Detection of facial emotions are mainly addressed by computer vision researchers based on facial display. Also detection of vocal expressions of emotions is found in research work done by acoustic researchers. Most of these research paradigms are devoted purely to visual or purely to auditory human emotion detection. However we found that it is very interesting to consider both of these auditory and visual informations together, for processing, since we hope this kind of multimodal information processing will become a datum of information processing in future multimedia era. By several intensive subjective evaluation studies we found that human beings recognize Anger, happiness, Surprise and Dislike by their visual appearance, compared to voice only detection. When the audio track of each emotion clip is dubbed with a different type of auditory emotional expression, still Anger, Happiness and Surprise were video dominant. However Dislike emotion gave mixed responses to different speakers. In both studies we found that Sadness and Fear emotions were audio dominant. As a conclusion to the paper we propose a method of facial emotion detection by using a hybrid approach, which uses multimodal informations for facial emotion recognition.

In this paper, a dynamic constructive algorithm for fault tolerant feedforward neural network, called DCFTA, is proposed. The algorithm starts with a network with single hidden neuron, and a new hidden unit is added dynamically to the network whenever it fails to converge. Before inserting the new hidden neuron into the network, only the weights connecting the new hidden neuron to the other neurons are trained (i. e. , updated) until there is no significant reduction of the output error. To generate a fault tolerant network, the relevance of each synaptic weight is estimated in each cycle, and only the weights which have their relevance less than a specified threshold are updated in that cycle. The loss of a connections between neurons (which are equivalent to stuck-at-0 faults) are assumed. The simulation results indicate that the network constructed by DCFTA has a significant fault tolerance ability.

We present a new scheduling policy named Value-based Processor Allocation (VPA-k) for scheduling value-based transactions in a multiprocessor real-time database system. The value of a transaction represents the profit the transaction contributes to the system if it is completed before its deadline. Using VPA-k policy, the transactions with higher values are given higher priorities to execute first, while at most k percentage of the total processors are allocated to the urgent transactions dynamically. Through simulation experiments, VPA-k policy is shown to outperform other scheduling policies substantially in both maximizing the totally obtained values and minimizing the number of missed transactions.

The address-based queues are widely used in shared buffer ATM switches to guarantee the order of the cell delivery. In this paper, we propose an address-based queue mechanism to achieve an efficient use of the shared memory under a multicast service. In the switch, both cells and the address queues share the common memory. Each queue length changes flexibly according to the number of the stored cells. Our approach significantly reduces the cell loss probability as compared with the previously proposed approaches.

An oversampling theorem for regular sampling in wavelet subspaces is established. The sufficient-necessary condition for which it holds is found. Meanwhile the truncation error and aliasing error are estimated respectively when the theorem is applied to reconstruct discretely sampled signals. Finally an algorithm is formulated and an example is calculated to show the algorithm.

We propose a new method for verifying synchronous circuits using the Boyer-Moore Theorem Prover (BMTP) based on an efficient use of induction. The method contains two techniques. The one is the representation method of signals. Each signal is represented not as a waveform, but as a time parameterized function. The other is the mechanical transformation of the circuit description. A simple description of the logical connection of the components of a circuit is transformed into such a form that is not only acceptable as a definition of BMTP but also adequate for applying induction. We formalize the method and show that it realizes an efficient proof.

This letter proposes a Perfect-Reconstruction (PR) encryption scheme based on a PR QMF bank. Using the proposed scheme, signals can be encrypted and reconstructed perfectly by using two Periodically Time-Varying (PTV) digital filters respectively. Also we find that the proposed scheme has a "good" encryption effect and compares favorably with frequency scramble in the aspects of computation complexity, PR property, and degree of security.

We propose new key exchange and authentication protocols, which are efficient in protecting a poorly-chosen weak secret from guessing attacks, based on the use of a one-time pad and a strong one-way hash function. Cryptographic protocols assume that a strong secret should be shared between communication participants for authentication, in the light of an ever-present threat of guessing attacks. Cryptographically long secret would be better for security only if ordinary users could remember it. But most users choose an easy-to-remember password as a secret and such a weak secret can be guessed easily. In our previous work, we made much of introducing a basic concept and its application. In this paper, we describe our idea in more detail and propose more protocols which correspond to variants of our basic protocol using well-defined notations. Formal verification and efficiency comparison of the proposed protocols are also presented. By our scheme the password guessing attacks are defeated efficiently, and a session key is exchanged and participants are authenticated securely.

This paper proposes to apply random mapping methods of a pseudo random function to find collisions of a hash function. We test a hash function including a block cipher (see ISO/IEC 10118-2) with computers, where users can select its initial vector. In particular, the paper shows that a hash function with multiple stages generates a lot of collision hash values, so our probabilistic consideration of a small model for the hash function well explains the computational results. We show that it's feasible to find collisions between the selected messages in advance for 64-bit-size hash functions with WSs linked via an ordinary LAN (Local Area Network). Thus, it is dangerous to use the hash function -- single block mode -- defined in [6] and [7].

This paper proposes a secure electronic sealed-bid auction protocol (SEAP) that provides an auction service on the Internet by combining three providers: an auction service provider, a key service provider, and a time service provider. The SEAP uses public key cryptography and the concept of a time-key certificate. The most important property of this protocol is that time-dependent security requirements can be strictly satisfied. The SEAP satisfies the following nine security requirements: (a) no one can deny having made a bid; (b) the protocol should be secure against malicious acts; (c) no bidder can act for another bidder; (d) no one can know who else is bidding until the time comes for the bids to be opened; (e) no one can discover the contents of any of the bids until the time comes for the bids to be opened; (f) the successful bid must have been submitted before the bidding deadline; (g) all bidders can verify that the auction policy has been correctly implemented; (h) the successful bidder can be identified without being required to make himself or herself known; and (i) the bidding contents cannot be altered. The protocol consists of three subprotocols: the Registration Subprotocol, the Bidding Subprotocol, and the Auction Subprotocol. The protocol parameters and algorithm are described in detail.

In Asiacrypt '96, Bleichenbacher et al. showed the upper limit of the efficiency of one-time digital signature scheme using a directed graph of tree structure as its base. They also claimed that there exists more effective signature scheme on general directed graphs, and showed an example of a method to construct more effective signature schemes as a witness. Unfortunately, their example does not achieve the efficiency as they claimed. This paper shows the upper limit of the efficiency of the signature scheme on general directed graphs by showing no signature scheme is more effective than the optimal signature scheme on trees (or forests). Further, we introduce another signature scheme named pseudo k-time signature scheme. This signature scheme allows signers to sign k-time which is no less efficient than the one time signature scheme.

For given integerp, a flow networkNwithn vertices, and sources inN, a problem of finding location ofp sinks inN which maximize the value of maximum flow from sources to sinks is calledp-collection problem. This problem is NP-hard even if a given network is a tree, but a pseudo-polynomial time algorithm is possible for such a network. This paper proposes a new pseudo-polynomial time algorithm for a tree-type network, which is simpler and more efficient than the previous algorithm, and has time complexity of O(p2n2Cc min {Cc,Cd}), whereCc andCd are the maximum edge capacity and the maximum vertex weight, respectively.

This paper applies linear cryptanalysis to FEAL and describes the experimental results of attacking FEAL-8 by linear cryptanalysis. The following points are important in linear cryptanalysis to reduce the processing amount and memory size in the attack: 1) to find linear expressions with as high a deviation as possible, and 2) to reduce the number of effective key bits and effective text bits. We have succeeded in attacking FEAL-8 in about 1 hour on a low-end workstation (SPARCstation 10 Model 30). We have confirmed that the entire set of subkeys of FEAL-8 can be derived from 225 known plaintexts with a success rate of over 70%, and from 226 known plaintexts with a success rate of almost 100%.

The addition chain (A-chain) and addition-subtraction chain (AS-chain) are efficient tools to calculate power Me (or multiplication eM), where integere is fixed andM is variable. Since the optimization problem to find the shortest A (or AS)-chain is NP-hard, many algorithms to get a sub-optimal A (or AS)-chain in polynomial time are proposed. In this paper, a window method for the AS-chain and an extended window method for the A-chain and AS-chain are proposed and their performances are theoretically evaluated by applying the theory of the optimal variable-to-fixed length code, i. e. , Tunstall code, in data compression. It is shown by theory and simulation that the proposed algorithms are more efficient than other algorithms in practical cases in addition to the asymptotic case.

S-boxes (vector output Boolean functions) should satisfy cryptographic criteria even if some input bits (say, k bits) are kept constant. However, this kind of security has been studied only for scalar output Boolean functions. SAC (k) is a criterion for scalar output Boolean functions of this type. This paper studies a generalization of SAC (k) to vector output Boolean functions as the first step toward the security of block ciphers against attacks which keep some input bits constant. We first show a close relationship between such Boolean functions and linear error correcting codes. Then we show the existence, bounds and enumeration of vector Boolean functions which satisfy the generalized SAC (k). A design method and examples are also presented.

The proposed high-power-factor converter is constructed with a flyback converter, and locates the energy-storage capacitor on the secondary side of the transformer. A high power-factor can be obtained without needing to detect any current, and the ZVS operation can be achieved without auxiliary switches. To make the best use of these advantages in the converter, ZVS operations and power-factor characteristics in the converter were analyzed. From the analytical results, the effective control method for achieving ZVS was examined. Using a bread-board circuit controlled by this method, a power-factor of 0.99 and a conversion efficiency of 88% were measured.

This paper proposes an analysis method of scaling-factor-quantization error in fractal image coding using a state-space approach with the statistical analysis method. It is shown that the statistical analysis method is appropriate and leads to a simple result, whereas the deterministic analysis method is not appropriate and leads to a complex result for the analysis of fractal image coding. We derive the output error variance matrix for the measure of error and define the output error variance by scalar quantity as the mean of diagonal elements of the output error variance matrix. Examples are given to show that the scaling-factor-quantization error due to iterative computation with finite-wordlength scaling factors degrades the quality of decoded images. A quantitative comparison of experimental scaling-factor-quantization error with analytical result is made for the output error variance. The result shows that our analysis method is valid for the fractal image coding.

We consider the situation that plural degraded images are obtained. When no prior knowledge about original images are known, these images are individually restored by an optimum restoration filter, for example, by Wiener Filter or by Projection Filter. If correlations between original images are obtained, some restoration filters based on Wiener Filter or Projection Filter are proposed. In this paper, we deal with the case that some pixels or some parts of original images overlap, and propose a restoration method using a formulae for overlapping. The method is based on Partial Projection Filter. Moreover, we confirm an efficacy of the proposed method by numerical examples.

A genetic algorithm is presented for the blind-deconvolution problem of image restoration without any a priori information about object image or blurring function. The restoration problem is modeled as an optimization problem, whose cost function is to be minimized based on mechanics of natural selection and natural genetics. The applicability of GA for blind-deconvolution problem was demonstrated.

In this paper, three issues concerning the linear adaptive receiver using the LMS algorithm for single-user demodulation in direct-sequence/code-division multiple-access (DS/CDMA) systems are considered. First, the convergence rate of the LMS algorithm in DS/CDMA environment is considered theoretically. Both upper and lower bounds of the eigenvalue spread of the autocorrelation matrix of receiver input signals are derived. It is cleared from the results that the convergence rate of the LMS algorithm becomes slow when the signal power of interferer is large. Second, fast converging technique using a prefilter is considered. The LMS based adaptive receiver using an adaptive prefilter adjusted by a Hebbian learning algorithm to decorrelate the input signals is proposed. Computer simulation results show that the proposed receiver provides faster convergence than the LMS based receiver. Third, the complexity reduction of the proposed receiver by prefiltering is considered. As for the reduced complexity receiver, it is shown that the performance degradation is little as compared with the full complexity receiver.