Jana BACKHUS Ichigaku TAKIGAWA Hideyuki IMAI Mineichi KUDO Masanori SUGIMOTO
In this paper, we introduce a self-constructive Normalized Gaussian Network (NGnet) for online learning tasks. In online tasks, data samples are received sequentially, and domain knowledge is often limited. Then, we need to employ learning methods to the NGnet that possess robust performance and dynamically select an accurate model size. We revise a previously proposed localized forgetting approach for the NGnet and adapt some unit manipulation mechanisms to it for dynamic model selection. The mechanisms are improved for more robustness in negative interference prone environments, and a new merge manipulation is considered to deal with model redundancies. The effectiveness of the proposed method is compared with the previous localized forgetting approach and an established learning method for the NGnet. Several experiments are conducted for a function approximation and chaotic time series forecasting task. The proposed approach possesses robust and favorable performance in different learning situations over all testbeds.
Aravind THARAYIL NARAYANAN Wei DENG Dongsheng YANG Rui WU Kenichi OKADA Akira MATSUZAWA
An all-digital fully-synthesizable PVT-tolerant clock data recovery (CDR) architecture for wireline chip-to-chip interconnects is presented. The proposed architecture enables the co-synthesis of the CDR with the digital core. By eliminating the resource hungry manual layout and interfacing steps, which are necessary for conventional CDR topologies, the design process and the time-to-market can be drastically improved. Besides, the proposed CDR architecture enables the re-usability of majority of the sub-systems which enables easy migration to different process nodes. The proposed CDR is also equipped with a self-calibration scheme for ensuring tolerence over PVT. The proposed fully-syntehsizable CDR was implemented in 28nm FDSOI. The system achieves a maximum data rate of 10.06Gbps while consuming a power of 16.1mW from a 1V power supply.
Huaning WU Yalong YAN Chao LIU Jing ZHANG
This paper introduces and uses spider monkey optimization (SMO) for synthesis sparse linear arrays, which are composed of a uniformly spaced core subarray and an extended sparse subarray. The amplitudes of all the elements and the locations of elements in the extended sparse subarray are optimized by the SMO algorithm to reduce the side lobe levels of the whole array, under a set of practical constraints. To show the efficiency of SMO, different examples are presented and solved. Simulation results of the sparse arrays designed by SMO are compared with published results to verify the effectiveness of the SMO method.
In this paper, we present two classes of zero difference balanced (ZDB) functions, which are derived by difference balanced functions, and a class of perfect ternary sequences respectively. The proposed functions have parameters not covered in the literature, and can be used to design optimal constant composition codes, and perfect difference systems of sets.
The solution of the standard 2-norm-based multiple kernel regression problem and the theoretical limit of the considered model space are discussed in this paper. We prove that 1) The solution of the 2-norm-based multiple kernel regressor constructed by a given training data set does not generally attain the theoretical limit of the considered model space in terms of the generalization errors, even if the training data set is noise-free, 2) The solution of the 2-norm-based multiple kernel regressor is identical to the solution of the single kernel regressor under a noise free setting, in which the adopted single kernel is the sum of the same kernels used in the multiple kernel regressor; and it is also true for a noisy setting with the 2-norm-based regularizer. The first result motivates us to develop a novel framework for the multiple kernel regression problems which yields a better solution close to the theoretical limit, and the second result implies that it is enough to use the single kernel regressors with the sum of given multiple kernels instead of the multiple kernel regressors as long as the 2-norm based criterion is used.
Characteristics of the bis-styrylbenzene derivatives with trifluoromethyl or methyl moieties were evaluated in each as-vapor-deposited film, thermally-treated film, and the crystal from the solution. Thermal treatment dramatically changed morphologies and photo-physical properties of the vapor-deposited film.
Takafumi FUJIMOTO Takaya ISHIKUBO Masaya TAKAMURA
In this paper, a printed elliptical monopole antenna for wideband circular polarization is proposed. The antenna's structure is asymmetric with regard to the microstrip line. The section of the ground plane that overlaps the elliptical patch is removed. With simulations, the relationship between the antenna's geometrical parameters and the antenna's axial ratio of circularly polarized wave is clarified. The operational principle for wideband circular polarization is explained by the simulated electric current distributions. The simulated and measured bandwidths of the 3dB-axial ratio with a 2-VSWR is approximately 88.4% (2.12GHz-5.47GHz) and 83.6% (2.20GHz-5.36GHz), respectively.
Oussama DERBEL René LANDRY, Jr.
Driver behavior assessment is a hard task since it involves distinctive interconnected factors of different types. Especially in case of insurance applications, a trade-off between application cost and data accuracy remains a challenge. Data uncertainty and noises make smart-phone or low-cost sensor platforms unreliable. In order to deal with such problems, this paper proposes the combination between the Belief and Fuzzy theories with a two-level fusion based architecture. It enables the propagation of information errors from the lower to the higher level of fusion using the belief and/or the plausibility functions at the decision step. The new developed risk models of the Driver and Environment are based on the accident statistics analysis regarding each significant driving risk parameter. The developed Vehicle risk models are based on the longitudinal and lateral accelerations (G-G diagram) and the velocity to qualify the driving behavior in case of critical events (e.g. Zig-Zag scenario). In case of over-speed and/or accident scenario, the risk is evaluated using our new developed Fuzzy Inference System model based on the Equivalent Energy Speed (EES). The proposed approach and risk models are illustrated by two examples of driving scenarios using the CarSim vehicle simulator. Results have shown the validity of the developed risk models and the coherence with the a-priori risk assessment.
Xiaoxia DAI Wei XIA Wenlong HE
Much attention has recently been paid to sparsity-aware space-time adaptive processing (STAP) algorithms. The idea of sparsity-aware technology is commonly based on the convex l1-norm penalty. However, some works investigate the lq (0 < q < 1) penalty which induces more sparsity owing to its lack of convexity. We herein consider the design of an lq penalized STAP processor with a generalized sidelobe canceler (GSC) architecture. The lq cyclic descent (CD) algorithm is utilized with the least squares (LS) design criterion. It is validated through simulations that the lq penalized STAP processor outperforms the existing l1-based counterparts in both convergence speed and steady-state performance.
Kyohei NAKAJIMA Koichi KOBAYASHI Yuh YAMASHITA
Event-triggered control is a control method that the measured signal is sent to the controller only when a certain triggering condition on the measured signal is satisfied. In this paper, we propose a linear quadratic regulator (LQR) with decentralized triggering conditions. First, a suboptimal solution to the design problem of LQRs with decentralized triggering conditions is derived. A state-feedback gain can be obtained by solving a convex optimization problem with LMI (linear matrix inequality) constraints. Next, the relation between centralized and decentralized triggering conditions is discussed. It is shown that control performance of an LQR with decentralized event-triggering is better than that with centralized event-triggering. Finally, a numerical example is illustrated.
David WONG Daisuke DEGUCHI Ichiro IDE Hiroshi MURASE
Advances in intelligent vehicle systems have led to modern automobiles being able to aid drivers with tasks such as lane following and automatic braking. Such automated driving tasks increasingly require reliable ego-localization. Although there is a large number of sensors that can be employed for this purpose, the use of a single camera still remains one of the most appealing, but also one of the most challenging. GPS localization in urban environments may not be reliable enough for automated driving systems, and various combinations of range sensors and inertial navigation systems are often too complex and expensive for a consumer setup. Therefore accurate localization with a single camera is a desirable goal. In this paper we propose a method for vehicle localization using images captured from a single vehicle-mounted camera and a pre-constructed database. Image feature points are extracted, but the calculation of camera poses is not required — instead we make use of the feature points' scale. For image feature-based localization methods, matching of many features against candidate database images is time consuming, and database sizes can become large. Therefore, here we propose a method that constructs a database with pre-matched features of known good scale stability. This limits the number of unused and incorrectly matched features, and allows recording of the database scales into “tracklets”. These “Feature scale tracklets” are used for fast image match voting based on scale comparison with corresponding query image features. This process reduces the number of image-to-image matching iterations that need to be performed while improving the localization stability. We also present an analysis of the system performance using a dataset with high accuracy ground truth. We demonstrate robust vehicle positioning even in challenging lane change and real traffic situations.
Wei HAN Xiongwei ZHANG Gang MIN Xingyu ZHOU Meng SUN
In this letter, we explore joint optimization of perceptual gain function and deep neural networks (DNNs) for a single-channel speech enhancement task. A DNN architecture is proposed which incorporates the masking properties of the human auditory system to make the residual noise inaudible. This new DNN architecture directly trains a perceptual gain function which is used to estimate the magnitude spectrum of clean speech from noisy speech features. Experimental results demonstrate that the proposed speech enhancement approach can achieve significant improvements over the baselines when tested with TIMIT sentences corrupted by various types of noise, no matter whether the noise conditions are included in the training set or not.
Takahiro KITADA Hiroto OTA Xiangmeng LU Naoto KUMAGAI Toshiro ISU
Compact and room-temperature operable terahertz emitting devices have been proposed using a semiconductor coupled multilayer cavity that consists of two functional cavity layers and three distributed Bragg reflector (DBR) multilayers. Two cavity modes with an optical frequency difference in the terahertz region are realized since two cavities are coupled by the intermediate DBR multilayer. In the proposed device, one cavity is used as the active layer for two-color lasing in the near-infrared region by current injection and the other is used as the second-order nonlinear optical medium for difference-frequency generation of the two-color fundamental laser light. The control of the nonlinear polarization by face-to-face bonding of two epitaxial wafers with different orientations is quite effective to achieve bright terahertz emission from the coupled cavity. In this study, two-color emission by optical excitation was measured for the wafer-bonded GaAs/AlGaAs coupled multilayer cavity containing self-assembled InAs quantum dots (QDs). We found that optical loss at the bonding interface strongly affects the two-color emission characteristics when the bonding was performed in the middle of the intermediate DBR multilayer. The effect was almost eliminated when the bonding position was carefully chosen by considering electric field distributions of the two modes. We also fabricated the current-injection type devices using the wafer-bonded coupled multilayer cavities. An assemble of self-assembled QDs is considered to be desirable as the optical gain medium because of the discrete nature of the electronic states and the relatively wide gain spectrum due to the inhomogeneous size distribution. The gain was, however, insufficient for two-color lasing even when the nine QD layers were used. Substituting two types of InGaAs multiple quantum wells (MQWs) for the QDs, we were able to demonstrate two-color lasing of the device when the gain peaks of MQWs were tuned to the cavity modes by lowering the operating temperature.
We present a novel receiver for reliable IoT communications. In this letter, it is assumed that IoT communications are based on ZigBee under frequency-selective indoor environments. The ZigBee includes IEEE 802.15.4 specification for low-power and low-cost communications. The presented receiver fully follows the specification. However, the specification exhibits extremely low performance under frequency-selective environments. Therefore, a channel estimation approach is proposed for reliable communications under frequency-selective fading indoor environments. The estimation method relies on FFT operations, which are usually embedded in cellular phones. We also suggest a correlation method for accurate recovery of original information. The simulation results show that the proposed receiver is very suitable for IoT communications under frequency-selective indoor environments.
Xibin WANG Fengji LUO Chunyan SANG Jun ZENG Sachio HIROKAWA
With the rapid development of information and Web technologies, people are facing ‘information overload’ in their daily lives. The personalized recommendation system (PRS) is an effective tool to assist users extract meaningful information from the big data. Collaborative filtering (CF) is one of the most widely used personalized recommendation techniques to recommend the personalized products for users. However, the conventional CF technique has some limitations, such as the low accuracy of of similarity calculation, cold start problem, etc. In this paper, a PRS model based on the Support Vector Machine (SVM) is proposed. The proposed model not only considers the items' content information, but also the users' demographic and behavior information to fully capture the users' interests and preferences. An improved Particle Swarm Optimization (PSO) algorithm is also proposed to improve the performance of the model. The efficiency of the proposed method is verified by multiple benchmark datasets.
Controlling synchrony as well as desynchrony in a network of neuronal oscillators has been one of the focus issues in nonlinear science and engineering. It has been well known that spike stimuli injected commonly to multiple neurons can synchronize them if the strength of the common spike stimuli is high enough. Our recent study showed that this common spike-induced synchrony could be suppressed by introducing heterogeneity to inhibitory connections, through which the common spikes are transmitted. The aim of the present study is apply this methodology to electronic neurons as a real physical hardware. Using an Axon-Hillock circuit that represents basic properties of the leaky integrate-and-fire (LIF) neuron, our experiment demonstrated that the method was quite effective for desynchronizing the neuron circuits. The experimental results are also in a good agreement with the linear response theory that describes the input-output relationship of LIF neurons. Our method of suppressing the neuronal synchrony should be of practical use for enhancement of neural information processing as well as for improvement of pathological state of the brain.
Keisuke TOMIDA Hiroshi FUJITA Satoshi USUI Kuniaki TANAKA Hiroaki USUI
Thin films of vinyl derivatives of naphthalene diimide were prepared by electron-assisted vapor deposition. Monomer materials of N, N'-bis(allyl)-naphthalene diimide (Allyl-NDI) and N,N'-bis(p-vinyl-benzyl)-naphthalene diimide (Sty-NDI) were newly synthesized for this purpose. Uniform films were obtained by vapor-depositing these materials, whereas spin-coating yielded nonuniform films. IR analysis suggested that Sty-NDI can be polymerized upon vapor deposition. An insoluble film of Sty-NDI was obtained by the electron-assisted vapor deposition. On the other hand, Allyl-NDI had lower reactivity for polymerization. It was concluded that Sty-NDI is a promising material for preparing thin films of vinyl polymer having naphthalene diimide units.
Recently, the Static Heterogeneous Particle Swarm Optimization (SHPSO) has been studied by more and more researchers. In SHPSO, the different search behaviours assigned to particles during initialization do not change during the search process. As a consequence of this, the inappropriate population size of exploratory particles could leave the SHPSO with great difficulties of escaping local optima. This motivated our attempt to improve the performance of SHPSO by introducing the dynamic heterogeneity. The self-adaptive heterogeneity is able to alter its heterogeneous structure according to some events caused by the behaviour of the swarm. The proposed triggering events are confirmed by keeping track of the frequency of the unchanged global best position (pg) for a number of iterations. This information is then used to select a new heterogeneous structure when pg is considered stagnant. According to the different types of heterogeneity, DHPSO-d and DHPSO-p are proposed in this paper. In, particles dynamically use different rules for updating their position when the triggering events are confirmed. In DHPSO-p, a global gbest model and a pairwise connection model are automatically selected by the triggering configuration. In order to investigate the scalability of and DHPSO-p, a series of experiments with four state-of-the-art algorithms are performed on ten well-known optimization problems. The scalability analysis of and DHPSO-p reveals that the dynamic self-adaptive heterogeneous structure is able to address the exploration-exploitation trade-off problem in PSO, and provide the excellent optimal solution of a problem simultaneously.
Masanori HOSHINO Shigemasa TAKAI
We consider a decentralized similarity control problem for composite nondeterministic discrete event systems, where each subsystem has its own local specification and the entire specification is described as the synchronous composition of local specifications. We present necessary and sufficient conditions for the existence of a complete decentralized supervisor that solves a similarity control problem under the assumption that any locally uncontrollable event is not shared by other subsystems. We also show that the system controlled by the complete decentralized supervisor that consists of maximally permissive local supervisors is bisimilar to the one controlled by the maximally permissive monolithic supervisor under the same assumption.
Shinichi HASHIMOTO Takaya SHIZUME Hiroaki TAKAMATSU Yoshifumi SHIMODAIRA Gosuke OHASHI
The Helmholtz-Kohlrausch (H-K) effect is a phenomenon in which the perceived brightness levels induced by two stimuli are different even when two color stimuli have the same luminance and different chroma in a particular hue. This phenomenon appears on display devices, and the wider the gamut these devices have, the more the perceived brightness is affected by the H-K effect. The quantification of this effect can be expected to be useful for the development and evaluation of a wide range of display devices. However, quantification of the H-K effect would require considerable subjective evaluation experimentation, which would be a major burden. Therefore, the authors have derived perceived brightness maps for natural images using an estimation equation for the H-K effect without experimentation. The results of comparing and analyzing the calculated maps and ground truth maps obtained through subjective evaluation experiments confirm strong correlation coefficients between such maps overall. However, a tendency for the estimation of the calculation map to be poor on high chroma strongly influenced by the H-K effect was also confirmed. In this study, we propose an accuracy improvement method for the estimation of the H-K effect by correcting the calculation maps using a correction coefficient obtained by focusing on this tendency, and we confirm the effectiveness of our method.