Hiroshi UEDA Masaya OHTA Akio OGIHARA Kunio FUKUNAGA
A pseudoinverse rule, one of major rule to determine a weight matrix for associative memory, has large capacity comparing with other determining rules. However, it is wellknown that the rule has small domains of attraction of memory vectors on account of many spurious states. In this paper, we try to improve the problem by means of subtracting a constant from all diagonal elements of a weight matrix. By this method, many spurious states disappear and eigenvectors with negative eigenvalues are introduced for the orthocomplement of the subspace spanned by memory vectors. This method can be applied to two types of networks: binary network and analog network. Some computer simulations are performed for both two models. The results of the simulations show our improvement is effective to extend error correcting ability for both networks.
Hiroshi UEDA Masaya OHTA Akio OGIHARA Kunio FUKUNAGA
In this article, the autocorrelation associative neural network that is one of well-known applications of neural networks is improved to extend its capacity and error correcting ability. Our approach of the improvement is based on the consideration that negative self-feedbacks remove spurious states. Therefore, we propose a method to determine the self-feedbacks as small as possible within the range that all stored patterns are stable. A state transition rule that enables to escape oscillation is also presented because the method has a possibility of falling into oscillation. The efficiency of the method is confirmed by means of some computer simulations.
Yoshikazu YAMAGUCHI Akio OGIHARA Yasuhisa HAYASHI Nobuyuki TAKASU Kunio FUKUNAGA
We propose a continuous speech recognition algorithm utilizing island-driven A* search. Conventional left-to-right A* search is probable to lose the optimal solution from a finite stack if some obscurities appear at the start of an input speech. Proposed island-driven A* search proceeds searching forward and backward from the clearest part of an input speech, and thus can avoid to lose the optimal solution from a finite stack.
Yoichiro ANZAI Koichi MATSUMOTO Shojiro YONEDA Akio OGIHARA
Recently, the hardware realizations of the neural networks for specially-purposed-use have been in focus. In this paper, two kinds of networks, a two-layer network and the Boltzmann machine, using the switched-capacitor circuit are proposed. The variable synaptic weights of neural circuit are realized by through the programmable capacitor array (PCA) in the switched-capacitor variable-coefficients multiplier. As a result, the recognition system of the handwritten character using a two-layer neural network is constructed by the discrete electronic elements and its desirable effects are shown by the experimental results. The stochastic operation in the processing element (PE) of the Boltzmann machine is realized by using the generation of noise voltage with the random number and is also confirmed by teh experimental results using the discrete electronic elements. Furthermore, the operations of the PE have been also confirmed by using the simulation of Traveling-Salesman Problem.
In the field of speech recognition, many researchers have proposed speech recognition methods using auditory information like acoustic signal or visual information like shape and motion of lips. Auditory information has valid features for speech recognition, but it is difficult to accomplish speech recognition in noisy environment. On the other side, visual information has advantage to accomplish speech recognition in noisy environment, but it is difficult to extract effective features for speech recognition. Thus, in case of using either auditory information or visual information, it is difficult to accomplish speech recognition perfectly. In this paper, we propose a method to fuse auditory information and visual information in order to realize more accurate speech recognition. The proposed method consists of two processes: (1) two probabilities for auditory information and visual information are calculated by HMM, (2) these probabilities are fused by using linear combination. We have performed speech recognition experiments of isolated words, whose auditory information (22.05kHz sampling, 8-bit quantization) and visual information (30-frame/s sampling, 24-bit quantization) are captured with multi-media personal computer, and have confirmed the validity of the proposed method.
Masaya OHTA Kazumichi MATSUMIYA Akio OGIHARA Shinobu TAKAMATSU Kunio FUKUNAGA
This article analyzes dynamics of the chaotic neural network and minimum searching principle of this network. First it is indicated that the dynamics of the chaotic newral network is described like a gradient decent, and the chaotic neural network can roughly find out a local minimum point of a quadratic function using its attractor. Secondly It is guaranteed that the vertex corresponding a local minimum point derived from the chaotic neural network has a lower value of the objective function. Then it is confirmed that the chaotic neural network can escape an invalid local minimum and find out a reasonable one.
Satoru IGAWA Akio OGIHARA Akira SHINTANI Shinobu TAKAMATSU
We propose a method to fuse auditory information and visual information for accurate speech recognition. This method fuses two kinds of information by using Iinear combination after calculating two kinds of probabilities by HMM for each word. In addition, we use full-frame color image as visual information in order to improve the accuracy of the proposed speech recognition system. We have performed experiments comparing the proposed method with the method using either auditory information or visual information, and confirmed the validity of the proposed method.
Harumi MURATA Akio OGIHARA Masaki UESAKA
Yajima et al. proposed a method based on amplitude and phase coding of audio signals. This method has relatively high sound quality because human auditory property is considered for embedding. However, in this method, the tolerance to attacks tends to be weak. Hence, we propose a high-tolerance watermarking method using BCH code which is one of error correcting code. This paper evaluates whether our method preserves the sound quality while ensuring high tolerance.
Akira SHINTANI Akio OGIHARA Yoshikazu YAMAGUCHI Yasuhisa HAYASHI Kunio FUKUNAGA
We propose two methods to fuse auditory information and visual information for accurate sppech recognition. The first method fuses two kinds of information by using linear combination after calculating two kinds of probabilities by HMM for each word. The second method fuses two kinds of information by using the histogram which expresses the correlation of them. We have performed experiments comparing the proposed methods with the conventional method and confirmed the validity of the proposed methods.
Satoshi KONDO Akio OGIHARA Shojiro YONEDA
This study proposes fuzzy matrix quantization (FMQ) which is a new coding technique developed to obtain discrete symbols employed for hidden Markov models (HMM's). FMQ is a coding technique combining fuzzy vector quantization with matrix quantization. The validity of FMQ is evaluated by a speaker-independent isolated word recognition task. First, the effect of FMQ is examined when FMQ is applied to the training phase and/or recognition phase. The effects of number of training data, codebook size and codeword matrix size for recognition accuracy are investigated. And the results of the speech recognition based on HMM recognizer using FMQ technique is compared with HMM recognizers using conventional quantization methods, vector quantization and fuzzy vector quantization. As a result, FMQ is the effective coding technique for isolated word recognition on condition that codebook size is large, above all, when FMQ is applied to the training phase and training data set is small.
Motoo YAMAMOTO Akira SHIOZAKI Motoi IWATA Akio OGIHARA
This paper presents a correlation-based watermarking method for video using the similarity of adjacent frames. In general, the adjacent frames of a video sequence is very similar. In the proposed scheme, we use an adjoining frame in detection process instead of an original image in the watermarking scheme of Cox et al. So the proposed method does not need an original video sequence in detection process. When a watermarked video sequence is attacked by overwriting copy or frame dropping, the pair of the frames that is not adjoining in an original video sequence is used in detection process. However, since a watermark is embedded in a part of each frame and embedding positions are different for each frame in the proposed method, we can detect the watermark even from an overwriting-copied video sequence and a frame-dropped video sequence. Experimental results show that the proposed method is robust against overwriting copy and frame dropping. Moreover, it is shown from experimental results that the method has robustness to low bitrate MPEG compression and StirMark attack.
Harumi MURATA Akio OGIHARA Shigetoshi HAYASHI
We have proposed an audio watermarking method based on modification of sound pressure level between channels. This method is focused on the invariability of sound localization against sound processing like MP3 and the imperceptibility about slightly change of sound localization. In this paper, we investigate about tolerance evaluation against various attacks in reference to IHC criteria.
Akira YAMAMOTO Masaya OHTA Hiroshi UEDA Akio OGIHARA Kunio FUKUNAGA
We propose an asymmetric neural network which can solve inequality-constrained combinatorial optimization problems that are difficult to solve using symmetric neural networks. In this article, a knapsack problem that is one of such the problem is solved using the proposed network. Additionally, we study condition for obtaining a valid solution. In computer simulations, we show that the condition is correct and that the proposed network produces better solutions than the simple greedy algorithm.
Akira YAMAMOTO Masaya OHTA Hiroshi UEDA Akio OGIHARA Kunio FUKUNAGA
The Traveling Salesman Problem (TSP) can be solved by a neural network using the coding scheme based on the adjacency of city in the tour. Using this coding scheme, the neural network generates a better solution than that using other coding schemes. We, however, often get the invalid solution consisting of some subtours. In this article, we propose a method of eliminating subtours using additional neurons. On the computer simulation it is shown that we get the optimum solution by means of taking only O(n2) additional neurons and trials.
Masaya OHTA Akio OGIHARA Kunio FUKUNAGA
This article deals with the binary neural network with negative self-feedback connections as a method for solving combinational optimization problems. Although the binary neural network has a high convergence speed, it hardly searches out the optimum solution, because the neuron is selected randomly at each state update. In thie article, an improvement using the negative self-feedback is proposed. First it is shown that the negative self-feedback can make some local minimums be unstable. Second a selection rule is proposed and its property is analyzed in detail. In the binary neural network with negative self-feedback, this selection rule is effective to escape a local minimum. In order to comfirm the effectiveness of this selection rule, some computer simulations are carried out for the N-Queens problem. For N=256, the network is not caught in any local minimum and provides the optimum solution within 2654 steps (about 10 minutes).
Naoshi DOI Akira SHINTANI Yasuhisa HAYASHI Akio OGIHARA Shinobu TAKAMATSU
Recently, some speech recognition methods using fusion of visual and auditory information have been researched. In this paper, a study on the mouth shape image suitable for fusion of visual and auditory information has been described. Features of mouth shape which are extracted from gray level image and binary image are adopted, and speech recognition using linear combination method has been performed. From results of speech recognition, the studies on the mouth shape features which are effective in fusion of visual and auditory information have been performed. And the effectiveness of using two kinds of mouth shape features also has been confirmed.
Hiroshi UEDA Yoichiro ANZAI Masaya OHTA Shojiro YONEDA Akio OGIHARA
In this paper, two models for associative memory based on a measure of manhattan length are proposed. First, we propose the two-layered model which has an advantage to its implementation by using PDN. We also refer to the way to improve the recalling ability of this model against noisy input patterns. Secondly, we propose the other model which always recalls the nearest memory pattern in a measure of manhattan length by lateral inhibition. Even if a noise of input pattern is so large that the first model can not recall, this model can recall correctly against such a noisy pattern. We also confirm the performance of the two models by computer simulations.
Masaya OHTA Yoichiro ANZAI Shojiro YONEDA Akio OGIHARA
This article analyzes the property of the fully interconnected neural networks as a method of solving combinatorial optimization problems in general. In particular, in order to escape local minimums in this model, we analyze theoretically the relation between the diagonal elements of the connection matrix and the stability of the networks. It is shown that the position of the global minimum point of the energy function on the hyper sphere in n dimensional space is given by the eigen vector corresponding the maximum eigen value of the connection matrix. Then it is shown that the diagonal elements of the connection matrix can be improved without loss of generality. The equilibrium points of the improved networks are classified according to their properties, and their stability is investigated. In order to show that the change of the diagonal elements improves the potential for the global minimum search, computer simulations are carried out by using the theoretical values. In according to the simulation result on 10 neurons, the success rate to get the optimum solution is 97.5%. The result shows that the improvement of the diagonal elements has potential for minimum search.
Akio OGIHARA Hitoshi UNNO Akira SHIOZAKI
We propose discrimination method of synthetic speech using pitch pattern of speech signal. By applying the proposed synthetic speech discrimination system as pre-process before the conventional HMM speaker verification system, we can improve the safety of conventional speaker verification system against imposture using synthetic speech. The proposed method distinguishes between synthetic speech and natural speech according to the pitch pattern which is distribution of value of normalized short-range autocorrelation function. We performed the experiment of user verification, and confirmed the validity of the proposed method.
Nobuyuki TAKASU Akio OGIHARA Satoshi KONDO Shojiro YONEDA
The authors propose a model of the top down parser for continuous speech recognition. It utilizes a subject of an input sentence for its top down process and a preceding transition among subjects for the determination of a new subject. A task, a washing machine operation, which has five subjects are examined.