著者

福島 邦彦

出版者

The Institute of Electronics, Information and Communication Engineers

雑誌

電子情報通信学会論文誌 A (ISSN:03736091)

巻号頁・発行日

vol.J62-A, no.10, pp.658-665, 1979-10-25

---

パターン認識における最大の難題は,入力パターンの位置がずれたり形がゆがんだりしたときに,どのような処理を施すべきかという問題であり,この問題に対する根本的な解決法はこれまでみいだされていなかった.筆者は,動物の視覚神経系の構造からヒントを得て,この問題を解決する新しいアルゴリズムを考察し,これを実現する多層の神経回路モデル(ネオコグニトロンと呼ぶ)を構成し,計算機シミュレーションによってその能力を確認したので報告する.この神経回路モデルは,自己組織化能力をも有する.認識すべき複数個のパターンを回路に繰返し呈示しているだけで,回路は,それらのパターンを区別して正しく認識する能力を,教師なし学習によって身につけていく.自己組織化が完了した状態では,回路は,入力パターンの呈示位置がずれても,その大きさや形が多少変形しても,多少の雑音が含まれていても,正しくパターンを認識する.

著者

Mariana RODRIGUES MAKIUCHI Tifani WARNITA Nakamasa INOUE Koichi SHINODA Michitaka YOSHIMURA Momoko KITAZAWA Kei FUNAKI Yoko EGUCHI Taishiro KISHIMOTO

出版者

The Institute of Electronics, Information and Communication Engineers

雑誌

IEICE Transactions on Information and Systems (ISSN:09168532)

巻号頁・発行日

vol.E104.D, no.11, pp.1930-1940, 2021-11-01 (Released:2021-11-01)

参考文献数

71

被引用文献数

6

---

We propose a non-invasive and cost-effective method to automatically detect dementia by utilizing solely speech audio data. We extract paralinguistic features for a short speech segment and use Gated Convolutional Neural Networks (GCNN) to classify it into dementia or healthy. We evaluate our method on the Pitt Corpus and on our own dataset, the PROMPT Database. Our method yields the accuracy of 73.1% on the Pitt Corpus using an average of 114 seconds of speech data. In the PROMPT Database, our method yields the accuracy of 74.7% using 4 seconds of speech data and it improves to 80.8% when we use all the patient's speech data. Furthermore, we evaluate our method on a three-class classification problem in which we included the Mild Cognitive Impairment (MCI) class and achieved the accuracy of 60.6% with 40 seconds of speech data.

著者

Daiki CHIBA Ayako AKIYAMA HASEGAWA Takashi KOIDE Yuta SAWABE Shigeki GOTO Mitsuaki AKIYAMA

出版者

The Institute of Electronics, Information and Communication Engineers

雑誌

IEICE Transactions on Information and Systems (ISSN:09168532)

巻号頁・発行日

vol.E103.D, no.7, pp.1493-1511, 2020-07-01 (Released:2020-07-01)

参考文献数

70

被引用文献数

3

---

Internationalized domain names (IDNs) are abused to create domain names that are visually similar to those of legitimate/popular brands. In this work, we systematize such domain names, which we call deceptive IDNs, and analyze the risks associated with them. In particular, we propose a new system called DomainScouter to detect various deceptive IDNs and calculate a deceptive IDN score, a new metric indicating the number of users that are likely to be misled by a deceptive IDN. We perform a comprehensive measurement study on the identified deceptive IDNs using over 4.4 million registered IDNs under 570 top-level domains (TLDs). The measurement results demonstrate that there are many previously unexplored deceptive IDNs targeting non-English brands or combining other domain squatting methods. Furthermore, we conduct online surveys to examine and highlight vulnerabilities in user perceptions when encountering such IDNs. Finally, we discuss the practical countermeasures that stakeholders can take against deceptive IDNs.

著者

Graham NEUBIG Masato MIMURA Shinsuke MORI Tatsuya KAWAHARA

出版者

The Institute of Electronics, Information and Communication Engineers

雑誌

IEICE Transactions on Information and Systems (ISSN:09168532)

巻号頁・発行日

vol.E95.D, no.2, pp.614-625, 2012-02-01 (Released:2012-02-01)

参考文献数

40

被引用文献数

11 24 6

---

We propose a novel scheme to learn a language model (LM) for automatic speech recognition (ASR) directly from continuous speech. In the proposed method, we first generate phoneme lattices using an acoustic model with no linguistic constraints, then perform training over these phoneme lattices, simultaneously learning both lexical units and an LM. As a statistical framework for this learning problem, we use non-parametric Bayesian statistics, which make it possible to balance the learned model's complexity (such as the size of the learned vocabulary) and expressive power, and provide a principled learning algorithm through the use of Gibbs sampling. Implementation is performed using weighted finite state transducers (WFSTs), which allow for the simple handling of lattice input. Experimental results on natural, adult-directed speech demonstrate that LMs built using only continuous speech are able to significantly reduce ASR phoneme error rates. The proposed technique of joint Bayesian learning of lexical units and an LM over lattices is shown to significantly contribute to this improvement.

著者

Takaharu KATO Ikuko SHIMIZU Tomas PAJDLA

出版者

The Institute of Electronics, Information and Communication Engineers

雑誌

IEICE Transactions on Information and Systems (ISSN:09168532)

巻号頁・発行日

vol.E105.D, no.9, pp.1590-1599, 2022-09-01 (Released:2022-09-01)

参考文献数

39

---

Selecting visually overlapping image pairs without any prior information is an essential task of large-scale structure from motion (SfM) pipelines. To address this problem, many state-of-the-art image retrieval systems adopt the idea of bag of visual words (BoVW) for computing image-pair similarity. In this paper, we present a method for improving the image pair selection using BoVW. Our method combines a conventional vector-based approach and a set-based approach. For the set similarity, we introduce a modified version of the Simpson (m-Simpson) coefficient. We show the advantage of this measure over three typical set similarity measures and demonstrate that the combination of vector similarity and the m-Simpson coefficient effectively reduces false positives and increases accuracy. To discuss the choice of vocabulary construction, we prepared both a sampled vocabulary on an evaluation dataset and a basic pre-trained vocabulary on a training dataset. In addition, we tested our method on vocabularies of different sizes. Our experimental results show that the proposed method dramatically improves precision scores especially on the sampled vocabulary and performs better than the state-of-the-art methods that use pre-trained vocabularies. We further introduce a method to determine the k value of top-k relevant searches for each image and show that it obtains higher precision at the same recall.

著者

Masayoshi Yamamoto Shinya Shirai Senanayake Thilak Jun Imaoka Ryosuke Ishido Yuta Okawauchi Ken Nakahara

出版者

The Institute of Electronics, Information and Communication Engineers

雑誌

IEICE Transactions on Fundamentals of Electronics, Communications and Computer Sciences (ISSN:09168508)

巻号頁・発行日

pp.2021GCI0001, (Released:2021-11-26)

---

In response to fast charging systems, Silicon Carbide (SiC) power semiconductor devices are of great interest of the automotive power electronics applications as the next generation of fast charging systems require high voltage batteries. For high voltage battery EVs (Electric Vehicles) over 800V, SiC power semiconductor devices are suitable for 3-phase inverters, battery chargers, and isolated DC-DC converters due to their high voltage rating and high efficiency performance. However, SiC-MOSFETs have two characteristics that interfere with high-speed switching and high efficiency performance operations for SiC MOS-FET applications in automotive power electronics systems. One characteristic is the low voltage rating of the gate-source terminal, and the other is the large internal gate-resistance of SiC MOS-FET. The purpose of this work was to evaluate a proposed hybrid gate drive circuit that could ignore the internal gate-resistance and maintain the gate-source terminal stability of the SiC-MOSFET applications. It has been found that the proposed hybrid gate drive circuit can achieve faster and lower loss switching performance than conventional gate drive circuits by using the current source gate drive characteristics. In addition, the proposed gate drive circuit can use the voltage source gate drive characteristics to protect the gate-source terminals despite the low voltage rating of the SiC MOS-FET gate-source terminals.

著者

Yuya KAMATAKI Yusuke KAMEDA Yasuyo KITA Ichiro MATSUDA Susumu ITOH

出版者

The Institute of Electronics, Information and Communication Engineers

雑誌

IEICE Transactions on Information and Systems (ISSN:09168532)

巻号頁・発行日

vol.E104.D, no.10, pp.1572-1575, 2021-10-01 (Released:2021-10-01)

参考文献数

11

被引用文献数

1

---

This paper proposes a lossless coding method for HDR color images stored in a floating point format called Radiance RGBE. In this method, three mantissa and a common exponent parts, each of which is represented in 8-bit depth, are encoded using the block-adaptive prediction technique with some modifications considering the data structure.

著者

Daisuke OKU Kotaro TERADA Masato HAYASHI Masanao YAMAOKA Shu TANAKA Nozomu TOGAWA

出版者

The Institute of Electronics, Information and Communication Engineers

雑誌

IEICE TRANSACTIONS on Information and Systems (ISSN:09168532)

巻号頁・発行日

vol.E102-D, no.9, pp.1696-1706, 2019-09-01

被引用文献数

23

---

Combinatorial optimization problems with a large solution space are difficult to solve just using von Neumann computers. Ising machines or annealing machines have been developed to tackle these problems as a promising Non-von Neumann computer. In order to use these annealing machines, every combinatorial optimization problem is mapped onto the physical Ising model, which consists of spins, interactions between them, and their external magnetic fields. Then the annealing machines operate so as to search the ground state of the physical Ising model, which corresponds to the optimal solution of the original combinatorial optimization problem. A combinatorial optimization problem can be firstly described by an ideal fully-connected Ising model but it is very hard to embed it onto the physical Ising model topology of a particular annealing machine, which causes one of the largest issues in annealing machines. In this paper, we propose a fully-connected Ising model embedding method targeting for CMOS annealing machine. The key idea is that the proposed method replicates every logical spin in a fully-connected Ising model and embeds each logical spin onto the physical spins with the same chain length. Experimental results through an actual combinatorial problem show that the proposed method obtains spin embeddings superior to the conventional de facto standard method, in terms of the embedding time and the probability of obtaining a feasible solution.

著者

森田 正典

出版者

The Institute of Electronics, Information and Communication Engineers

雑誌

電子情報通信学会論文誌 D (ISSN:09135713)

巻号頁・発行日

vol.J70-D, no.11, pp.2047-2057, 1987-11-25

---

まず,入力の対象となる日本文の特徴,入力方式に関係する人間工学的要素,および入力方式として望ましい条件の3者を明確にした.上記3要件を踏まえて,最適の日本文入力方式は何であるかを追求した結果,日本文入力用に最適化したローマ字方式である通称M方式が最も優れているとの結論を導いた.M方式の特徴は,子音キーと母音キーを,右手と左手に分類してそれぞれを50音順に配置し,漢字入力の際の打鍵数節減のための特別な複合キーを配置して,漢字入力の高速化を図ったことである.一方鍵盤方式としては,現在一般に使用されているキーボードの欠点を明確にし,筆者らが,それらの欠点を改善のために努力して改善を重ねてきた各種の製品を紹介し,最新型の鍵盤方式としては,仮想キー方式の採用によって機能キーの数を減少させ,常時頻繁に使用する機能キーのみを左右の手の形に合わせたデータキーの周辺に配置した,左右分離型の鍵盤を紹介している.

著者

Yoshinao ISOBE Hisabumi HATSUGAI Akira TANAKA Yutaka OIWA Takanori AMBE Akimasa OKADA Satoru KITAMURA Yamato FUKUTA Takashi KUNIFUJI

出版者

The Institute of Electronics, Information and Communication Engineers

雑誌

IEICE TRANSACTIONS on Fundamentals of Electronics, Communications and Computer Sciences (ISSN:09168508)

巻号頁・発行日

vol.E102-A, no.2, pp.325-335, 2019-02-01

---

This paper presents a formal approach for generating train timetables in a mesoscopic level that is more concrete than the macroscopic level, where each station is simply expressed in a black-box, and more abstract than the microscopic level, where the infrastructure in each station-area is expressed in detail. The accuracy of generated timetable and the computational effort for the generation is a trade-off. In this paper, we design a formal mesoscopic modeling language by analyzing real railways, for example Tazawako-line as the first step of this work. Then, we define the constraint formulae for generating train timetables with the help of SMT (Satisfiability Module Theories)-Solver, and explain our tool RW-Solver that is an implementation of the constraint formulae. Finally, we demonstrate how RW-Solver with the help of SMT-Solver can be used for generating timetables in a case study of Tazawako-line.

著者

Kei SAWADA Akira TAMAMORI Kei HASHIMOTO Yoshihiko NANKAKU Keiichi TOKUDA

出版者

The Institute of Electronics, Information and Communication Engineers

雑誌

IEICE TRANSACTIONS on Information and Systems (ISSN:09168532)

巻号頁・発行日

vol.E99-D, no.12, pp.3119-3131, 2016-12-01

---

This paper proposes a Bayesian approach to image recognition based on separable lattice hidden Markov models (SL-HMMs). The geometric variations of the object to be recognized, e.g., size, location, and rotation, are an essential problem in image recognition. SL-HMMs, which have been proposed to reduce the effect of geometric variations, can perform elastic matching both horizontally and vertically. This makes it possible to model not only invariances to the size and location of the object but also nonlinear warping in both dimensions. The maximum likelihood (ML) method has been used in training SL-HMMs. However, in some image recognition tasks, it is difficult to acquire sufficient training data, and the ML method suffers from the over-fitting problem when there is insufficient training data. This study aims to accurately estimate SL-HMMs using the maximum a posteriori (MAP) and variational Bayesian (VB) methods. The MAP and VB methods can utilize prior distributions representing useful prior information, and the VB method is expected to obtain high generalization ability by marginalization of model parameters. Furthermore, to overcome the local maximum problem in the MAP and VB methods, the deterministic annealing expectation maximization algorithm is applied for training SL-HMMs. Face recognition experiments performed on the XM2VTS database indicated that the proposed method offers significantly improved image recognition performance. Additionally, comparative experiment results showed that the proposed method was more robust to geometric variations than convolutional neural networks.

著者

Tomoki Yamada Yu Yonezawa Masayoshi Yamamoto

出版者

The Institute of Electronics, Information and Communication Engineers

雑誌

IEICE Electronics Express (ISSN:13492543)

巻号頁・発行日

vol.20, no.21, pp.20230351, 2023-11-10 (Released:2023-11-10)

参考文献数

30

---

This letter proposes a simpler and more efficient circuit scheme than the conventional current fed full bridge converter. In the proposed circuit, an auxiliary resonant capacitor is added to the conventional circuit to eliminate the adverse effects of leakage inductance of the transformer on the circuit operation, and at the same time, soft switching is achieved. As a result, switching losses were reduced and the efficiency was improved by approximately 4.7% compared to the conventional method, which was confirmed on LTspice. First, the basic structure of the proposed circuit is shown, followed by the principle of operation. Next, the advantages and disadvantages of the proposed method are presented. Finally, we compare the proposed method with the conventional method.

著者

橋本 新一郎

出版者

The Institute of Electronics, Information and Communication Engineers

雑誌

電子情報通信学会論文誌 D (ISSN:09135713)

巻号頁・発行日

vol.J56-D, no.11, pp.654-661, 1973-11-25

---

日本語単語アクセントの言語学的,音響学的,聴覚的諸性質について,統一的に論じた.まず,単語アクセントの種類が,日本語東京方言では,せいぜい十数種であり,0形から5形までで,全単語の98%以上をしめること,また3形以上で,nモーラ長の単語を取り上げた場合,8形の例外を除けば,第(n-2)モーラにアクセント核が存在する確率が最も高いことを見い出した.つぎに,単語の各モーラについて,母音のエネルギー重心点で求めた基本周波数は,単語の種類によらず(同一アクセント形をもつ単語について),きわめて安定であり,単語アクセントの形を反映する音響パラメータとなることを示した.また,この基本周波数と振幅および音韻継続時間の三要素がアクセント感形成に,上記の順序で寄与していることを明らかにした.最後に,合成音声を用いて,種々なピッチパターンのアクセント感に及ぼす影響を調べた結果,単語の種類や被験者によらない,各アクセントの形に固有なピッチパターンが存在し,その聴覚的許容範囲は,一人の話者の発音によるピッチパターンのばらつきよりも一般に広いことが明らかとなった.

著者

興梠 紗和 木村 昭悟 藤代 裕之 西川 仁

出版者

The Institute of Electronics, Information and Communication Engineers

雑誌

電子情報通信学会論文誌 D (ISSN:18804535)

巻号頁・発行日

vol.J99-D, no.4, pp.403-414, 2016-04-01

---

SNSの隆盛によりニュースを取り巻く環境は大きく変化している.新聞やテレビから一方的に配信される記事を受け取るのではなく,膨大な情報で溢れるSNS上から関心のある記事を選択して購読する新たなニュースの読まれ方が生まれている.この変化により,ニュースメディアはSNS上で記事を読者に対して効果的にアピールする必要に迫られている.その一方で,刺激的な言葉を用いてむやみに拡散させるのではなく,記事を正確に説明し,その内容に興味をもつ読者に記事を届ける必要がある.本研究では,ニュース配信者がニュース消費者に適切なニュース記事を提供するための一手段として,ニュース記事を的確に説明する説明文が,SNS上でより多くの読者に読まれるために備えるべき性質を特定することを目指す.この目標に向け,本論文ではまず記者と編集者を対象としたヒアリング調査と,ニュースサイトがSNSに投稿している説明文の調査を行った.これらの調査を分析することで明らかになった,説明文がもつべき性質を利用することで,与えられたニュース記事をSNS上で紹介する説明文を幾つかの候補の中から自動的に選択する手法を提案する.

著者

Ken Umeno

出版者

The Institute of Electronics, Information and Communication Engineers

雑誌

Nonlinear Theory and Its Applications, IEICE (ISSN:21854106)

巻号頁・発行日

vol.7, no.1, pp.14-20, 2016 (Released:2016-01-01)

参考文献数

11

被引用文献数

2 13

---

We consider a family of ergodic transformations on the real line R preserving Cauchy laws. A dualistic nature between the ergodic transformation and the associated transformation of the scale parameter of a Cauchy law is proven to be hold, which provides a systematic view of explicit mixing property with the ergodic transformation having the Cauchy law as the limiting distribution.

著者

佐藤 宏介 井口 征士

出版者

The Institute of Electronics, Information and Communication Engineers

雑誌

電子情報通信学会論文誌 D (ISSN:09135713)

巻号頁・発行日

vol.J68-D, no.3, pp.369-375, 1985-03-25

---

距離画像(Range-Picture)は,画像中の各画素が物体面までの距離情報を担ったものであり,三次元物体の認識に有効である.本論文では距離画像入力の新しい計測法について述べる.本方法ではCCDカメラとパターン光投影器を用いて,能動ステレオ法により距離情報を得る.パターン光投影器は2進符号化された縦縞状の2値パターン光を測定空間に投光して,空間を細いクサビ状に分割する.各々の領域は1本1本のスリット光に見なすことができ,割り当てられたコードにより識別が可能である.n回のパターン光投影で2n本のスリット光投影と等価な距離画像が得られるため,高速な計測が期待できる.空間のコード化には交番2進符号(グレイコード)を用いて,パターン光の明暗部境界でのコード化誤りを最小にする.最後に,実際の計測例により,テクスチャを含む物体にも有効であることを示し,また多面体の観測も行う.

著者

Kazuhiro NAKAMURA Kei HASHIMOTO Yoshihiko NANKAKU Keiichi TOKUDA

出版者

The Institute of Electronics, Information and Communication Engineers

雑誌

IEICE TRANSACTIONS on Information and Systems (ISSN:09168532)

巻号頁・発行日

vol.E97-D, no.6, pp.1438-1448, 2014-06-01

---

This paper proposes a novel approach for integrating spectral feature extraction and acoustic modeling in hidden Markov model (HMM) based speech synthesis. The statistical modeling process of speech waveforms is typically divided into two component modules: the frame-by-frame feature extraction module and the acoustic modeling module. In the feature extraction module, the statistical mel-cepstral analysis technique has been used and the objective function is the likelihood of mel-cepstral coefficients for given speech waveforms. In the acoustic modeling module, the objective function is the likelihood of model parameters for given mel-cepstral coefficients. It is important to improve the performance of each component module for achieving higher quality synthesized speech. However, the final objective of speech synthesis systems is to generate natural speech waveforms from given texts, and the improvement of each component module does not always lead to the improvement of the quality of synthesized speech. Therefore, ideally all objective functions should be optimized based on an integrated criterion which well represents subjective speech quality of human perception. In this paper, we propose an approach to model speech waveforms directly and optimize the final objective function. Experimental results show that the proposed method outperformed the conventional methods in objective and subjective measures.

著者

中村 和寛 大浦 圭一郎 南角 吉彦 徳田 恵一

出版者

The Institute of Electronics, Information and Communication Engineers

雑誌

電子情報通信学会論文誌 D (ISSN:18804535)

巻号頁・発行日

vol.J97-D, no.10, pp.1572-1581, 2014-10-01

---

本論文では隠れマルコフモデル(Hidden Markov Model; HMM) に基づく英語歌声合成について述べる.HMM歌声合成システムは,学習用の歌声データに基づいて,あらかじめスペクトル,基本周波数,ビブラートをHMMにより同時にモデル化しておき,合成時には合成したい歌声の楽譜に合わせてHMMを連結し,歌声を生成する.これまでに,日本語の楽譜から歌声を合成するシステムが提案され,一般ユーザによる楽曲作成の際のボーカルとして利用されてきている.本論文ではこのシステムを,英語の歌声を合成できるように拡張するために,英語歌声合成のコンテクストを定義し,楽譜の音符と実際の発音を対応付ける手法を提案する.客観・主観評価実験により効果を確認し,また,日本語歌声合成との比較実験も行う.

著者

Manabu HAGIWARA

出版者

The Institute of Electronics, Information and Communication Engineers

雑誌

IEICE Transactions on Fundamentals of Electronics, Communications and Computer Sciences (ISSN:09168508)

巻号頁・発行日

pp.2022TAP0008, (Released:2022-09-21)

---

This paper considers error-correction for information in array design, i.e., two-dimensional design such as QR-codes. The error model is multi deletion/substitution/erasure errors. Code construction for the errors and an application of the code are provided. The decoding technique uses an error-locator for deletion codes.

著者

Ayumu Nakayama Manabu Hagiwara

出版者

The Institute of Electronics, Information and Communication Engineers

雑誌

IEICE Communications Express (ISSN:21870136)

巻号頁・発行日

pp.2019XBL0154, (Released:2020-01-22)

被引用文献数

13

---

A quantum error-correcting code for single deletion errors is provided. To the authors' best knowledge, this is the first code for deletion errors.