computational complexity, with the enormous parameters being one of the primary challenges. To address this issue, we propose a novel lightweight multi-scale group shift fusion network (LMGSNet) for low-quality 3D FR. Specifically, we construct a multi-scale group attention layer-by-layer shift fusion module (GALSF) based on our proposed channel shift fusion (CSF) method, which integrates attention convolution and grouping operation to capture critical local features (such as the nose, mouth, and forehead) while significantly reducing parameters. Furthermore, we design a novel split-aggregate local feature fusion module (SALF) to enhance local features representation and capturing rich discriminative features. Extensive experiments on three challenging low-quality 3D face datasets demonstrate that our model achieves competitive recognition accuracy with the lowest parameters.

Code review is a common software engineering practice of practical importance to reduce software defects. Review today is often with the help of specialized tools, such as Gerrit. However, even in a tool-supported code review involves a significant amount of human effort to understand the code change, because the information required to inspect code changes may distribute across multiple files that reviewers are not familiar with. Code changes are often organized as commits for review. In this paper, we found that most of the commits contain a salient class(es), which is saliently modified and causes the modification of the rest classes in a commit. Our user studies confirmed that identifying the salient class in a commit can facilitate reviewers in understanding code change. Inspired by the effectiveness of machine learning techniques in the classification field, we model the salient class identification as a binary classification problem and a number of discriminative features is extracted for a commit and used to characterize the salience of a class. The experiments results show that our approach achieves an accuracy of 88 percent. A user study with industrial developers shows that our approach can really improve the efficiency of reviewers understanding code changes in a reviewing scenario without using comment.

Colorization is the process of coloring monochrome images. It has been widely used in photo processing and scientific illustration. However, colorizing grayscale images is an intrinsic ill-posed and ambiguous problem, with multiple plausible solutions. To address this issue, we develop a novel progressive automatic colorization via iterative generative models (iGM) that can produce satisfactory colorization in an unsupervised manner. In particular, the generative model is exploited in multi-color spaces (e.g., RGB, YCbCr) jointly and enforced with linearly autocorrelative constraint. This is regarded as the key prior information to pave the way for producing the most probable colorization in high-dimensional space. Experiments on indoor and outdoor scenes reveal that iGM produces more realistic and finer results, compared to state-of-the-arts.

Code review involves a significant amount of human effort to understand the code change, because the information required to inspect code changes may distribute across multiple files that reviewers are not familiar with. Code changes are often organized as commits for review. In this paper, we found that most of the commits contain a salient class, which is saliently modified and causes the modification of the rest classes in a commit. Our user studies confirmed that identifying the salient class in a commit can facilitate reviewers in understanding code change. We model the salient class identification as a binary classification problem and extract a number of discriminative features from commit to characterize the salience of a class. The initial experiment result shows that the proposed approach can improve the efficiency of reviewers understanding code changes in code review.

We present a novel framework of system combination for multi-document summarization.For each input set (input), we generate candidate summaries by combining whole sentences from the summaries generated by different systems.We show that the oracle among these candidates is much better than the summaries that we have combined.We then present a supervised model to select among the candidates.The model relies on a rich set of features that capture content importance from different perspectives.Our model performs better than the systems that we combined based on manual and automatic evaluations.We also achieve very competitive performance on six DUC/TAC datasets, comparable to the state-of-the-art on most datasets.

We introduce a supervised model for predicting word importance that incorporates a rich set of features.Our model is superior to prior approaches for identifying words used in human summaries.Moreover we show that an extractive summarizer using these estimates of word importance is comparable in automatic evaluation with the state-of-the-art.

Recently, system-on-chip (SoC) solutions have been realized by integrating not only complex digital processing but also extensive analog and radio frequency (RF) circuits in a single chip. This paper presents a novel RF self test methodology suitable for complex radio SoC's. The proposed methodology employs the digital processor embedded in an SoC to enable self-testing using low-cost digital automatic test equipments (ATE). Moreover, to achieve increased RF test coverage achievable through internal loopback, the embedded processor also utilizes a compact assisted test board and interfaces to it through simple general purpose I/O's (GPIO). The proposed self-test methodology has been successfully applied in the mass production (MP) of a 65nm CMOS combo SoC that includes Wi-Fi, Bluetooth, GPS and FM. The final test (FT) and wafer probe test (PT) of the SoC have been accomplished with a 3X reduction in total test time compared to conventional test methodology using RF instruments.

We investigate the problem of entity ranking towards descriptive queries, that aims to match entities referred in user queries to entities of a large knowledge base (KB). Entity ranking faces the primary challenge of the sparseness of entity related data, such as various ways of referring to an entity. The lack of sufficient variations of entity referring expressions in KB makes it difficult to find entities referred in user queries, especially when the queries are descriptive. We tackle this problem by enriching KB entries using web documents and query click logs. First, we propose a novel method of injecting textual information from web documents to the KB on a large scale. Since the number of web documents can be large, we propose to use keyword extraction and summarization techniques for compactly representing entity-related information. Second, we mine web search query logs to link entities to existing queries. Experiments show significant improvements after the KB enrichment, compared with two competitive baselines. We also achieve further improvements by combining the data from these two resources.

In this paper, we present SpiderRadio, a software defined cognitive radio (CR) prototype for dynamic spectrum access (DSA) networking. The medium access control (MAC) layer of SpiderRadio is implemented in software on top of commodity IEEE 802.11a/b/g hardware. However, the proposed architecture and implementation are applicable to other spectrum bands as well. We also present a dynamic spectrum sensing methodology for primary incumbent detection. The proposed method is based on observing the PHY errors, received signal strength and statistical model building. For coordination among radio nodes, synchronization and fast channel switching, we present new communication protocols, design extended management frame structure and modify the hardware abstraction layer. Several fundamental tradeoffs (e.g., complexity versus network performance) to be considered during a dynamic spectrum access radio network prototype implementation are also discussed in detail. To demonstrate the practical capabilities of the proposed SpiderRadio prototype, we also present various testbed experimental measurement results.

OBJECTIVE: A system that translates narrative text in the medical domain into structured representation is in great demand. The system performs three sub-tasks: concept extraction, assertion classification, and relation identification. DESIGN: The overall system consists of five steps: (1) pre-processing sentences, (2) marking noun phrases (NPs) and adjective phrases (APs), (3) extracting concepts that use a dosage-unit dictionary to dynamically switch two models based on Conditional Random Fields (CRF), (4) classifying assertions based on voting of five classifiers, and (5) identifying relations using normalized sentences with a set of effective discriminating features. MEASUREMENTS: Macro-averaged and micro-averaged precision, recall and F-measure were used to evaluate results. RESULTS: The performance is competitive with the state-of-the-art systems with micro-averaged F-measure of 0.8489 for concept extraction, 0.9392 for assertion classification and 0.7326 for relation identification. CONCLUSIONS: The system exploits an array of common features and achieves state-of-the-art performance. Prudent feature engineering sets the foundation of our systems. In concept extraction, we demonstrated that switching models, one of which is especially designed for telegraphic sentences, improved extraction of the treatment concept significantly. In assertion classification, a set of features derived from a rule-based classifier were proven to be effective for the classes such as conditional and possible. These classes would suffer from data scarcity in conventional machine-learning methods. In relation identification, we use two-staged architecture, the second of which applies pairwise classifiers to possible candidate classes. This architecture significantly improves performance.

We present a game theoretic power control of overlay/overt communications to maximize the goodput (effective throughput of error-free bits) of multi-terminal covert timing channels. Most approaches in the literature on covert timing channels discuss capacities of the timing channels but do not study how the overlay communication can be controlled to maximize the goodput of covert timing channels. We study the factors of the overlay communication that affect the goodput of each timing channel in a multi-terminal covert timing network. We show that the goodput of the covert timing channel can be enhanced by increasing the rate of overlay transmission and by game theoretic power control of overlay communication. We finally extend the game theoretic power control to maximize the goodput of each covert timing channel in a multi-terminal covert timing network by maximizing the asymptotic spectral efficiency of the overlay communication.

We address a unique security vulnerability in long term evolution (LTE) advanced and high speed packet access (HSPA+) wireless networks due to carrier/channel bonding. This vulnerability is shown to result in various amounts of service disruption based on the radio network parameters and the user locations. Typically, channel bonding have been perceived as a means to enhance the bandwidth and throughput for the users. However, this could also result in the loss of orthogonality between the bonded spectrum bands. We show that this leads to a security vulnerability that can be exploited by an attacker to cause service disruption. In this case, the attacker need not even operate in the same bands as the user, to be effective. We present an analysis to compare the loss in throughput caused by the vulnerability due to channel bonding in advanced LTE and HSPA+ networks. Results indicate that channel bonding is susceptible to about 70% loss of throughput in LTE networks and about 11-15% in HSPA+ networks compared to systems with no bonding. Also, users farther away from the base station suffer larger throughput degradation due to channel bonding in LTE networks, while it causes larger degradation in throughput for near users in HSPA+ networks. To the best of our knowledge, this is the first attempt to identify and analyze a significant security vulnerability in LTE and HSPA+ networks.

In this paper, we investigate a new type of denialof- service attack in dynamic spectrum access networks - Sybilenabled attack. In this attack, the attacker not only launches the primary user emulation (PUE) attacks but also creates and infiltrates multiple Sybil identities to compromise the decision making process of the secondary network via Byzantine attacks. We implement this attack in our cognitive radio testbed to show its feasibility and attack impact. We further analyze the optimal attack strategy from the perspective of the malicious attacker, i.e., the optimal allocation of Sybil interfaces for different attacks, to maximize the impact on the secondary network. The attack models are analyzed under two different scenarios: with and without a reputation system in the network fusion center. Numerical analysis and simulations are conducted to solve the optimal attack strategy and demonstrate the impact of attacks on the secondary network.

We address a unique security threat that arises due to channel fragmentation (or aggregation or bonding) in dynamic spectrum access (DSA) based IEEE 802.22 networks. Typically, channel fragmentation, aggregation and bonding have been studied in the literature as a means to enhance the spectrum utilization. However, the loss of orthogonality between the spectrum bands due to channel fragmentation, aggregation or bonding can be exploited by malicious attackers to cause a cognitive service disruption. We present an analysis of such a threat. We determine the optimal transmit powers a malicious attacker transmits on each fragment, so as to create maximum service disruption. Numerical results indicate that a malicious attacker can cause up to about 16% loss in the capacity of the system as a consequence of fragmentation. Detailed analysis is presented for channel fragmentation and can be easily applied to channel aggregation and bonding. To the best of our knowledge,this is the first analysis of such cognitive service disruption threats due to fragmentation.

In this paper, we investigate Sybil attacks on spectrum allocation in distributed dynamic spectrum access (DSA) networks. Using IEEE 802.11 devices as secondary nodes, we demonstrate the feasibility of mounting Sybil attacks in the cognitive radio testbed, in which the malicious node poses as multiple normal secondary nodes with different identities in order to steal more spectrum bands. We also show the impact of the attack through an example and simulation results. A defense strategy using the statistics of beacon intervals is also proposed. Through experimental results, we show the effectiveness of this defense mechanism when there is no interference from external sources as well as in the presence of interference.

Efficient routing is a fundamental issue in multi-hop wireless ad hoc networks. In this paper, we study the limitation of traditional routing structure in multi-hop wireless ad hoc networks due to (a) the layered structure of a wireless protocol stack and (b) the lack of coordination between medium access control (MAC) and routing protocols. These limitations result in long processing delays in a relay/forwarding node. In order to alleviate these issues, we propose a solution based on cross-layer MAC design, which improves the coordination between MAC and routing layers using an idea we call ``virtual link". The virtual link idea was implemented and tested in an ad hoc wireless network testbed. Experimental results show that the proposed cross-layer design significantly improves the performance in terms of reduced round trip time (RTT), reduced processing time in the intermediate relay/forwarding nodes and increased throughput compared to a legacy architecture.

Spectrum sensing is one of the critical features in cognitive radio based dynamic spectrum access networking. In this paper, we discuss a new spectrum sensing technique for primary incumbent detection. The proposed method is based on observing the PHY errors, received signal strength and n-moving window averaging of the observed measurement. The sensing parameters are dynamically optimized based on the operating radio environment. This sensing method is implemented in SpiderRadio, a cognitive radio testbed based on off-the-shelf IEEE 802.11 devices. Experimental results show that the proposed technique results in very low sensing delay and failure probability.