Fairuz Safwan MAHAD Masakazu IWAMURA Koichi KISE
Neural network-based three-dimensional (3D) reconstruction methods have produced promising results. However, they do not pay particular attention to reconstructing detailed parts of objects. This occurs because the network is not designed to capture the fine details of objects. In this paper, we propose a network designed to capture both the coarse and fine details of objects to improve the reconstruction of the fine parts of objects.
Object contour detection is a task of extracting the shape created by the boundaries between objects in an image. Conventional methods limit the detection targets to specific categories, or miss-detect edges of patterns inside an object. We propose a new method to represent a contour image where the pixel value is the distance to the boundary. Contour detection becomes a regression problem that estimates this contour image. A deep convolutional network for contour estimation is combined with stereo vision to detect unspecified object contours. Furthermore, thanks to similar inference targets and common network structure, we propose a network that simultaneously estimates both contour and disparity with fully shared weights. As a result of experiments, the multi-tasking network drew a good precision-recall curve, and F-measure was about 0.833 for FlyingThings3D dataset. L1 loss of disparity estimation for the dataset was 2.571. This network reduces the amount of calculation and memory capacity by half, and accuracy drop compared to the dedicated networks is slight. Then we quantize both weights and activations of the network to 3-bit. We devise a dedicated hardware architecture for the quantized CNN and implement it on an FPGA. This circuit uses only internal memory to perform forward propagation calculations, that eliminates high-power external memory accesses. This circuit is a stall-free pixel-by-pixel pipeline, and performs 8 rows, 16 input channels, 16 output channels, 3 by 3 pixels convolution calculations in parallel. The convolution calculation performance at the operating frequency of 250 MHz is 9 TOPs/s.
Xiaoyu CHEN Huanchang LI Yihan ZHANG Yubo LI
A new construction of shift sequences is proposed under the condition of P|L, and then the inter-group complementary (IGC) sequence sets are constructed based on the shift sequence. By adjusting the parameter q, two or three IGC sequence sets can be obtained. Compared with previous methods, the proposed construction can provide more sequence sets for both synchronous and asynchronous code-division multiple access communication systems.
Naoki HATTORI Jun SHIOMI Yutaka MASUDA Tohru ISHIHARA Akihiko SHINYA Masaya NOTOMI
With the rapid progress of the integrated nanophotonics technology, the optical neural network architecture has been widely investigated. Since the optical neural network can complete the inference processing just by propagating the optical signal in the network, it is expected more than one order of magnitude faster than the electronics-only implementation of artificial neural networks (ANN). In this paper, we first propose an optical vector-matrix multiplication (VMM) circuit using wavelength division multiplexing, which enables inference processing at the speed of light with ultra-wideband. This paper next proposes optoelectronic circuit implementation for batch normalization and activation function, which significantly improves the accuracy of the inference processing without sacrificing the speed performance. Finally, using a virtual environment for machine learning and an optoelectronic circuit simulator, we demonstrate the ultra-fast and accurate operation of the optical-electronic ANN circuit.
Jin HOKI Kosei SAKAMOTO Kazuhiko MINEMATSU Takanori ISOBE
In this paper, we explore the security against integral attacks on well-known stream ciphers SNOW 3G and KCipher-2. SNOW 3G is the core of the 3GPP confidentiality and integrity algorithms UEA2 and UIA2, and KCipher-2 is a standard algorithm of ISO/IEC 18033-4 and CRYPTREC. Specifically, we investigate the propagation of the division property inside SNOW 3G and KCipher-2 by the Mixed-Integer Linear Programming to efficiently find an integral distinguisher. As a result, we present a 7-round integral distinguisher with 23 chosen IVs for KCipher-2. As far as we know, this is the first attack on a reduced variant of KCipher-2 by the third party. In addition, we present a 13-round integral distinguisher with 27 chosen IVs for SNOW 3G, whose time/data complexity is half of the previous best attack by Biryukov et al.
Kyohei ONO Shoichiro YAMASAKI Shinichiro MIYAZAKI Tomoko K. MATSUSHIMA
Optical code-division multiple-access (CDMA) techniques provide multi-user data transmission services in optical wireless and fiber communication systems. Several signature codes, such as modified prime sequence codes (MPSCs), generalized MPSCs (GMPSCs) and modified pseudo-orthogonal M-sequence sets, have been proposed for synchronous optical CDMA systems. In this paper, a new scheme is proposed for synchronous optical CDMA to increase the number of users and, consequently, to increase the total data rate without increasing the chip rate. The proposed scheme employs a GMPSC and an extended bi-orthogonal code which is a unipolar code generated from a bipolar Walsh code. Comprehensive comparisons between the proposed scheme and several conventional schemes are shown. Moreover, bit error rate performance and energy efficiency of the proposed scheme are evaluated comparing with those of the conventional optical CDMA schemes under atmospheric propagation environment.
Yufeng CHEN Siqi LI Xingya LI Jinan XU Jian LIU
Relation extraction is one of the key basic tasks in natural language processing in which distant supervision is widely used for obtaining large-scale labeled data without expensive labor cost. However, the automatically generated data contains massive noise because of the wrong labeling problem in distant supervision. To address this problem, the existing research work mainly focuses on removing sentence-level noise with various sentence selection strategies, which however could be incompetent for disposing word-level noise. In this paper, we propose a novel neural framework considering both intra-sentence and inter-sentence relevance to deal with word-level and sentence-level noise from distant supervision, which is denoted as Sentence-Related Gated Piecewise Convolutional Neural Networks (SR-GPCNN). Specifically, 1) a gate mechanism with multi-head self-attention is adopted to reduce word-level noise inside sentences; 2) a soft-label strategy is utilized to alleviate wrong-labeling propagation problem; and 3) a sentence-related selection model is designed to filter sentence-level noise further. The extensive experimental results on NYT dataset demonstrate that our approach filters word-level and sentence-level noise effectively, thus significantly outperforms all the baseline models in terms of both AUC and top-n precision metrics.
Yang YAN Yao YAO Zhi CHEN Qiuyan WANG
Codebooks with small inner-product correlation have applied in direct spread code division multiple access communications, space-time codes and compressed sensing. In general, it is difficult to construct optimal codebooks achieving the Welch bound or the Levenstein bound. This paper focuses on constructing asymptotically optimal codebooks with characters of cyclic groups. Based on the proposed constructions, two classes of asymptotically optimal codebooks with respect to the Welch bound are presented. In addition, parameters of these codebooks are new.
Yusuke HARA Xueting WANG Toshihiko YAMASAKI
Video inpainting is a task of filling missing regions in videos. In this task, it is important to efficiently use information from other frames and generate plausible results with sufficient temporal consistency. In this paper, we present a video inpainting method jointly using affine transformation and deformable convolutions for frame alignment. The former is responsible for frame-scale rough alignment and the latter performs pixel-level fine alignment. Our model does not depend on 3D convolutions, which limits the temporal window, or troublesome flow estimation. The proposed method achieves improved object removal results and better PSNR and SSIM values compared with previous learning-based methods.
Takumi UCHIDA Keisuke ISHIBASHI Kensuke FUKUDA
This paper introduces a method to estimate latent traffic from its origin to destination from the link packet loss rate and traffic volume. In addition, we propose a method for the joint optimization of routing and link provisioning based on the estimated latent traffic. Observed traffic could deviate from the original traffic demand and become latent when the traffic passes through congested links because of changes in user behavioral and/or applications as a result of degraded quality of experience (QoE). The latent traffic is actualized by improving congested link capacity. When link provisioning is based on observed traffic, actual traffic might cause new congestion at other links. Thus, network providers need to estimate the origin-destination (OD) original traffic demand for network planning. Although the estimation of original traffic has been well studied, the estimation was only applicable for links. In this paper, we propose a method to estimate latent OD traffic by combining and expanding techniques. The method consists of three steps. The first step is to estimate the actual OD traffic and loss rate from the actual traffic and packet loss rate of the links. The second step is to estimate the latent traffic demand. Finally, using this estimated demand, the link capacity and routing matrix are optimized. We evaluate our method by simulation and confirm that congestion could be avoided by capacity provisioning based on estimated latent traffic, while provisioning based on observed traffic retains the congestion. The combined method can avoid congestion with an increment of 23% compared with capacity provisioning only. We also evaluated our method's adaptability, i.e., the ability to estimate the required parameter for the estimations using fewer given values, but values obtained in the environment.
Fumihiro YAMASHITA Daisuke GOTO Yasuyoshi KOJIMA Jun-ichi ABE Takeshi ONIZAWA
We have developed a direct spectrum division transmission (DSDT) technique that can divide a single-carrier signal into multiple sub-spectra and assign them to dispersed frequency resources of the satellite transponder to improve the spectrum efficiency of the whole system. This paper summarizes the satellite experiments on DSDT over a single and/or multiple satellite transponders, while changing various parameters such as modulation schemes, roll-off ratios, and symbol rates. In addition, by considering practical use conditions, we present an evaluation of the performance when the spectral density of each sub-spectrum differed across transponders. The satellite experiments demonstrate that applying the proposal does not degrade the bit error rate (BER) performance. Thus, the DSDT technique is a practical approach to use the scattered unused frequency resources over not only a single transponder but also multiple ones.
Xiaoran CHEN Xin QIU Xurong CHAI
Orthogonal frequency division multiplexing (OFDM) technique has been widely used in communication systems in pursuit of the most efficient utilization of spectrum. However, the increase of the number of orthogonal subcarriers will lead to the rise of the peak-to-average power ratio (PAPR) of the waveform, thus reducing the efficiency of the power amplifiers. In this letter we propose a phase-changed PAPR reduction technique based on windowing function architecture for OFDM systems. This technique is based on the idea of phase change, which makes the spectrum of output signal almost free of regrowth caused by peak clipping. It can reduce more than 28dBc adjacent channel power ratio (ACPR) compared with the traditional peak windowing clipping methods in situation that peak is maximally suppressed. This technique also has low algorithm complexity so it can be easily laid out on hardware. The proposed algorithm has been laid out on a low-cost field-programmable gate array (FPGA) to verify its effectiveness and feasibility. A 64-QAM modulated 20M LTE-A waveform is used for measurement, which has a sampling rate of 245.67M.
To reduce peak-to-average power ratio, we propose a method of choosing suitable vectors in a partial transmit sequence technique. Conventional approaches require that a suitable vector be selected from a large number of candidates. By contrast, our method does not include such a selecting procedure, and instead generates random vectors from the Gaussian distribution whose covariance matrix is a solution of a relaxed problem. The suitable vector is chosen from the random vectors. This yields lower peak-to-average power ratio than a conventional method.
Fang LIU Kenneth W. SHUM Yijin ZHANG Wing Shing WONG
We consider all-to-all broadcast and unicast among nodes in a multi-channel single-hop ad hoc network, with no time synchronization. Motivated by the hard delay requirement for ultra-reliable and low-latency communication (URLLC) in 5G wireless networks, we aim at designing medium access control (MAC) schemes to guarantee successful node-to-node transmission within a bounded delay. To provide a hard guarantee on the transmission delay, deterministic sequence schemes are preferred to probabilistic schemes such as carrier sense multiple access (CSMA). Therefore, we mainly consider sequence schemes, with the goal to design schedule sequence set to guarantee successful broadcast/unicast within a common sequence period. This period should be as short as possible since it determines an upper bound on the transmission delay. In previous works, we have considered sequence design under time division duplex (TDD). In this paper, we focus on another common duplex mode, frequency division duplex (FDD). For the FDD case, we present a lower bound on period of feasible sequence sets, and propose a sequence construction method by which the sequence period can achieve the same order as the lower bound, for both broadcast and unicast models. We also compare the sequence length for FDD with that for TDD.
Takahiro MATSUMOTO Hideyuki TORII Yuta IDA Shinya MATSUFUJI
In this paper, we propose new generation methods of two-dimensional (2D) optical zero-correlation zone (ZCZ) sequences with the high peak autocorrelation amplitude. The 2D optical ZCZ sequence consists of a pair of a binary sequence which takes 1 or 0 and a bi-phase sequence which takes 1 or -1, and has a zero-correlation zone in the two-dimensional correlation function. Because of these properties, the 2D optical ZCZ sequence is suitable for optical code-division multiple access (OCDMA) system using an LED array having a plurality of light-emitting elements arranged in a lattice pattern. The OCDMA system using the 2D optical ZCZ sequence can be increased the data rate and can be suppressed interference by the light of adjacent LEDs. By using the proposed generation methods, we can improve the peak autocorrelation amplitude of the sequence. This means that the BER performance of the OCDMA system using the sequence can be improved.
Longfei CHEN Yuichi NAKAMURA Kazuaki KONDO Dima DAMEN Walterio MAYOL-CUEVAS
We propose a novel framework for integrating beginners' machine operational experiences with those of experts' to obtain a detailed task model. Beginners can provide valuable information for operation guidance and task design; for example, from the operations that are easy or difficult for them, the mistakes they make, and the strategy they tend to choose. However, beginners' experiences often vary widely and are difficult to integrate directly. Thus, we consider an operational experience as a sequence of hand-machine interactions at hotspots. Then, a few experts' experiences and a sufficient number of beginners' experiences are unified using two aggregation steps that align and integrate sequences of interactions. We applied our method to more than 40 experiences of a sewing task. The results demonstrate good potential for modeling and obtaining important properties of the task.
Expectation propagation (EP) decoding is proposed for sparse superposition coding in orthogonal frequency division multiplexing (OFDM) systems. When a randomized discrete Fourier transform (DFT) dictionary matrix is used, the EP decoding has the same complexity as approximate message-passing (AMP) decoding, which is a low-complexity and powerful decoding algorithm for the additive white Gaussian noise (AWGN) channel. Numerical simulations show that the EP decoding achieves comparable performance to AMP decoding for the AWGN channel. For OFDM systems, on the other hand, the EP decoding is much superior to the AMP decoding while the AMP decoding has an error-floor in high signal-to-noise ratio regime.
Gil-Mo KANG Cheolsoo PARK Oh-Soon SHIN
We propose an optimal power allocation scheme that maximizes the transmission rate of device-to-device (D2D) communications underlaying a cellular system based on orthogonal frequency division multiplexing (OFDM). The proposed algorithm first calculates the maximum allowed transmission power of a D2D transmitter to restrict the interference caused to a cellular link that share the same OFDM subchannels with the D2D link. Then, with a constraint on the maximum transmit power, an optimization of water-filling type is performed to find the optimal transmit power allocation across subchannels and within each subchannel. The performance of the proposed power allocation scheme is evaluated in terms of the average achievable rate of the D2D link.
Yuto SAGAE Takashi MATSUI Taiji SAKAMOTO Kazuhide NAKAJIMA
We propose an ultra-low inter-core crosstalk (XT) multi-core fiber (MCF) with standard 125-μm cladding. We show the fiber design and fabrication results of an MCF housing four cores with W-shaped index profile; it offers XT of less than -67dB/km over the whole C+L band. This enables us to realize 10,000-km transmission with negligible XT penalty. We also observe a low-loss of 0.17dB/km (average) at a wavelength of 1.55μm and other optical properties compatible with ITU-T G.654.B fiber. We also elucidate its good micro-bend resistance in terms of both the loss and XT to confirm its applicability to high-density optical fiber cables. Finally, we show that the fabricated MCF is feasible along with long-distance transmission by confirming that the XT noise performance corresponds to transmission distances of 10,000km or more.
Ryota TSUJI Daisuke HISANO Ken MISHINA Akihiro MARUTA
Wavelength division multiplexing (WDM) scheme is used widely in photonic metro-core networks. In a WDM network, wavelength continuity constraint is employed to simply construct relay nodes. This constraint reduces the wavelength usage efficiency of each link. To improve the same, an all-optical wavelength converter (AO-WC) has been attracting attention in recent years. In particular, an AO-WC is a key device because it enables simultaneous conversion of multiple wavelengths of signal lights to other wavelengths, independent of the modulation format. However, each AO-WC requires installation of multiple laser sources with narrow bandwidth because the lights emitted by the laser sources are used as pump lights when the wavelengths of the signal lights are converted by the four-wave mixing (FWM) process. To reduce the number of laser sources, we propose a remote pumped AO-WC, in which the laser sources of the pump lights are aggregated into several relay nodes. When the request for the wavelength conversion from the relay node without the laser source is conveyed, the relay node with the laser source transmits the pump light through the optical link. The proposed scheme enables reduction in the number of laser sources of the pump lights. Herein we analyze the distortion of the pump light by propagating it through the optical link We also evaluate the effect of the noise in optical amplifiers and nonlinearities in optical fibers using numerical simulations employing the representative parameters for a practical WDM network.