Search

Object detection techniques have been widely studied, utilized in various works, and have exhibited robust performance on images with sufficient luminance. However, these approaches typically struggle to extract valuable features from low-luminance images, which often exhibit blurriness and dim appearence, leading to detection failures. To overcome this issue, we introduce an innovative unsupervised feature domain knowledge distillation (KD) framework. The proposed framework enhances the generalization capability of neural networks across both low- and high-luminance domains without incurring additional computational costs during testing. This improvement is made possible through the integration of generative adversarial networks and our proposed unsupervised KD process. Furthermore, we introduce a region-based multiscale discriminator designed to discern feature domain discrepancies at the object level rather than from the global context. This bolsters the joint learning process of object detection and feature domain distillation tasks. Both qualitative and quantitative assessments shown that the proposed method, empowered by the region-based multiscale discriminator and the unsupervised feature domain distillation process, can effectively extract beneficial features from low-luminance images, outperforming other state-of-the-art approaches in both low- and sufficient-luminance domains.

Quantum computation and optimization have recently garnered considerable attention, with a noticeable focus on their floating-point and arithmetic designs. In classical computing, numerical optimization problems are commonly employed to assess the performance of optimization algorithms before their application to real-world issues. Nevertheless, precision issues impact the performance analysis of these algorithms, and neglecting these predictable exceptions can lead to unforeseen consequences. Therefore, this study systematically organizes potential precision issues related to how the floating-point storage format impacts optimization. These issues cause the optimal solution to deviate from the theoretical value, introducing imprecision or multiple optimal values, which in turn affects the usability of optimization algorithms, the direction of the search, and the assessment of convergence levels. These analyses offer valuable insights into the practical behavior of optimization algorithms when applied to function optimization problems, aiding researchers in accurately assessing and enhancing algorithm performance. Moreover, these findings contribute to the advancement of both classical and quantum computation.

Advanced object detection (OD) techniques have been widely studied in recent years and have been successfully applied in real-world applications. However, existing algorithms may struggle with nighttime image detection, especially in low-luminance conditions. Researchers have attempted to overcome this issue by collecting large amounts of multidomain data, but performance remains poor because these methods train images from both low- and sufficient-luminance domains without a specific training policy. In this work, we present a lightweight framework for multidomain OD using feature domain transformation with generative adversarial networks (GANs). The proposed GAN framework trains a generator network to transform features from the low-luminance domain to a sufficient-luminance domain, making the discriminator networks unable to distinguish whether the features were generated from a low-luminance or a normal image and thus achieving luminance-invariant feature extraction. To preserve semantic meaning in the transformed features, a training policy has been introduced for OD and feature transformation in various domains. The proposed method achieves the state-of-the-art performance with a 9.95 improvement in average precision without incurring additional computational costs.

Quantum computing is an essential issue for taking advantage of quantum properties in real-world applications. Realizing quantum computing with a corresponding quantum circuit is a critical step. This study introduces the AngelQ system, which provides quantum computing-as-a-service (QCaaS) for the cloud. The AngelQ system integrates quantum design automation (QDA) to achieve quantum and reversible circuit synthesis by a quantum-inspired optimization (QiO) technique and has visualization software interfaces to verify the QiO process and the correctness of the synthesized quantum and reversible circuits. First, the proposed novel QiO method has four significant features: Adaptive strategy, quantum-Not gate, update mechanism by Global solutions, and Entanglement Local search, named Angel QiO. which is a flexible technique to realize various circuits under different gate libraries with a few alterations. Angel QiO can construct diversified function-equivalent circuits to assess the circuit composition and boost the development of circuit synthesis. Second, researchers can observe the QiO algorithm process and inspect the intermediate result formula through AngelQ system's visualization interface to further adjust the parameters and improve the performance. Moreover, it benefits beginners’ learning and adds convenience to education. In summary, the AngelQ system can extend and facilitate the future design and automation of quantum computing.

Quantum computing and secure communication have gained considerable interest in recent years, as quantum computing has a potent ability to accelerate certain complex problems, such as factoring problems and discrete logarithms, which are important foundations of traditional computer security. With the rapid development of quantum computers, some secure communication protocols based on certain mathematical problems will face a significant danger, and quantum properties can be used to construct quantum secure communication and quantum cryptography that can achieve unconditional security. Among many topics of quantum cryptography, quantum secret sharing (QSS) is an important and challenging issue that provides scenarios to store highly sensitive, important, and confidential information. In this article, we systematically illustrate the development of QSS and how it operates, as well as its challenging issues and future directions.

Quantum computation has extraordinary capabilities for solving complicated problems. As quantum computations are reversible by nature, reversible circuits are important for the development of quantum computation techniques. Designing an effective and efficient method for synthesizing reversible circuits to reduce costs and stabilize circuit efficiency is crucial. The traditional synthesis methods of solving reversible circuits focus on the conversion efficiency rather than discussing the properties of the reversible function. Thus, this paper aims to propose a novel synthesis method that directly and efficiently optimizes reversible circuit synthesis with the properties of the reversible circuit. The proposed method converts the reversible function into a hypercube, allowing visual observations of the overall circuit. Two new indicators, the adjacent Hamming distance (AHD) and total cycle distance (TCD), aid in effective decision-making, generating shorter circuits. Furthermore, we use the generalized Toffoli gate set, which without requiring any additional ancilla bits and has applications in error correction and fault tolerance. The experimental results show that our method can find better solutions than traditional methods, significantly reducing the gate count, while the hypercube assists in synthesizing the reversible circuit.

This paper presents a new low complexity design of Reed Solomon model, which is the key technology for RAID system advanced codec. The real-time constraint of codec leads to a heavy computational bottleneck on today's data storage devices. To overcome this problem, design analysis and optimization of Reed Solomon code are addressed at the algorithmic level. First, the dominant calculation of spreading function is replaced with small lookup tables. Second, a method which greatly reduces the number of multipliers in performing the multiplication operations of the codec process is proposed to reduce the RAID (Redundant Array of Inexpensive Disks) system codec complexity while maintaining the quality. Moreover, our algorithm can be expanded to correct multiple failed disks. Therefore, the failed data in disks can be recovered and system still can work as usual without broken. This technique is different from the traditional methods. The proposed design could be implemented in a real-time Reed Solomon codec RAID system with reduction of hardware complexity by 28%. A hybrid strategy considering both data layout and load of storage nodes has a 13-25% higher I/O performance than conventional RS RAID. Index Terms--Reed Solomon codes, Reduced Static-Checksum, Reliability, RAID.

Restoration of visibility in hazy images is the first relevant step of information analysis in many outdoor computer vision applications. To this aim, the restored image must feature clear visibility with sufficient brightness and visible edges, while avoiding the production of noticeable artifacts. In this paper, we propose a haze removal approach based on the radial basis function (RBF) through artificial neural networks dedicated to effectively removing haze formation while retaining not only the visible edges but also the brightness of restored images. Unlike traditional haze-removal methods that consist of single atmospheric veils, the multiatmospheric veil is generated and then dynamically learned by the neurons of the proposed RBF networks according to the scene complexity. Through this process, more visible edges are retained in the restored images. Subsequently, the activation function during the testing process is employed to represent the brightness of the restored image. We compare the proposed method with the other state-of-the-art haze-removal methods and report experimental results in terms of qualitative and quantitative evaluations for benchmark color images captured in typical hazy weather conditions. The experimental results demonstrate that the proposed method is able to produce brighter and more vivid haze-free images with more visible edges than can the other state-of-the-art methods.

This study uses event trees and fault trees to analyze dependability issues on instrumentation and control systems in safety-critical systems. In an event tree, each state can be constructed as a fault tree and each fault tree can be converted to an ordered binary decision diagram (OBDD). The state probability can then be evaluated by quantifying the corresponding OBDD. We present an algorithm to combine OBDDs and evaluate each outcome probability for both statistically-independent (s-independent) and statistically-dependent (s-dependent) events. For s-independent events, the outcome probability can be easily obtained by probability multiplications of the associated branch states. For s-dependent events, we present an algorithm to combine OBDDs and compute each outcome probability. In the security domain, we analyze cyberphysical security issues with attack trees. The countermeasures and different levels/layers of protection are also presented. The integration of the dependability and security analysis are adopted by combining event trees, fault trees, and attack trees. A case study is presented to demonstrate that the integration of dependability and security is feasible and the improvement of outcome risk is significant through adopting security countermeasures.

We propose a channel state information acquisition mechanism to reduce the training overhead in frequency-division duplexing (FDD) massive multiple-input multiple-output (MIMO) downlink (DL). While the massive MIMO has demonstrated the great potentials in many aspects, the carrier multiplexing designs in massive MIMO mostly adopt time-division duplexing (TDD) system, which stems from a stereotype that the uplink/DL channel reciprocity only holds for TDD. On the other hand, FDD massive MIMO is generally considered infeasible due to the unaffordable temporal overhead for sending training symbols at base station. Our proposed FDD with reverse training (FDD-RT) modifies the DL band training procedure in typical FDD system to reduce pilot overhead, where the DL band channel reciprocity is exploited in FDD-RT. FDD-RT contributes to providing a feasible and relatively low complexity method to implement massive MIMO in FDD system. Without FDD-RT, communication operators who currently use FDD system may be enforced to change to TDD system in the future to accommodate massive MIMO, where the cost is prohibitive. The most different property of FDD-RT from other solutions is that its training overhead does not scale with the number of antennas at base station. The detail analysis/comparison of FDD-RT will be given in this paper.

A smart grid is considered as the next-generation power supply network, where the electricity is generated according to the real-time demand from household appliances. However, while authentication is still needed in the power request, because the power request could leak the life pattern to the malicious adversary, privacy preservation of the sensitive information in power request also needs to be considered. In this paper, we propose a privacy-preserving power request (PPP) scheme to fulfill the security requirements. In particular, when requesting for the power with the authentication guarantee, the source smart meter will not leak the identity to the power generator and substations of the smart grid. In addition, our proposed PPP scheme is efficient in terms of communication and computation overhead, which can be demonstrated by numerical simulations.

Rectilinear Steiner trees are used to route signal nets by global and detail routers in VLSI design for a long time. However, in current IC industry, there are significantly increasing obstacles to be considered, such as large-scale power networks, pre-routed nets, IP blocks, and antenna jumpers. Accordingly, the obstacle-avoiding rectilinear Steiner minimal tree (OARSMT) problem has become more important. In this article, we propose a new routing graph, obstacle-avoiding routing graph (OARG), for the OARSMT problem. Due to the important properties of OARG, we construct a 3-step algorithm and a local refinement scheme, which both can take advantage of these properties, to find a suboptimal solution efficiently. Furthermore, each step of our 3-step algorithm as well as the local refinement scheme has theoretical or practical benefits. Therefore, each of them can be applicable to other existing works for general or specific considerations such as efficiency or effectiveness. Extensive experimental results show that our method outperforms all existing works in terms of wirelength and achieves the best speed performance.

Time-based coordinated checkpointing protocols are well suited for mobile computing systems because no explicit coordination message is needed while the advantages of coordinated checkpointing are kept. However, without coordination, every process has to take a checkpoint during a checkpointing process. In this paper, an efficient time-based coordinated checkpointing protocol for mobile computing systems over Mobile IP is proposed. The protocol reduces the number of checkpoints per checkpointing process to nearly minimum, so that fewer checkpoints need to be transmitted through the costly wireless link. Our protocol also performs very well in the aspects of minimizing the number and size of messages transmitted in the wireless network. In addition, the protocol is nonblocking because inconsistencies can be avoided by the piggybacked information in every message. Therefore, the protocol brings very little overhead to a mobile host with limited resource. Additionally, by taking advantage of reliable timers in mobile support stations, the time-based checkpointing protocol can adapt to wide area networks.

This paper proposes an optical crossconnect (MFOXC) architecture that supports multicasting and fault tolerance. First, a tap-based and two splitter-based MFOXC node architectures are presented for wavelength routed all-optical networks. Compared to the existing optical crossconnects, the proposed MFOXC node not only performs multicasting efficiently but also improves reliability significantly. The MFOXC introduces a new feature (fault tolerance) and keeps the multicasting capability. It can be used at some critical points in a network to improve the overall reliability and multicast performance. Furthermore, the probability of maintaining fault free operations has been investigated for both MFOXC architectures. We present our evaluation results with a commonly used reliability measure, the mean time between failures (MTBF). Finally, we have proposed the cost and the sensitivity analysis for these MFOXC structures. The cost model and the sensitivity analysis show that the cost reduction in different components has various different impacts on the total cost of a MFOXC architecture. It can help us to know which component dominates the total cost and how to make a decision to choose among different MFOXC structures. The simulation results show that (1) the decrease of 75% in the cost of the N×N switch will result in the reduction of 20% in the total cost of the tap-based MFOXC, (2) the 1×2 switch has a big impact on the cost of the splitter-based MFOXC structures, and (3) the variation in the cost of the splitter does not introduce significant disadvantage to the type II splitter-based MFOXC structure.

This paper presents a new technique for constructing IP over photonic systems. The use of lable switching is assumed in the IP routers, while a new routing architecture is introduced to transport IP packets across an optical backbone network. The architecture is based on a two-level TDMA structure with wavelength division multiplexing (WDM). Many IP-based network applications such as high-resolution image, distributed database, and real-time video/audio service generally require high-speed transmissions in WAN/LAN. The network traffic in these applications usually exhibits traffic locality. As a result, traditional TDMA is not efficient for such traffic. Consequently, based on the traffic parameters such as locality and loading, an architecture named a PG (Partition-Group) Network is proposed. Furthermore, the interleaved control slot (ICS) with cross-group section (CGS) or non-cross-group section (NCGS) for reducing collisions is also presented. The slot reuse can be easily achieved by using the ICS scheme, and the slot utilization of the network can be improved within the high traffic locality.

Yield degradation from physical failures in large memories and processor arrays is of significant concern to semiconductor manufacturers. One method of increasing the yield for iterated arrays of memory cells or processing elements is to incorporate spare rows and columns in the die or wafer. These spare rows and columns can then be programmed into the array. The authors discuss the use of CAD approaches to reconfigure such arrays. The complexity of optimal reconfiguration is shown to be NP-complete. The authors present two algorithms for spare allocation that are based on graph-theoretic analysis. The first uses a branch-and-bound approach with early screening based on bipartite graph matching. The second is an efficient polynomial time-approximation algorithm. In contrast to existing greedy and exhaustive search algorithms, these algorithms provide highly efficient and flexible reconfiguration analysis.

Knowledge acquisition by interviewing a domain expert is one of the most problematic aspects of the development of expert systems. As an alternative, methods for inducing concept descriptions from examples have proven useful in eliminating this bottleneck. In this paper, we propose a probabilistic induction method (PIM), which is an improvement of the Chan and Wong method, for detecting relevant patterns implicit in a given data set. PIM uses the technique of residual analysis and several heuristics to effectively detect complex relevant patterns and to avoid the problem of combinatorial explosion. A reasonable trade-off between the induction time and the classification ratio is achieved. Moreover, PIM quickly classifies unknown objects using classification rules converted from the positively relevant patterns detected. Three experiments are conducted to confirm the validity of PIM.

A systematic efficient fault diagnosis method for reconfigurable VLSI/WSI array architectures is presented. The basic idea is to utilize the output data path independence among a subset of processing elements (PEs) based on the topology of the array under test. The “divide and conquer” technique is applied to reduce the complexity of test application and enhance the controllability and observability of a processor array. The array under test is divided into nonoverlapping diagnosis blocks. Those PEs in the same diagnosis block can be diagnosed concurrently. The problem of finding diagnosis blocks is shown equivalent to a generalizedEight Queensproblem. Three types of PEs and one type of switches, which are designed to be easily testable and reconfigurable, are used to show how to apply this approach. The main contribution of this paper is an efficient switch and link testing procedure, and a novel PE fault diagnosis approach which can speed up the testing by at leastO(⋎V⋎1/2) for the processor arrays considered in this paper, where ⋎V⋎ is the number of PEs. The significance of our approach is the ability to detect as well as to locate multiple PE, switch, and link faults with little or no hardware overhead.

An efficient method for evaluating the terminal-pair reliability based on an edge expansion tree and using an OBDD (ordered binary decision diagram) is presented. The effectiveness of the algorithm is demonstrated on the larger benchmarks collected in previous work. One notable case of the experimental results for a 2 × 20 lattice network is that the number of nodes in the OBDD is linearly proportional to the number of stages. This is significantly superior to previous algorithms which are based on the sum of disjoint products and has exponential complexity.

A technique for constructing fault-tolerant q-ary k-cube networks for an arbitrary positive integer q is investigated. Extra sets of links are added to construct the fault-tolerant network such that in the presence of link faults, the remaining healthy portion is guaranteed to contain an original fault-free network. Linear codes over an integer ring are used for this approach.