The efficient implementation strategy for speeding up high-quality clustering algorithms is developed on the basis of general purpose graphic processing units (GPGPUs) in this work. Among various clustering algorithms, a sophisticated Gaussian mixture model (GMM) by estimating parameters through variational Bayesian (VB) mechanism is conducted due to its superior performances. Since the VB-GMM methodology is computation-hungry, the GPGPU is employed to carry out massive matrix-computations. To efficiently migrate the conventional CPU-oriented schemes of VB-GMM onto GPGPU platforms, an entire migration-flow with thirteen stages is presented in detail. The CPU-GPGPU co-operation scheme, execution re-order, and memory access optimization are proposed for optimizing the GPGPU utilization and maximizing the clustering speed. Five types of real-world applications along with relevant data-sets are introduced for the cross-validation. From the experimental results, the feasibility of implementing VB-GMM algorithm by GPGPU is verified with practical benefits. The proposed GPGPU migration achieves 192x speedup in maximum. Furthermore, it succeeded in identifying the proper number of clusters, which is hardly conducted by the EM-algotihm.

For a graph G=(V,E), finding a set of disjoint edges that do not share any vertices is called a matching problem, and finding the maximum matching is a fundamental problem in the theory of distributed graph algorithms. Although local algorithms for the approximate maximum matching problem have been widely studied, exact algorithms have not been much studied. In fact, no exact maximum matching algorithm that is faster than the trivial upper bound of O(n2) rounds is known for general instances. In this paper, we propose a randomized $O(s_{max}^{3/2})$-round algorithm in the CONGEST model, where smax is the size of maximum matching. This is the first exact maximum matching algorithm in o(n2) rounds for general instances in the CONGEST model. The key technical ingredient of our result is a distributed algorithms of finding an augmenting path in O(smax) rounds, which is based on a novel technique of constructing a sparse certificate of augmenting paths, which is a subgraph of the input graph preserving at least one augmenting path. To establish a highly parallel construction of sparse certificates, we also propose a new characterization of sparse certificates, which might also be of independent interest.

Android operating system occupies a high share in the mobile terminal market. It promotes the rapid development of Android applications (apps). However, the emergence of Android malware greatly endangers the security of Android smartphone users. Existing research works have proposed a lot of methods for Android malware detection, but they did not make the utilization of apps' functional category information so that the strong similarity between benign apps in the same functional category is ignored. In this paper, we propose an Android malware detection scheme based on the functional classification. The benign apps in the same functional category are more similar to each other, so we can use less features to detect malware and improve the detection accuracy in the same functional category. The aim of our scheme is to provide an automatic application functional classification method with high accuracy. We design an Android application functional classification method inspired by the hyperlink induced topic search (HITS) algorithm. Using the results of automatic classification, we further design a malware detection method based on app similarity in the same functional category. We use benign apps from the Google Play Store and use malware apps from the Drebin malware set to evaluate our scheme. The experimental results show that our method can effectively improve the accuracy of malware detection.

The usefulness and usability of existing knowledge graphs (KGs) are mostly limited because of the incompleteness of knowledge compared to the growing number of facts about the real world. Most existing ontology-based KG completion methods are based on the closed-world assumption, where KGs are fixed. In these methods, entities and relations are defined, and new entity information cannot be easily added. In contrast, in open-world assumptions, entities and relations are not previously defined. Thus there is a vast scope to find new entity information. Despite this, knowledge acquisition under the open-world assumption is challenging because most available knowledge is in a noisy unstructured text format. Nevertheless, Open Information Extraction (OpenIE) systems can extract triples, namely (head text; relation text; tail text), from raw text without any prespecified vocabulary. Such triples contain noisy information that is not essential for KGs. Therefore, to use such triples for the KG completion task, it is necessary to identify competent triples for KGs from the extracted triple set. Here, competent triples are the triples that can contribute to add new information to the existing KGs. In this paper, we propose the Competent Triple Identification (CTID) model for KGs. We also propose two types of feature, namely syntax- and semantic-based features, to identify competent triples from a triple set extracted by a state-of-the-art OpenIE system. We investigate both types of feature and test their effectiveness. It is found that the performance of the proposed features is about 20% better compared to that of the ReVerb system in identifying competent triples.

Virtual Adversarial Training (VAT) has shown impressive results among recently developed regularization methods called consistency regularization. VAT utilizes adversarial samples, generated by injecting perturbation in the input space, for training and thereby enhances the generalization ability of a classifier. However, such adversarial samples can be generated only within a very small area around the input data point, which limits the adversarial effectiveness of such samples. To address this problem we propose LVAT (Latent space VAT), which injects perturbation in the latent space instead of the input space. LVAT can generate adversarial samples flexibly, resulting in more adverse effect and thus more effective regularization. The latent space is built by a generative model, and in this paper we examine two different type of models: variational auto-encoder and normalizing flow, specifically Glow. We evaluated the performance of our method in both supervised and semi-supervised learning scenarios for an image classification task using SVHN and CIFAR-10 datasets. In our evaluation, we found that our method outperforms VAT and other state-of-the-art methods.

A new adaptive binarization method is proposed for the vehicle state images obtained from the intelligent operation and maintenance system of rail transit. The method can check the corresponding vehicle status information in the intelligent operation and maintenance system of rail transit more quickly and effectively, track and monitor the vehicle operation status in real time, and improve the emergency response ability of the system. The advantages of the proposed method mainly include two points. For decolorization, we use the method of contrast preserving decolorization[1] obtain the appropriate ratio of R, G, and B for the grayscale of the RGB image which can retain the color information of the vehicle state images background to the maximum, and maintain the contrast between the foreground and the background. In terms of threshold selection, the mean value and standard deviation of gray value corresponding to multi-color background of vehicle state images are obtained by using major cluster estimation[2], and the adaptive threshold is determined by the 2 sigma principle for binarization, which can extract text, identifier and other target information effectively. The experimental results show that, regarding the vehicle state images with rich background color information, this method is better than the traditional binarization methods, such as the global threshold Otsu algorithm[3] and the local threshold Sauvola algorithm[4],[5] based on threshold, Mean-Shift algorithm[6], K-Means algorithm[7] and Fuzzy C Means[8] algorithm based on statistical learning. As an image preprocessing scheme for intelligent rail transit data verification, the method can improve the accuracy of text and identifier recognition effectively by verifying the optical character recognition through a data set containing images of different vehicle statuses.

Various electromagnetic (EM) wave applications have become commonplace, and humans are frequently exposed to EM waves. Therefore, the effect of EM waves on the human body should be evaluated. In this study, we focused on the specific absorption rate (SAR) due to the EM waves emitted from smartphones, developed high-resolution numerical smartphone models, and studied the SAR variation by changing the position and tilt angle (the angle between the display of the smartphone model and horizontal plane) of the smartphone models vis-à-vis the human abdomen, assuming the use of the smartphone at various tilt angles in front of the abdomen. The calculations showed that the surface shape of the human model influenced the SAR variation.

It is an important research area to construct a cryptosystem that satisfies the security for multi-user setting. In addition, it is desirable that such a cryptosystem is tightly secure and the ciphertext size is small. For IND-CCA public key encryption schemes for multi-user setting with constant-size ciphertexts tightly secure under the DH assumptions, in 2020, Y. Sakai and G. Hanaoka firstly proposed such a scheme (implicitly based on hybrid encryption paradigm) under the DDH assumption. More recently, Y. Lee et al. proposed such a hybrid encryption scheme (with slightly stronger security) where the assumption for the KEM part is weakened to the CDH assumption. In this paper, we revisit the twin-DH hashed ElGamal KEM with even shorter ciphertexts than those schemes, and prove that its IND-CCA security for multi-user setting is in fact tightly reducible to the CDH assumption.

Due to the property of supporting arbitrary operation over the encrypted data, fully homomorphic encryption (FHE) has drawn considerable attention since it appeared. Some FHE schemes have been constructed based on the general approximate common divisor (GACD) problem, which is widely believed intractable. Therefore, studying the GACD problem's hardness can provide proper security parameters for these FHE schemes and their variants. This paper aims to study an orthogonal lattice algorithm introduced by Ding and Tao (Ding-Tao algorithm) to solve the GACD problem. We revisit the condition that Ding-Tao algorithm works and obtain a new bound of the GACD samples' number based on geometric series assumption. Simultaneously, we also give an analysis of the bound given in the previous work. To further verify the theoretical results, we conduct experiments on Ding-Tao algorithm under our bound. We show a comparison with the experimental results under the previous bound, which indicates the success probability under our bound is higher than that of the previous bound with the growth of the bound.

Smart contracts are protocols that can automatically execute a transaction including an electronic contract when a condition is satisfied without a trusted third party. In a representative use-case, a smart contract is executed when multiple parties fairly trade on a blockchain asset. On blockchain systems, a smart contract can be regarded as a system participant, responding to the information received, receiving and storing values, and sending information and values outwards. Also, a smart contract can temporarily keep assets, and always perform operations in accordance with prior rules. Many cryptocurrencies have implemented smart contracts. At POST2018, Atzei et al. give formulations of seven fair exchange protocols using smart contract on Bitcoin: oracle, escrow, intermediated payment, timed commitment, micropayment channels, fair lotteries, and contingent payment. However, they only give an informal discussion on security. In this paper, we verify the fairness of their seven protocols by using the formal verification tool ProVerif. As a result, we show that five protocols (the oracle, intermediated payment, timed commitment, micropayment channels and fair lotteries protocols) satisfy fairness, which were not proved formally. Also, we re-find known attacks to break fairness of two protocols (the escrow and contingent payment protocols). For the escrow protocol, we formalize the two-party scheme and the three-party scheme with an arbitrator, and show that the two-party scheme does not satisfy fairness as Atzei et al. showed. For the contingent payment protocol, we formalize the protocol with the non-interactive zero-knowledge proof (NIZK), and re-find the attack shown by Campanelli et al. at CCS 2017. Also, we show that a countermeasure with subversion NIZK against the attack works properly while it is not formally proved.

In spite of the research for a linear layer of Type-2 Generalized Feistel Network (Type-2 GFN) over more than 10 years, finding a good 32-branch permutation for Type-2 GFN is still a very hard task due to a huge search space. In terms of the diffusion property, Suzaki and Minematsu investigated the required number of rounds to achieve the full diffusion when the branch number is up to 16. After that, Derbez et al. presented a class of 32-branch permutations that achieves the 9-round full diffusion and they prove that this is optimal. However, this class is not suitable to be used in Type-2 GFN because it requires a large number of rounds to ensure a sufficient number of active S-boxes. In this paper, we present how to find a good class of 32-branch permutations for Type-2 GFN. To achieve this goal, we convert Type-2 GFN into a LBlock-like structure, and then we evaluate the diffusion property and the resistance against major attacks, such as differential, linear, impossible differential and integral attacks by an MILP. As a result, we present a good class of 32-branch permutations that achieves the 10-round full diffusion, ensures differentially/linearly active S-boxes of 66 at 19 round, and has the 18/20-round impossible differential/integral distinguisher, respectively. The 32-branch permutation used in WARP was chosen among this class.

Physical attacks against cryptographic devices and their countermeasures have been studied for over a decade. Physical attacks on block-cipher algorithms usually target a few rounds near the input or the output of cryptographic algorithms. Therefore, in order to reduce the implementation cost or increase the performance, countermeasures tend to be applied to the rounds that can be targeted by physical attacks. For example, for AES, the conventional physical attacks have practical complexity when the target leakage is as deep as 4 rounds. In general, the deeper rounds are targeted, the greater the cost required for attackers. In this paper, we focus on the physical attack that uses the leakage as deep as 5 rounds. Specifically, we consider the recently proposed 5-round mixture differential cryptanalysis, which is not physical attack, into the physical attack scenarios, and propose the corresponding physical attack. The proposed attack can break AES-128 with data complexity and time complexity of 225.31. As a result, it is clear that the rounds as deep as 5 must be protected for AES. Furthermore, we evaluated the proposed attack when the information extracted from side-channel leakage contains noise. In the means of theoretical analysis and simulated attacks, the relationship between the accuracy of information leakage and the complexity of the attack is evaluated.

Autonomous robots are controlled using physical information acquired by various sensors. The sensors are susceptible to physical attacks, which tamper with the observed values and interfere with control of the autonomous robots. Recently, sensor spoofing attacks targeting subsequent algorithms which use sensor data have become large threats. In this paper, we introduce a new attack against the LiDAR-based simultaneous localization and mapping (SLAM) algorithm. The attack uses an adversarial LiDAR scan to fool a pose graph and a generated map. The adversary calculates a falsification amount for deceiving pose estimation and physically injects the spoofed distance against LiDAR. The falsification amount is calculated by gradient method against a cost function of the scan matching algorithm. The SLAM algorithm generates the wrong map from the deceived movement path estimated by scan matching. We evaluated our attack on two typical scan matching algorithms, iterative closest point (ICP) and normal distribution transform (NDT). Our experimental results show that SLAM can be fooled by tampering with the scan. Simple odometry sensor fusion is not a sufficient countermeasure. We argue that it is important to detect or prevent tampering with LiDAR scans and to notice inconsistencies in sensors caused by physical attacks.

A backdoor attack is a type of attack method inducing deep neural network (DNN) misclassification. An adversary mixes poison data, which consist of images tampered with adversarial marks at specific locations and of adversarial target classes, into a training dataset. The backdoor model classifies only images with adversarial marks into an adversarial target class and other images into the correct classes. However, the attack performance degrades sharply when the location of the adversarial marks is slightly shifted. An adversarial mark that induces the misclassification of a DNN is usually applied when a picture is taken, so the backdoor attack will have difficulty succeeding in the physical world because the adversarial mark position fluctuates. This paper proposes a new approach in which an adversarial mark is applied using fault injection on the mobile industry processor interface (MIPI) between an image sensor and the image recognition processor. Two independent attack drivers are electrically connected to the MIPI data lane in our attack system. While almost all image signals are transferred from the sensor to the processor without tampering by canceling the attack signal between the two drivers, the adversarial mark is injected into a given location of the image signal by activating the attack signal generated by the two attack drivers. In an experiment, the DNN was implemented on a Raspberry pi 4 to classify MNIST handwritten images transferred from the image sensor over the MIPI. The adversarial mark successfully appeared in a specific small part of the MNIST images using our attack system. The success rate of the backdoor attack using this adversarial mark was 91%, which is much higher than the 18% rate achieved using conventional input image tampering.

In the source coding problem with cost constraint, a cost function is defined over the code alphabet. This can be regarded as a noiseless channel coding problem with cost constraint. In this case, we will not distinguish between the input alphabet and the output alphabet of the channel. However, we must distinguish them for a noisy channel. In the channel coding problem with cost constraint so far, the cost function is defined over the input alphabet of the noisy channel. In this paper, we define the cost function over the output alphabet of the channel. And, the cost is paid only after the received word is observed. Note that the cost is a random variable even if the codeword is fixed. We show the channel capacity with cost constraint defined over the output alphabet. Moreover, we generalize it to tolerate some decoding error and some cost overrun. Finally, we show that the cost constraint can be described on a subset of arbitrary set which may have no structure.

In the user identification (UI) scheme for a multiple-access fading channel based on a randomly generated (0, 1, -1)-signature code, previous studies used the signature code over a noisy multiple-access adder channel, and only the user state information (USI) was decoded by the signature decoder. However, by considering the communication model as a compressed sensing process, it is possible to estimate the channel coefficients while identifying users. In this study, to improve the efficiency of the decoding process, we propose an iterative deep neural network (DNN)-based decoder. Simulation results show that for the randomly generated (0, 1, -1)-signature code, the proposed DNN-based decoder requires less computing time than the classical signal recovery algorithm used in compressed sensing while achieving higher UI and channel estimation (CE) accuracies.

This letter proposes a transmission scheme called spatial vector (SV), which is effective for Nakagami-m fading multiple-input multiple-output channels. First, the analytical error rate of SV is derived for Nakagami-m fading MIMO channels. Next, an example of SV called integer SV (ISV) is introduced. The error performance was evaluated over Nakagami-m fading from m = 1 to m = 50 and compared with spatial modulation (SM), enhanced SM, and quadrature SM. The results show that for m > 1, ISV outperforms the SM schemes and is robust to m variations.

There have been some researchers that investigate the accuracy of the approximation to a function that shows a generating pattern of data by a deep neural network. However, they have confirmed only whether at least one function close to the function showing a generating pattern exists in function classes of deep neural networks whose parameter values are changing. Therefore, we propose a new criterion to infer the approximation accuracy. Our new criterion shows the existence ratio of functions close to the function showing a generating pattern in the function classes. Moreover, we show a deep neural network with a larger number of layers approximates the function showing a generating pattern more accurately than one with a smaller number of layers under the proposed criterion, with numerical simulations.

This work proposes a design methodology that saves the power dissipation under voltage over-scaling (VOS) operation. The key idea of the proposed design methodology is to combine critical path isolation (CPI) and bit-width scaling (BWS) under the constraint of computational quality, e.g., Peak Signal-to-Noise Ratio (PSNR) in the image processing domain. Conventional CPI inherently cannot reduce the delay of intrinsic critical paths (CPs), which may significantly restrict the power saving effect. On the other hand, the proposed methodology tries to reduce both intrinsic and non-intrinsic CPs. Therefore, our design dramatically reduces the supply voltage and power dissipation while satisfying the quality constraint. Moreover, for reducing co-design exploration space, the proposed methodology utilizes the exclusiveness of the paths targeted by CPI and BWS, where CPI aims at reducing the minimum supply voltage of non-intrinsic CP, and BWS focuses on intrinsic CPs in arithmetic units. From this key exclusiveness, the proposed design splits the simultaneous optimization problem into three sub-problems; (1) the determination of bit-width reduction, (2) the timing optimization for non-intrinsic CPs, and (3) investigating the minimum supply voltage of the BWS and CPI-applied circuit under quality constraint, for reducing power dissipation. Thanks to the problem splitting, the proposed methodology can efficiently find quality-constrained minimum-power design. Evaluation results show that CPI and BWS are highly compatible, and they significantly enhance the efficacy of VOS. In a case study of a GPGPU processor, the proposed design saves the power dissipation by 42.7% with an image processing workload and by 51.2% with a neural network inference workload.

With the demand for energy-efficient and high- performance computing, multicore architecture has become more appealing than ever. Multicore task scheduling is one of domains in parallel computing which exploits the parallelism of multicore. Unlike traditional scheduling, multicore task scheduling has recently been studied on the assumption that tasks have inherent parallelism and can be split into multiple sub-tasks in data parallel fashion. However, it is still challenging to properly determine the degree of parallelism of tasks and mapping on multicores. Our proposed scheduling techniques determine the degree of parallelism of tasks, and sub-tasks are decided which type of cores to be assigned to heterogeneous multicores. In addition, two approaches to hardware/software codesign for heterogeneous multicore systems are proposed. The works optimize the types of cores organized in the architecture simultaneously with scheduling of the tasks such that the overall energy consumption is minimized under a deadline constraint, a warm start approach is also presented to effectively solve the problem. The experimental results show the simultaneous scheduling and core-type optimization technique remarkably reduces the energy consumption.