Fast agglomerative hierarchical clustering algorithm using Locality-Sensitive Hashing

The single linkage method is a fundamental agglomerative hierarchical clustering algorithm. This algorithm regards each point as a single cluster initially. In the agglomeration step, it connects a pair of clusters such that the distance between the nearest members is the shortest. This step is repeated until only one cluster remains. The single linkage method can efficiently detect clusters in arbitrary shapes. However, a drawback of this method is a large time complexity of O(n ²), where n represents the number of data points. This time complexity makes this method infeasible for large data. This paper proposes a fast approximation algorithm for the single linkage method. Our algorithm reduces the time complexity to O(nB) by rapidly finding the near clusters to be connected by Locality-Sensitive Hashing, a fast algorithm for the approximate nearest neighbor search. Here, B represents the maximum number of points going into a single hash entry and it practically diminishes to a small constant as compared to n for sufficiently large hash tables. Experimentally, we show that (1) the proposed algorithm obtains clustering results similar to those obtained by the single linkage method and (2) it runs faster for large data than the single linkage method. Hisashi Koga received the M.S. and Ph.D. degree in information science in 1995 and 2002, respectively, from the University of Tokyo. From 1995 to 2003, he worked as a researcher at Fujitsu Laboratories Ltd. Since 2003, he has been a faculty member at the University of Electro-Communications, Tokyo (Japan). Currently, he is an associate professor at the Graduate School of Information Systems, University of Electro-Communications. His research interest includes various kinds of algorithms such as clustering algorithms, on-line algorithms, and algorithms in network communications. Tetsuo Ishibashi received the M.E. degree in information systems design from the Graduate School of Information Systems at the University of Electro-Communications in 2004. Presently, he is a system engineer at Fujitsu Broad Solution & Consulting Inc. Toshinori Watanabe received the B.E. degree in aeronautical engineering in 1971 and the D.E. degree in 1985, both from the University of Tokyo. In 1971, he worked at Hitachi as a researcher in the field of information systems design. His experience includes demand forecasting, inventory and production management, VLSI design automation, knowledge-based nonlinear optimizer, and a case-based evolutionary learning system nicknamed TAMPOPO. He also engaged in FGCS (Fifth Generation Computer System) project of Japan and developed a new hierarchical message-passing parallel cooperative VLSI layout problem solver that ran on PIM (Parallel Inference Machine) in 1991. Since 1992, he has been a professor at the Graduate School of Information Systems, University of Electro-Communications, Tokyo, Japan. His areas of interest include media analysis, learning intelligence, and the semantics of information systems. He is a member of the IEEE.     

An ESR study of the film formation steps of vapor deposition polymerization of polyamic acid

A film of polyamic acid is formed by vapor deposition polymerization of pyromellitic dianhydride (PMDA) and 4,4′-diaminodiphenyl ether (ODA). We have taken ESR spectrum during the polymerization process and compared it with the ESR spectrum of films obtained from solution polymerization. In the intermediate polymers during vapor deposition polymerization, the amide bond (? CONH? ) is coplanar with the benzene ring and two protons in the PMDA molecule. This intermediate polymer has one unpaired electron that interacts with the two nitrogen nuclei equally. On the other hand, in the polymer obtained by solution polymerization, the amide bond and the benzene ring of PMDA are not coplanar. In this polymer, too, some of the molecules have an unpaired electron that seems to have almost no coupling with NH groups. These results imply that the polymer formation via vapor deposition proceeds through different intermediates and different molecular configurations from that via the solution process.     

Sub-millimeter-long carbon nanotubes repeatedly grown on and separated from ceramic beads in a single fluidized bed reactor

A semi-continuous fluidized-bed process is reported which rapidly converts acetylene into carbon nanotubes (CNTs). Catalysts are first immobilized on ceramic beads and CNTs are then grown on the beads and then separated from them in a repetitive process accomplished within a single reactor simply by switching gases at a fixed temperature. CNTs of 6–10 nm diameter, three walls on average, 0.4 mm length and 99 wt.% purity were synthesized at an yield of over 70% in a reactor residence time shorter than 0.3 s. The easy and efficient production of such CNTs with in situ separation from the catalysts may accelerate the development of CNT-based nanotechnology industries.     

A reinforcement learning with switching controllers for a continuous action space

Reinforcement learning (RL) attracts much attention as a technique for realizing computational intelligence such as adaptive and autonomous decentralized systems. In general, however, it is not easy to put RL to practical use. This difficulty includes the problem of designing a suitable action space for an agent, i.e., satisfying two requirements in trade-off: (i) to keep the characteristics (or structure) of an original search space as much as possible in order to seek strategies that lie close to the optimal, and (ii) to reduce the search space as much as possible in order to expedite the learning process. In order to design a suitable action space adaptively, in this article, we propose a RL model with switching controllers based on Q-learning and an actor-critic to mimic the process of an infant’s motor development in which gross motor skills develop before fine motor skills. Then a method for switching controllers is constructed by introducing and referring to the “entropy.” Further, through computational experiments by using a path-planning problem with continuous action space, the validity and potential of the proposed method have been confirmed.     

新一代超小型DIP-IPM

为了满足空调、洗衣机、冰箱等白色家电对高效、高性能的需要,三菱电机开发了新一代超小型第四代(Ver.4)双列直插封装的智能功率模块(Dual-In-linePackageIntelligentPowerModule,简称DIP-IPM)。本文首先分析了第四代超小型DIP-IPM内部电路的构成和功能,然后详细介绍了第四代DIP-IPM生产过程中的关键技术,包括无铅化、高热传导性绝缘片、chip-to-chip导线直接连接以及ASIC技术,最后与第三代(Ver.3)DIP-IPM进行了比较。     

An FPGA implementation of hard‐wired sequence control system based on PLC software

Although a programmable logic controller (PLC) has been widely adopted for the sequence control of industrial machinery, its performance does not always satisfy the recent requirements in large and highly responsive systems. With the state‐of‐the‐art field programmable gate array (FPGA) technology, it is possible to implement a control program with hard‐wired logic for higher response and reduced implementation cost/space. This approach is also worthwhile for transmigration of legacy PLC software into forthcoming FPGA‐based control hardware. This study presents a systematic method to implement a hard‐wired sequence control from PLC software. PLC instructions are converted into VHDL codes, and then implemented as logic circuit with various peripheral functions. Productive PLC programs were examined with Mitsubishi Electric FX2N PLC and Altera Stratix II FPGA, and were shown to fit into a common FPGA chip. A straightforward Sequential design was estimated to be 184 times faster than PLC, while a performance‐oriented Flat design was estimated to be 44 times faster than Sequential design (i.e., 8050 times faster than PLC). A practical perfect layer winder system was actually built and successfully operated with our FPGA control board, whose logic design was implemented with our tools. © 2011 Institute of Electrical Engineers of Japan. Published by John Wiley & Sons, Inc.     

Application of step‐response test for evaluation of uncertainty of standard measuring system for lightning‐impulse high voltage

In evaluating the uncertainty of the standard measuring system for lightning‐impulse high voltages, which is composed of a standard voltage divider, a digital recorder, and calibrators, step‐response tests of the standard voltage divider may be useful. In this paper, a convolution algorithm is employed to calculate the output impulse voltage waveforms from measured step‐response waveforms. The uncertainties of peak‐value measurement due to the influence of the nominal epoch, uncertainty of the peak‐value measurement due to dispersion of the AC scale factor, and uncertainty of the virtual front‐time measurement due to long‐term stability are evaluated. Furthermore, the error of the virtual front time of the output waveforms is discussed. The front part of the step‐response waveform, t≤ T₃₀%, does not influence the error of the virtual front time. Therefore, for the standard voltage divider, the step‐response parameters, that is, the experimental response time, partial response time, settling time, and overshoot, have almost nothing to do with the error of the virtual front time. © 2012 Wiley Periodicals, Inc. Electr Eng Jpn, 180(2): 24–32, 2012; Published online in Wiley Online Library ( wileyonlinelibrary.com ). DOI 10.1002/eej.21279     

Links between Immune Cells from the Periphery and the Brain in the Pathogenesis of Epilepsy: A Narrative Review

Accumulating evidence has demonstrated that the pathogenesis of epilepsy is linked to neuroinflammation and cerebrovascular dysfunction. Peripheral immune cell invasion into the brain, along with these responses, is implicitly involved in epilepsy. This review explored the current literature on the association between the peripheral and central nervous systems in the pathogenesis of epilepsy, and highlights novel research directions for therapeutic interventions targeting these reactions. Previous experimental and human studies have demonstrated the activation of the innate and adaptive immune responses in the brain. The time required for monocytes (responsible for innate immunity) and T cells (involved in acquired immunity) to invade the central nervous system after a seizure varies. Moreover, the time between the leakage associated with blood–brain barrier (BBB) failure and the infiltration of these cells varies. This suggests that cell infiltration is not merely a secondary disruptive event associated with BBB failure, but also a non-disruptive event facilitated by various mediators produced by the neurovascular unit consisting of neurons, perivascular astrocytes, microglia, pericytes, and endothelial cells. Moreover, genetic manipulation has enabled the differentiation between peripheral monocytes and resident microglia, which was previously considered difficult. Thus, the evidence suggests that peripheral monocytes may contribute to the pathogenesis of seizures.     

Sieving of aerosol particles with metal screens

Metal screens with uniform micrometer-sized opening were employed to sieve aerosol particles by suppressing the adhesion of particles smaller than the openings. The collection efficiencies of monodispersed polystyrene latex (PSL) particles were experimentally determined using the metal screens with 1.2, 1.8, 2.5, and 4.2 μm openings at various filtration velocities. The particles smaller than the mesh opening adhered on the metal screen at a low filtration velocity, but the bounce-off of particles on the mesh surface suppressed the adhesion at a high velocity. As a result, we found that the adhesion of PSL particles larger than 0.3 μm mostly suppressed at a filtration velocity higher than 10 m s^?1 and therefore we can sieve aerosol particles according to the opening size of metal screens. We also found that the particle number concentration could be determined by measuring the increase in pressure drop since the clogging of metal screen openings takes place by the individual particles.

© 2016 American Association for Aerosol Research     

Preserving worker privacy in crowdsourcing

This paper proposes a crowdsourcing quality control method with worker-privacy preservation. Crowdsourcing allows us to outsource tasks to a number of workers. The results of tasks obtained in crowdsourcing are often low-quality due to the difference in the degree of skill. Therefore, we need quality control methods to estimate reliable results from low-quality results. In this paper, we point out privacy problems of workers in crowdsourcing. Personal information of workers can be inferred from the results provided by each worker. To formulate and to address the privacy problems, we define a worker-private quality control problem, a variation of the quality control problem that preserves privacy of workers. We propose a worker-private latent class protocol where a requester can estimate the true results with worker privacy preserved. The key ideas are decentralization of computation and introduction of secure computation. We theoretically guarantee the security of the proposed protocol and experimentally examine the computational efficiency and accuracy.