A new software based in-situ training (SBIST) method to achieve high accuracies is proposed for binarized neural networks inference accelerator chips in which measured offsets in sense amplifiers (activation binarizers) are transformed into biases in the training software. To expedite this individual training, the initial values for the weights are taken from results of a common forming training process which is conducted in advance by using the offset fluctuation distribution averaged over the fabrication line. SPICE simulation inference results for the accelerator predict that the accuracy recovers to higher than 90% even when the amplifier offset is as large as 40mV only after a few epochs of the individual training.
Yuki SHIMIZU Tomotaka KIMURA Jun CHENG
In this study, we consider fake message attacks in sparse mobile ad hoc networks, in which nodes are chronically isolated. In these networks, messages are delivered to their destination nodes using store-carry-forward routing, where they are relayed by some nodes. Therefore, when a node has messages in its buffer, it can falsify the messages easily. When malicious nodes exist in the network, they alter messages to create fake messages, and then they launch fake message attacks, that is, the fake messages are spread over the network. To analyze the negative effects of a fake message attack, we model the system dynamics without attack countermeasures using a Markov chain, and then formalize some performance metrics (i.e., the delivery probability, mean delivery delay, and mean number of forwarded messages). This analysis is useful for designing countermeasures. Moreover, we consider a hash-based countermeasure against fake message attacks using a hash of the message. Whenever a node that has a message and its hash encounters another node, it probabilistically forwards only one of them to the encountered node. By doing this, the message and the hash value can be delivered to the destination node via different relay nodes. Therefore, even if the destination node receives a fake message, it can verify the legitimacy of the received message. Through simulation experiments, we evaluate the effectiveness of the hash-based countermeasure.
The increasing amount of fake news is a growing problem that will progressively worsen in our interconnected world. Machine learning, particularly deep learning, is being used to detect misinformation; however, the models employed are essentially black boxes, and thus are uninterpretable. This paper presents an overview of explainable fake news detection models. Specifically, we first review the existing models, datasets, evaluation techniques, and visualization processes. Subsequently, possible improvements in this field are identified and discussed.
Loan default prediction has been a significant problem in the financial domain because overdue loans may incur significant losses. Machine learning methods have been introduced to solve this problem, but there are still many challenges including feature multicollinearity, imbalanced labels, and small data sample problems. To replicate the success of deep learning in many areas, an effective regularization technique named muddling label regularization is introduced in this letter, and an ensemble of feed-forward neural networks is proposed, which outperforms machine learning and deep learning baselines in a real-world dataset.
Combinatorial testing is an effective testing technique for detecting faults in a software or hardware system with multiple factors using combinatorial methods. By performing a test, which is an assignment of possible values to all the factors, and verifying whether the system functions as expected (pass) or not (fail), the presence of faults can be detected. The failures of the tests are possibly caused by combinations of multiple factors assigned with specific values, called faulty interactions. Martínez et al. [1] proposed the first deterministic adaptive algorithm for discovering faulty interactions involving at most two factors where each factor has two values, for which graph representations are adopted. In this paper, we improve Martínez et al.'s algorithm by an adaptive algorithmic approach for discovering faulty interactions in the so-called “non-2-locatable” graphs. We show that, for any system where each “non-2-locatable factor-component” involves two faulty interactions (for example, a system having at most two faulty interactions), our improved algorithm efficiently discovers all the faulty interactions with an extremely low mistaken probability caused by the random selection process in Martínez et al.'s algorithm. The effectiveness of our improved algorithm are revealed by both theoretical discussions and experimental evaluations.
In this paper, we study the number of failed components in a consecutive-k-out-of-n:G system. The distributions and expected values of the number of failed components when system is failed or working at a particular time t are evaluated. We also apply them to the optimization problems concerned with the optimal number of components and the optimal replacement time. Finally, we present the illustrative examples for the expected number of failed components and give the numerical results for the optimization problems.
Sho OBATA Koichi KOBAYASHI Yuh YAMASHITA
In the state estimation of steady-state power networks, a cyber attack that cannot be detected from the residual (i.e., the estimation error) is called a false data injection (FDI) attack. In this letter, to enforce the security of power networks, we propose a method of detecting an FDI attack. In the proposed method, an FDI attack is detected by randomly choosing sensors used in the state estimation. The effectiveness of the proposed method is presented by two examples including the IEEE 14-bus system.
Hitoshi SUDA Gaku KOTANI Daisuke SAITO
In this paper, we propose a new training framework named the INmfCA algorithm for nonparallel voice conversion (VC) systems. To train conversion models, traditional VC frameworks require parallel corpora, in which source and target speakers utter the same linguistic contents. Although the frameworks have achieved high-quality VC, they are not applicable in situations where parallel corpora are unavailable. To acquire conversion models without parallel corpora, nonparallel methods are widely studied. Although the frameworks achieve VC under nonparallel conditions, they tend to require huge background knowledge or many training utterances. This is because of difficulty in disentangling linguistic and speaker information without a large amount of data. In this work, we tackle this problem by exploiting NMF, which can factorize acoustic features into time-variant and time-invariant components in an unsupervised manner. The method acquires alignment between the acoustic features of a source speaker's utterances and a target dictionary and uses the obtained alignment as activation of NMF to train the source speaker's dictionary without parallel corpora. The acquisition method is based on the INCA algorithm, which obtains the alignment of nonparallel corpora. In contrast to the INCA algorithm, the alignment is not restricted to observed samples, and thus the proposed method can efficiently utilize small nonparallel corpora. The results of subjective experiments show that the combination of the proposed algorithm and the INCA algorithm outperformed not only an INCA-based nonparallel framework but also CycleGAN-VC, which performs nonparallel VC without any additional training data. The results also indicate that a one-shot VC framework, which does not need to train source speakers, can be constructed on the basis of the proposed method.
Zhi WENG Longzhen FAN Yong ZHANG Zhiqiang ZHENG Caili GONG Zhongyue WEI
As the basis of fine breeding management and animal husbandry insurance, individual recognition of dairy cattle is an important issue in the animal husbandry management field. Due to the limitations of the traditional method of cow identification, such as being easy to drop and falsify, it can no longer meet the needs of modern intelligent pasture management. In recent years, with the rise of computer vision technology, deep learning has developed rapidly in the field of face recognition. The recognition accuracy has surpassed the level of human face recognition and has been widely used in the production environment. However, research on the facial recognition of large livestock, such as dairy cattle, needs to be developed and improved. According to the idea of a residual network, an improved convolutional neural network (Res_5_2Net) method for individual dairy cow recognition is proposed based on dairy cow facial images in this letter. The recognition accuracy on our self-built cow face database (3012 training sets, 1536 test sets) can reach 94.53%. The experimental results show that the efficiency of identification of dairy cows is effectively improved.
In this study, we aim to improve the performance of audio source separation for monaural mixture signals. For monaural audio source separation, semisupervised nonnegative matrix factorization (SNMF) can achieve higher separation performance by employing small supervised signals. In particular, penalized SNMF (PSNMF) with orthogonality penalty is an effective method. PSNMF forces two basis matrices for target and nontarget sources to be orthogonal to each other and improves the separation accuracy. However, the conventional orthogonality penalty is based on an inner product and does not affect the estimation of the basis matrix properly because of the scale indeterminacy between the basis and activation matrices in NMF. To cope with this problem, a new PSNMF with cosine similarity between the basis matrices is proposed. The experimental comparison shows the efficacy of the proposed cosine similarity penalty in supervised audio source separation.
A variety of smart services are being provided on multiple virtual networks embedded into a common inter-cloud substrate network. The substrate network operator deploys critical substrate nodes so that multiple service providers can achieve enhanced services due to the secure sharing of their service data. Even if one of the critical substrate nodes incurs damage, resiliency of the enhanced services can be assured due to reallocation of the workload and periodic backup of the service data to the other normal critical substrate nodes. However, the connectivity of the embedded virtual networks must be maintained so that the enhanced services can be continuously provided to all clients on the virtual networks. This paper considers resilient virtual network embedding (VNE) that ensures the connectivity of the embedded virtual networks after critical substrate node failures have occurred. The resilient VNE problem is formulated using an integer linear programming model and a distance-based method is proposed to solve the large-scale resilient VNE problem efficiently. Simulation results demonstrate that the distance-based method can derive a sub-optimum VNE solution with a small computational effort. The method derived a VNE solution with an approximation ratio of less than 1.2 within ten seconds in all the simulation experiments.
Yutaro BESSHO Yuto HAYAMIZU Kazuo GODA Masaru KITSUREGAWA
Parallel processing is a typical approach to answer analytical queries on large database. As the size of the database increases, we often try to increase the parallelism by incorporating more processing nodes. However, this approach increases the possibility of node failure as well. According to the conventional practice, if a failure occurs during query processing, the database system restarts the query processing from the beginning. Such temporal cost may be unacceptable to the user. This paper proposes a fault-tolerant query processing mechanism, named PhoeniQ, for analytical parallel database systems. PhoeniQ continuously takes a checkpoint for every operator pipeline and replicates the output of each stateful operator among different processing nodes. If a single processing node fails during query processing, another can promptly take over the processing. Hence, PhoneniQ allows the database system to efficiently resume query processing after a partial failure event. This paper presents a key design of PhoeniQ and prototype-based experiments to demonstrate that PhoeniQ imposes negligible performance overhead and efficiently continues query processing in the face of node failure.
Chao LI Korkut Kaan TOKGOZ Ayuka OKUMURA Jim BARTELS Kazuhiro TODA Hiroaki MATSUSHIMA Takumi OHASHI Ken-ichi TAKEDA Hiroyuki ITO
Cow behavior monitoring is critical for understanding the current state of cow welfare and developing an effective planning strategy for pasture management, such as early detection of disease and estrus. One of the most powerful and cost-effective methods is a neural-network-based monitoring system that analyzes time series data from inertial sensors attached to cows. For this method, a significant challenge is to improve the quality and quantity of teaching data in the development of neural network models, which requires us to collect data that can cover various realistic conditions and assign labels to them. As a result, the cost of data collection is significantly high. This work proposes a data augmentation method to solve two major quality problems in the collection process of teaching data. One is the difficulty and randomicity of teaching data acquisition and the other is the sensor position changes during actual operation. The proposed method can computationally emulate different rotating states of the collar-type sensor device from the measured acceleration data. Furthermore, it generates data for actions that occur less frequently. The verification results showed significantly higher estimation performance with an average accuracy of over 98% for five main behaviors (feeding, walking, drinking, rumination, and resting) based on learning with long short-term memory (LSTM) network. Compared with the estimation performance without data augmentation, which was insufficient with a minimum of 60.48%, the recognition rate was improved by 2.52-37.05pt for various behaviors. In addition, comparison of different rotation intervals was investigated and a 30-degree increment was selected based on the accuracy performances analysis. In conclusion, the proposed data expansion method can improve the accuracy in cow behavior estimation by a neural network model. Moreover, it contributes to a significant reduction of the teaching data collection cost for machine learning and opens many opportunities for new research.
Jun SHIBAYAMA Takuma KURODA Junji YAMAUCHI Hisamatsu NAKANO
A periodic array of InSb spheres on a substrate is numerically analyzed at terahertz frequencies. The incident field is shown to be coupled to the substrate due to the guided-mode resonance. The effect of the background refractive index on the transmission characteristics is investigated for sensor applications.
Haotian CHEN Sukhoon LEE Di YAO Dongwon JEONG
High Frequency Surface Wave Radar (HFSWR) can achieve over-the-horizon detection, which can effectively detect and track the ships and ultra-low altitude aircrafts, as well as the acquisition of sea state information such as icebergs and ocean currents and so on. However, HFSWR is seriously affected by the clutters, especially sea clutter and ionospheric clutter. In this paper, we propose a deep learning image semantic segmentation method based on optimized Deeplabv3+ network to achieve the automatic detection of sea clutter and ionospheric clutter using the measured R-D spectrum images of HFSWR during the typhoon as experimental data, which avoids the disadvantage of traditional detection methods that require a large amount of a priori knowledge and provides a basis for subsequent the clutter suppression or the clutter characteristics research.
The energy efficiency of intelligent reflecting surface (IRS) enabled internet of things (IoT) networks is studied in this letter. The energy efficiency is mathematically expressed, respectively, as the number of reflecting elements and the spectral efficiency of the network and is shown to scale in the logarithm of the reflecting elements number in the high regime of transmit power from source node. Furthermore, it is revealed that the energy efficiency scales linearly over the spectral efficiency in the high regime of transmit power, in contrast to conventional studies on energy and spectral efficiency trade-offs in the non-IRS wireless IoT networks. Numerical simulations are carried out to verify the derived results for the IRS enabled IoT networks.
Kanghui ZHAO Tao LU Yanduo ZHANG Yu WANG Yuanzhi WANG
In recent years, compared with the traditional face super-resolution (SR) algorithm, the face SR based on deep neural network has shown strong performance. Among these methods, attention mechanism has been widely used in face SR because of its strong feature expression ability. However, the existing attention-based face SR methods can not fully mine the missing pixel information of low-resolution (LR) face images (structural prior). And they only consider a single attention mechanism to take advantage of the structure of the face. The use of multi-attention could help to enhance feature representation. In order to solve this problem, we first propose a new pixel attention mechanism, which can recover the structural details of lost pixels. Then, we design an attention fusion module to better integrate the different characteristics of triple attention. Experimental results on FFHQ data sets show that this method is superior to the existing face SR methods based on deep neural network.
Yuta FUKUDA Kota YOSHIDA Takeshi FUJINO
Deep learning applications have often been processed in the cloud or on servers. Still, for applications that require privacy protection and real-time processing, the execution environment is moved to edge devices. Edge devices that implement a neural network (NN) are physically accessible to an attacker. Therefore, physical attacks are a risk. Fault attacks on these devices are capable of misleading classification results and can lead to serious accidents. Therefore, we focus on the softmax function and evaluate a fault attack using a clock glitch against NN implemented in an 8-bit microcontroller. The clock glitch is used for fault injection, and the injection timing is controlled by monitoring the power waveform. The specific waveform is enrolled in advance, and the glitch timing pulse is generated by the sum of absolute difference (SAD) matching algorithm. Misclassification can be achieved by appropriately injecting glitches triggered by pattern detection. We propose a countermeasure against fault injection attacks that utilizes the randomization of power waveforms. The SAD matching is disabled by random number initialization on the summation register of the softmax function.
Jion HIROSE Junya NAKAMURA Fukuhito OOSHITA Michiko INOUE
We study the gathering problem requiring a team of mobile agents to gather at a single node in arbitrary networks. The team consists of k agents with unique identifiers (IDs), and f of them are weakly Byzantine agents, which behave arbitrarily except falsifying their identifiers. The agents move in synchronous rounds and cannot leave any information on nodes. If the number of nodes n is given to agents, the existing fastest algorithm tolerates any number of weakly Byzantine agents and achieves gathering with simultaneous termination in O(n4·|Λgood|·X(n)) rounds, where |Λgood| is the length of the maximum ID of non-Byzantine agents and X(n) is the number of rounds required to explore any network composed of n nodes. In this paper, we ask the question of whether we can reduce the time complexity if we have a strong team, i.e., a team with a few Byzantine agents, because not so many agents are subject to faults in practice. We give a positive answer to this question by proposing two algorithms in the case where at least 4f2+9f+4 agents exist. Both the algorithms assume that the upper bound N of n is given to agents. The first algorithm achieves gathering with non-simultaneous termination in O((f+|&Lambdagood|)·X(N)) rounds. The second algorithm achieves gathering with simultaneous termination in O((f+|&Lambdaall|)·X(N)) rounds, where |&Lambdaall| is the length of the maximum ID of all agents. The second algorithm significantly reduces the time complexity compared to the existing one if n is given to agents and |&Lambdaall|=O(|&Lambdagood|) holds.
Yasuaki KOBAYASHI Shin-ichi NAKANO Kei UCHIZAWA Takeaki UNO Yutaro YAMAGUCHI Katsuhisa YAMANAKA
Given a set P of n points and an integer k, we wish to place k facilities on points in P so that the minimum distance between facilities is maximized. The problem is called the k-dispersion problem, and the set of such k points is called a k-dispersion of P. Note that the 2-dispersion problem corresponds to the computation of the diameter of P. Thus, the k-dispersion problem is a natural generalization of the diameter problem. In this paper, we consider the case of k=3, which is the 3-dispersion problem, when P is in convex position. We present an O(n2)-time algorithm to compute a 3-dispersion of P.