Tsutomu SASAO Yuto HORIKAWA Yukihiro IGUCHI
A classification function maps a set of vectors into several classes. A machine learning problem is treated as a design problem for partially defined classification functions. To realize classification functions for MNIST hand written digits, three different architectures are considered: Single-unit realization, 45-unit realization, and 45-unit ×r realization. The 45-unit realization consists of 45 ternary classifiers, 10 counters, and a max selector. Test accuracy of these architectures are compared using MNIST data set.
Xueqing ZHANG Xiaoxia LIU Jun GUO Wenlei BAI Daguang GAN
As scientific and technological resources are experiencing information overload, it is quite expensive to find resources that users are interested in exactly. The personalized recommendation system is a good candidate to solve this problem, but data sparseness and the cold starting problem still prevent the application of the recommendation system. Sparse data affects the quality of the similarity measurement and consequently the quality of the recommender system. In this paper, we propose a matrix factorization recommendation algorithm based on similarity calculation(SCMF), which introduces potential similarity relationships to solve the problem of data sparseness. A penalty factor is adopted in the latent item similarity matrix calculation to capture more real relationships furthermore. We compared our approach with other 6 recommendation algorithms and conducted experiments on 5 public data sets. According to the experimental results, the recommendation precision can improve by 2% to 9% versus the traditional best algorithm. As for sparse data sets, the prediction accuracy can also improve by 0.17% to 18%. Besides, our approach was applied to patent resource exploitation provided by the wanfang patents retrieval system. Experimental results show that our method performs better than commonly used algorithms, especially under the cold starting condition.
Keiichiro SATO Ryoichi SHINKUMA Takehiro SATO Eiji OKI Takanori IWAI Takeo ONISHI Takahiro NOBUKIYO Dai KANETOMO Kozo SATODA
Predictive spatial-monitoring, which predicts spatial information such as road traffic, has attracted much attention in the context of smart cities. Machine learning enables predictive spatial-monitoring by using a large amount of aggregated sensor data. Since the capacity of mobile networks is strictly limited, serious transmission delays occur when loads of communication traffic are heavy. If some of the data used for predictive spatial-monitoring do not arrive on time, prediction accuracy degrades because the prediction has to be done using only the received data, which implies that data for prediction are ‘delay-sensitive’. A utility-based allocation technique has suggested modeling of temporal characteristics of such delay-sensitive data for prioritized transmission. However, no study has addressed temporal model for prioritized transmission in predictive spatial-monitoring. Therefore, this paper proposes a scheme that enables the creation of a temporal model for predictive spatial-monitoring. The scheme is roughly composed of two steps: the first involves creating training data from original time-series data and a machine learning model that can use the data, while the second step involves modeling a temporal model using feature selection in the learning model. Feature selection enables the estimation of the importance of data in terms of how much the data contribute to prediction accuracy from the machine learning model. This paper considers road-traffic prediction as a scenario and shows that the temporal models created with the proposed scheme can handle real spatial datasets. A numerical study demonstrated how our temporal model works effectively in prioritized transmission for predictive spatial-monitoring in terms of prediction accuracy.
Takahiro MATSUDA Fumie ONO Shinsuke HARA
In wireless links between ground stations and UAVs (Unmanned Aerial Vehicles), wireless signals may be attenuated by obstructions such as buildings. A three-dimensional RSS (Received Signal Strength) map (3D-RSS map), which represents a set of RSSs at various reception points in a three-dimensional area, is a promising geographical database that can be used to design reliable ground-to-air wireless links. The construction of a 3D-RSS map requires higher computational complexity, especially for a large 3D area. In order to sequentially estimate a 3D-RSS map from partial observations of RSS values in the 3D area, we propose a graph Laplacian-based sequential smooth estimator. In the proposed estimator, the 3D area is divided into voxels, and a UAV observes the RSS values at the voxels along a predetermined path. By considering the voxels as vertices in an undirected graph, a measurement graph is dynamically constructed using vertices from which recent observations were obtained and their neighboring vertices, and the 3D-RSS map is sequentially estimated by performing graph Laplacian regularized least square estimation.
Takaaki SAEKI Yuki SAITO Shinnosuke TAKAMICHI Hiroshi SARUWATARI
This paper proposes two high-fidelity and computationally efficient neural voice conversion (VC) methods based on a direct waveform modification using spectral differentials. The conventional spectral-differential VC method with a minimum-phase filter achieves high-quality conversion for narrow-band (16 kHz-sampled) VC but requires heavy computational cost in filtering. This is because the minimum phase obtained using a fixed lifter of the Hilbert transform often results in a long-tap filter. Furthermore, when we extend the method to full-band (48 kHz-sampled) VC, the computational cost is heavy due to increased sampling points, and the converted-speech quality degrades due to large fluctuations in the high-frequency band. To construct a short-tap filter, we propose a lifter-training method for data-driven phase reconstruction that trains a lifter of the Hilbert transform by taking into account filter truncation. We also propose a frequency-band-wise modeling method based on sub-band multi-rate signal processing (sub-band modeling method) for full-band VC. It enhances the computational efficiency by reducing sampling points of signals converted with filtering and improves converted-speech quality by modeling only the low-frequency band. We conducted several objective and subjective evaluations to investigate the effectiveness of the proposed methods through implementation of the real-time, online, full-band VC system we developed, which is based on the proposed methods. The results indicate that 1) the proposed lifter-training method for narrow-band VC can shorten the tap length to 1/16 without degrading the converted-speech quality, and 2) the proposed sub-band modeling method for full-band VC can improve the converted-speech quality while reducing the computational cost, and 3) our real-time, online, full-band VC system can convert 48 kHz-sampled speech in real time attaining the converted speech with a 3.6 out of 5.0 mean opinion score of naturalness.
The purpose of this paper is to find an automated pricing algorithm to calculate the real cost of each product by considering the associate costs of the business. The methodology consists of two main stages. A brief semi-structured survey and a mathematical calculation the expenses and adding them to the original cost of the offered products and services. The output of this process obtains the minimum recommended selling price (MRSP) that the business should not go below, to increase the likelihood of generating profit and avoiding the unexpected loss. The contribution of this study appears in filling the gap by calculating the minimum recommended price automatically and assisting businesses to foresee future budgets. This contribution has a certain limitation, where it is unable to calculate the MRSP of the in-house created products from raw materials. It calculates the MRSP only for the products bought from the wholesaler to be sold by the retailer.
Yucong ZHANG Stefan HOLST Xiaoqing WEN Kohei MIYASE Seiji KAJIHARA Jun QIAN
Loading test vectors and unloading test responses in shift mode during scan testing cause many scan flip-flops to switch simultaneously. The resulting shift switching activity around scan flip-flops can cause excessive local IR-drop that can change the states of some scan flip-flops, leading to test data corruption. A common approach solving this problem is partial-shift, in which multiple scan chains are formed and only one group of the scan chains is shifted at a time. However, previous methods based on this approach use random grouping, which may reduce global shift switching activity, but may not be optimized to reduce local shift switching activity, resulting in remaining high risk of test data corruption even when partial-shift is applied. This paper proposes novel algorithms (one optimal and one heuristic) to group scan chains, focusing on reducing local shift switching activity around scan flip-flops, thus reducing the risk of test data corruption. Experimental results on all large ITC'99 benchmark circuits demonstrate the effectiveness of the proposed optimal and heuristic algorithms as well as the scalability of the heuristic algorithm.
Manlin XIAO Zhibo DUAN Zhenglong YANG
Based on TLS-ESPRIT algorithm, this paper proposes a weighted spatial smoothing DOA estimation algorithm to address the problem that the conventional TLS-ESPRIT algorithm will be disabled to estimate the direction of arrival (DOA) in the scenario of coherent sources. The proposed method divides the received signal array into several subarrays with special structural feature. Then, utilizing these subarrays, this paper constructs the new weighted covariance matrix to estimate the DOA based on TLS-ESPRIT. The auto-correlation and cross-correlation information of subarrays in the proposed algorithm is extracted sufficiently, improving the orthogonality between the signal subspace and the noise subspace so that the DOA of coherent sources could be estimated accurately. The simulations show that the proposed algorithm is superior to the conventional spatial smoothing algorithms under different signal to noise ratio (SNR) and snapshot numbers with coherent sources.
Kijung JUNG Hyukki LEE Yon Dohn CHUNG
Deep learning has shown outstanding performance in various fields, and it is increasingly deployed in privacy-critical domains. If sensitive data in the deep learning model are exposed, it can cause serious privacy threats. To protect individual privacy, we propose a novel activation function and stochastic gradient descent for applying differential privacy to deep learning. Through experiments, we show that the proposed method can effectively protect the privacy and the performance of proposed method is better than the previous approaches.
Leilei KONG Yong HAN Haoliang QI Zhongyuan HAN
Source retrieval is the primary task of plagiarism detection. It searches the documents that may be the sources of plagiarism to a suspicious document. The state-of-the-art approaches usually rely on the classical information retrieval models, such as the probability model or vector space model, to get the plagiarism sources. However, the goal of source retrieval is to obtain the source documents that contain the plagiarism parts of the suspicious document, rather than to rank the documents relevant to the whole suspicious document. To model the “partial matching” between documents, this paper proposes a Partial Matching Convolution Neural Network (PMCNN) for source retrieval. In detail, PMCNN exploits a sequential convolution neural network to extract the plagiarism patterns of contiguous text segments. The experimental results on PAN 2013 and PAN 2014 plagiarism source retrieval corpus show that PMCNN boosts the performance of source retrieval significantly, outperforming other state-of-the-art document models.
Haisong JIANG Mahmoud NASEF Kiichi HAMAMOTO
This paper reports a single dimensional mode based multiplexer / de-multiplexer using the slab waveguide to realize high modes multiplexing and high integration in the non-MIMO (multi-in multi-out) multimode transmission system. A sufficient mode crosstalk of -20 dB was obtained by selecting suitable parameters of the spacing between the connecting positions of each arrayed waveguide Di, the radius slab waveguide R0 and lateral V-parameter.
Kwenga ISMAEL MUNENE Nobuo FUNABIKI Hendy BRIANTORO Md. MAHBUBUR RAHMAN Fatema AKHTER Minoru KURIBAYASHI Wen-Chung KAO
Currently, the IEEE 802.11n wireless local-area network (WLAN) has been extensively deployed world-wide. For the efficient channel assignment to access-points (APs) from the limited number of partially overlapping channels (POCs) at 2.4GHz band, we have studied the throughput drop estimation model for concurrently communicating links using the channel bonding (CB). However, non-CB links should be used in dense WLANs, since the CB links often reduce the transmission capacity due to high interferences from other links. In this paper, we examine the throughput drop estimation model for concurrently communicating links without using the CB in 802.11n WLAN, and its application to the POC assignment to the APs. First, we verify the model accuracy through experiments in two network fields. The results show that the average error is 9.946% and 6.285% for the high and low interference case respectively. Then, we verify the effectiveness of the POC assignment to the APs using the model through simulations and experiments. The results show that the model improves the smallest throughput of a host by 22.195% and the total throughput of all the hosts by 22.196% on average in simulations for three large topologies, and the total throughput by 12.89% on average in experiments for two small topologies.
This paper addresses pilot contamination in massive multiple-input multiple-output (MIMO) uplink. Pilot contamination is caused by reuse of identical pilot sequences in adjacent cells. To solve pilot contamination, the base station utilizes differences between the transmission frames of different users, which are detected via joint channel and data estimation. The joint estimation is regarded as a bilinear inference problem in compressed sensing. Expectation propagation (EP) is used to propose an iterative channel and data estimation algorithm. Initial channel estimates are attained via time-shifted pilots without exploiting information about large scale fading. The proposed EP modifies two points in conventional bilinear adaptive vector approximate message-passing (BAd-VAMP). One is that EP utilizes data estimates after soft decision in the channel estimation while BAd-VAMP uses them before soft decision. The other point is that EP can utilize the prior distribution of the channel matrix while BAd-VAMP cannot in principle. Numerical simulations show that EP converges much faster than BAd-VAMP in spatially correlated MIMO, in which approximate message-passing fails to converge toward the same fixed-point as EP and BAd-VAMP.
Chenyu WANG Kengo IOKIBE Yoshitaka TOYOTA
The plain bend in a pair of differential transmission lines causes a path difference, which leads to differential-to-common mode conversion due to the phase difference. This conversion can cause serious common-mode noise issues. We previously proposed a tightly coupled asymmetrically tapered bend to suppress forward differential-to-common mode conversion and derived the constraint conditions for high-density wiring. To provide sufficient suppression of mode conversion, however, the additional correction was required to make the effective path difference vanish. This paper proposes a practical and straightforward design methodology by using a very tightly coupled bend (decreasing the line width and the line separation of the tightly coupled bend). Full-wave simulations below 20GHz demonstrated that sufficient suppression of the forward differential-to-common mode conversion is successfully achieved as designed. Measurements showed that our design methodology is effective.
Tatsuya SUGIYAMA Keigo TAKEUCHI
Sparse orthogonal matrices are proposed to improve the convergence property of expectation propagation (EP) for sparse signal recovery from compressed linear measurements subject to known dense and ill-conditioned multiplicative noise. As a typical problem, this letter addresses generalized spatial modulation (GSM) in over-loaded and spatially correlated multiple-input multiple-output (MIMO) systems. The proposed sparse orthogonal matrices are used in precoding and constructed efficiently via a generalization of the fast Walsh-Hadamard transform. Numerical simulations show that the proposed sparse orthogonal precoding improves the convergence property of EP in over-loaded GSM MIMO systems with known spatially correlated channel matrices.
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.
The circuit satisfiability problem has been intensively studied since Ryan Williams showed a connection between the problem and lower bounds for circuit complexity. In this letter, we present a #SAT algorithm for synchronous Boolean circuits of n inputs and s gates in time $2^{nleft(1 - rac{1}{2^{O(s/n)}} ight)}$ if s=o(n log n).
Mingxing ZHANG Zhengchun ZHOU Meng YANG Haode YAN
The partial-period autocorrelation of sequences is an important performance measure of communication systems employing them, but it is notoriously difficult to be analyzed. In this paper, we propose an algorithm to design unimodular sequences with low partial-period autocorrelations via directly minimizing the partial-period integrated sidelobe level (PISL). The proposed algorithm is inspired by the monotonic minimizer for integrated sidelobe level (MISL) algorithm. Then an acceleration scheme is considered to further accelerate the algorithms. Numerical experiments show that the proposed algorithm can effectively generate sequences with lower partial-period peak sidelobe level (PPSL) compared with the well-known Zadoff-Chu sequences.
Giang-Truong NGUYEN Van-Quyet NGUYEN Van-Hau NGUYEN Kyungbaek KIM
In a smart home environment, sensors generate events whenever activities of residents are captured. However, due to some factors, abnormal events could be generated, which are technically reasonable but contradict to real-world activities. To detect abnormal events, a number of methods has been introduced, e.g., clustering-based or snapshot-based approaches. However, they have limitations to deal with complicated anomalies which occur with large number of events and blended within normal sensor readings. In this paper, we propose a novel method of detecting sensor anomalies under smart home environment by considering spatial correlation and dependable correlation between sensors. Initially, we pre-calculate these correlations of every pair of two sensors to discover their relations. Then, from periodic sensor readings, if it has any unmatched relations to the pre-computed ones, an anomaly is detected on the correlated sensor. Through extensive evaluations with real datasets, we show that the proposed method outperforms previous approaches with 20% improvement on detection rate and reasonably low false positive rate.
Guoqiang ZHANG Lingjin CAO Kosuke YAYAMA Akio KATSUSHIMA Takahiro MIKI
A differential on chip oscillator (OCO) is proposed in this paper for low supply voltage, high frequency accuracy and fast startup. The differential architecture helps the OCO achieve a good power supply rejection ratio (PSRR) without using a regulator so as to make the OCO suitable for a low power supply voltage of 1.38V. A reference voltage generator is also developed to generate two output voltages lower than Vbe for low supply voltage operation. The output frequency is locked to 48MHz by a frequency-locked loop (FLL) and a 3.3-ppm/°C temperature coefficient of frequency is realized by the differential voltage ratio adjusting (differential VRA) technique. The startup time is only 1.47μs because the differential OCO is not necessary to charge a big capacitor for ripple reduction.