University of Electro-Communications

About: University of Electro-Communications is a education organization based out in Tokyo, Japan. It is known for research contribution in the topics: Laser & Robot. The organization has 8041 authors who have published 16950 publications receiving 235832 citations. The organization is also known as: UEC & Denki-Tsūshin Daigaku.

DNN-Based Source Enhancement to Increase Objective Sound Quality Assessment Score

Abstract: We propose a training method for deep neural network (DNN) based source enhancement to increase objective sound quality assessment (OSQA) scores such as the perceptual evaluation of speech quality. In many conventional studies, DNNs have been used as a mapping function to estimate time–frequency masks and trained to minimize an analytically tractable objective function such as the mean squared error (MSE). Since OSQA scores have been used widely for sound-quality evaluation, constructing DNNs to increase OSQA scores would be better than using the minimum MSE to create high-quality output signals. However, since most OSQA scores are not analytically tractable, i.e., they are black boxes, the gradient of the objective function cannot be calculated by simply applying backpropagation. To calculate the gradient of the OSQA-based objective function, we formulated a DNN optimization scheme on the basis of black-box optimization , which is used for training a computer that plays a game. For a black-box-optimization scheme, we adopt the policy gradient method for calculating the gradient on the basis of a sampling algorithm. To simulate output signals using the sampling algorithm, DNNs are used to estimate the probability density function of the output signals that maximize OSQA scores. The OSQA scores are calculated from the simulated output signals, and the DNNs are trained to increase the probability of generating the simulated output signals that achieve high OSQA scores. Through several experiments, we found that OSQA scores significantly increased by applying the proposed method, even though the MSE was not minimized.

Linker stoichiometry-controlled stepwise supramolecular growth of a flexible Cu2Tb single molecule magnet from monomer to dimer to one-dimensional Chain

Abstract: A trinuclear complex [(CuL)2Tb(H2O)(NO3)3]·MeOH·H2O (1) is synthesized by using a metalloligand [CuL] (H2L = N,N′-bis(salicylidene)-1,3-propanediamine or salpn). This complex is used as a new 3d-4f metallatecton for stepwise supramolecular oligomerization along with a simple diimine linker, 4,4′-bipyridine (bpy). A ratio of metallatecton/linker = 2:1 is conducive for the formation of a dimer [{(CuL)2Tb(H2O)(NO3)3}2bpy]·2MeOH·4H2O (2), whereas a higher proportion of linker at a ratio of 1:1 gives rise to a zigzag one-dimensional coordination polymer {[(CuL)2Tb(NO3)3bpy]·MeOH·2H2O}∞ (3). The inherent conformational flexibility of the trinuclear (CuL)2Tb node which is folded (cisoid) for 1 and 2 (asymmetric TbO9 tricapped trigonal prism) and unfolded (transoid) for 3 (a C2-symmetric TbO10 tetradecahedron) seems to stabilize the crystal packing of these coordination-driven self-assemblies. The rapidly saturated magnetization at 1.8 K for 1 (11 μB) and 2 (22 μB) clearly indicates the presence of Tb–Cu ferromag...

Meet-in-the-Middle Technique for Integral Attacks against Feistel Ciphers

Abstract: In this paper, an improvement for integral attacks against Feistel ciphers is discussed. The new technique can reduce the complexity of the key recovery phase. This possibly leads to an extension of the number of attacked rounds. In the integral attack, an attacker guesses a part of round keys and performs the partial decryption. The correctness of the guess is judged by examining whether the XOR sum of the results becomes 0 or not. In this paper, it is shown that the computation of the XOR sum of the partial decryptions can be divided into two independent parts if the analysis target adopts the Feistel network or its variant. Then, correct key candidates are efficiently obtained with the meet-in-the-middle approach. The effect of our technique is demonstrated for several Feistel ciphers. Improvements on integral attacks against LBlock, HIGHT, and CLEFIA are presented. Particularly, the number of attacked rounds with integral analysis is extended for LBlock.

Coding 'corrective recasts' : The maintenance of meaning and more fundamental problems

Abstract: A fair amount of descriptive research in the field of second language acquisition has looked at the presence of what have been labeled corrective recasts. This research has relied on the methodological practice of coding to identify particular turns as 'corrective recasts'. Often, the coding criteria make use of the notion of the maintenance of meaning across turns, implicitly incorporating a theory of meaning. Through the use of empirically-based arguments, this paper demonstrates the problematic nature of this theory of meaning, and of the notion of the maintenance of meaning. However, the purpose of this paper is not to suggest ways that problems with such criteria may be rectified or mitigated. Rather, the exploration of problems with this notion is used to introduce more fundamental problems related to the use of coding. It is argued that the nature of the methodological practice of coding may be different from how it is generally understood. It is also argued that the use of coding obscures what is happening for participants in interaction. These arguments apply beyond research on corrective recasts within the field of second language acquisition, to all research on interaction which involves the methodological practice of coding.

A Reinforcement Learning-Based Data Storage Scheme for Vehicular Ad Hoc Networks

Abstract: Vehicular ad hoc networks (VANETs) have been attracting interest for their potential roles in intelligent transport systems (ITS). In order to enable distributed ITS, there is a need to maintain some information in the vehicular networks without the support of any infrastructure such as road side units. In this paper, we propose a protocol that can store the data in VANETs by transferring data to a new carrier (vehicle) before the current data carrier is moving out of a specified region. For the next data carrier node selection, the protocol employs fuzzy logic to evaluate instant reward by taking into account multiple metrics, specifically throughput, vehicle velocity, and bandwidth efficiency. In addition, a reinforcement learning-based algorithm is used to consider the future reward of a decision. For the data collection, the protocol uses a cluster-based forwarding approach to improve the efficiency of wireless resource utilization. We use theoretical analysis and computer simulations to evaluate the proposed protocol.