Next: Computer Education Laboratory Up: Department of Computer Previous: Computer Communications Laboratory

Multimedia Devices Laboratory


/ Hikaru Date / Professor
/ Qiangfu Zhao / Associate Professor
/ Jintae Lee / Assistant Professor

The multimedia devices presently available add sound, still pictures, and moving pictures to traditional computer displays which show only text and diagrams. The addition of these few functions, however, offers remarkable convenience to the user of such multimedia devices. For this reason, it is widely believed that the impact of these devices on human society in the near future will be extremely great.

The properties desirable for multimedia devices can be visualized on a graph with two axes. The horizontal axis represents a device's capacity for depth and variety of communication and the vertical axis represents the capacity for naturalness and high fidelity. Ideally, multimedia devices should be far along both axes if they are to offer the most comfortable and efficient human interfaces. Unfortunately, progress along these dimensions has not gone far enough. Prolonged use of computers still fatigues the body and fails to match the depth and quality of communication that can exist between humans in face-to-face encounters.

For example, human beings can communicate with great ease and variety through the spoken word. Computers can't. In spite of the fact that speech synthesis by computers is almost perfect, speech recognition has not reached the level of practical application that natural speech can achieve. Furthermore speech recognition algorithms are overly sensitive to existing noise. The area of speech recognition is still in the primitive stages and needs to evolve further to make multimedia devices more effective.

Future multimedia systems should be able to accept and recognize natural information. While traditional multimedia systems provide more and more information, future systems should be able to select, and provide useful and important information. One project in this laboratory is to study how to build such kind of intelligent multimedia systems. Our short-term goal is to build a system that can recognize 3D characters and documents. The long-term goal is to construct a system that can recognize nature images and speeches.

Sign language is another means of human communication that computers cannot participate in as easily as those who are forced to use it because of hearing impairment. If a computer could be made to communicate via this medium, far more people could access intellectual services and actively participate in multimedia that have been unable to do so thus far. At present, pioneer research to incorporate this dimension of communication as far as synthesis and recognition technology has just begun in our laboratory.

As for the dimension of high fidelity, the digitalization of high definition TV is one example of progress that has been made along the vertical axis. Signal processing technology is indispensable for the generation, processing, and recognition of visual, auditory and control signals. Furthermore, multimedia devices are preferable to work in real-time. These requirements can be fulfilled not only by high computing processor speeds and large memories, but also by fast algorithms. Therefore, the study of signal processing is fundamental in our laboratory.


Refereed Journal Papers

  1. Zhao, Q. F. and Higuchi, T., Efficient learning of NN-MLP based on individual evolutionary algorithm. Neurocomputing, vol. 13, p. 201-215, 1996.

    The nearest neighbor based multilayer perceptron (NN-MLP) is a suitable model for self-organization, and has been studied by many authors in different forms. However, a large number of neurons are usually required in this kind of networks. To obtain smaller or the smallest NN-MLP, this paper introduces the concept of individual evolutionary algorithm (IEA), and proposes a new method for NN-MLP learning. There are four basic operations in the IEA: competition, gain, loss and retraining. The basic rule is: all individuals compete for surviving, winners gain more, losers lose more, and the individuals are retrained to function better than before. The learning algorithm based on the IEA is simple and suitable for parallel realization, and is able to produce the ``smallest-at-present'' networks from random ones in an evolutionary manner. Its efficiency is shown by experimental results.

  2. Zhao, Q. F. and Higuchi, T., Evolutionary learning of nearest neighbor MLP. IEEE Trans. on Neural Networks, 1996. vol. 7, No. 3, p. 762-767.

    The nearest neighbor multilayer perceptron (NN-MLP) is a single-hidden-layer network suitable for pattern recognition. To design an NN-MLP efficiently, this paper proposes a new evolutionary algorithm consisting of four basic operations: recognition, remembrance, reduction and review. Experimental results show that this algorithm can produce the smallest or nearly smallest networks from random initial ones.

  3. Zhao, Q. F. and Higuchi, T., Minimization of Nearest Neighbor Classifiers Based on Individual Evolutionary Algorithm. Pattern Recognition Letters, vol. 17, p. 125-131, 1996.

    This paper introduces the individual evolutionary algorithm (IEA). Using IEA, the minimum or nearly minimum nearest neighbor classifiers can be obtained iteratively by performing four operations: competition, gain, loss and retraining. The efficiency of IEA is verified by experimental results.

Refereed Proceeding Papers

  1. Zhao, Q. F., Structural Learning of MTM-MLP Based on Individual Evolutionary Algorithm. Proc. Inter. Conf. on Evolutionary Computation, p. 341-345, IEEE, May 1996.

    The purpose of structural learning is to determine the structure as well as the weights of a neural network. For large-scaled networks, structural learning is a very difficult problem. To solve this problem efficiently, we have proposed the individual evolutionary algorithm (IEA), which can produce NN-MLP (nearest neighbor based multilayer perceptron) with the least or almost least hidden neurons. In this paper, we apply IEA to the learning of MTM-MLP (multi-template matching based MLP). The goal here is to obtain networks with the least hidden neurons as well as the least synapses. The performance of the IEA is shown by experimental results.

  2. Zhao, Q. F., On-Line Evolutionary Learning of NN-MLP Based on the Attentional Learning Concept. Proc. Inter. Conf. on Neural Networks, p. 403-408, IEEE, June 1996.

    To design the nearest neighbor based multilayer perceptron (NN-MLP) efficiently, the author has proposed a new evolutionary learning algorithm called the R4-rule. For off-line learning, the R4-rule can produce the smallest or nearly smallest networks with high generalization ability by iteratively performing four basic operations: recognition, remembrance, reduction and review. To apply the algorithm to on-line evolutionary learning of NN-MLP, this paper proposes some improvements for the R4-rule based on the attentional learning concept. The performance of the improved algorithm is verified by experimental results.

  3. Zhao, Q. F., Neural Network Learning Based on Co-Evolution. Proc. Inter. Conf. on Neural Networks, Plenary, panel and special sessions, p. 124-129, IEEE, June 1996.

    Learning of large-scaled neural networks is a very difficult problem. The genetic algorithms (GAs) provide high possibility for obtaining the global optimal solutions, but the computational amount is so large that only toy-like problems can be solved. This paper proposes a new learning algorithm based on the co-evolution concept. The basic idea is divide-and-conquer: divide the network into many small subnets, find the subnets using GAs, and put the subnets together again. Three questions are to be answered in this paper: 1) how to define the individuals ? 2) how to define the fitness of individuals ? and 3) how to get the desired network from the individuals ?

  4. Zhao, Q. F., Co-Evolutionary Algorithm: A New Approach to Evolutionary Learning. Proc. Inter. Conf. on SOFT Computing, Yamakawa, T., p. 529-532, IFSA, INNS, etc, World Scientific. Sept. 1996.

    Learning of large-scaled neural networks is a very difficult problem. The evolutionary algorithms (EAs) provide higher possibility for obtaining the global optimal solutions. However, the computational amount imposed by existing EAs is so large that only toy-like problems can be solved. This paper introduces an algorithm based on the co-evolution concept. The basic idea is : divide the network into many smaller modules, define the modules as individuals and find them by co-evolution, and then put the modules together to form the whole network.

  5. Zhao, Q. F., Hierarchical Evolutionary Algorithm Based on IEA and Its Application to Structural Learning. Proc. World Congress on Neural Networks, p. 1076-1081, INNS, Sept.1996.

    The individual evolutionary algorithm (IEA) was proposed by the author for efficient learning of NN-MLP (nearest neighbor based multilayer perceptron). In this paper, we propose the hierarchical evolutionary algorithm (HEA) based on IEA, and apply HEA to structural learning of neural networks. To make the study more concrete, the multi-template matching based multilayer perceptron (MTM-MLP ) is used as an example. The performance of HEA is shown by experimental results.

  6. Lee, J., Physically-Based Modeling of Brush Painting. 5-th International Conference on Computational Graphics and Visualization Techniques, Santo, H. P., p. 253--259. GRASP Press, Dec. 1996.

    This paper introduces physically-based modeling of brush painting. To provide the user with 3D brushes that can be manipulated and interact with the surface of the paper like real brushes, elastic bristles are designed and fixed at the bottom of the brush holder. A linear equation system is derived to calculate the physical deflection of bristles according to the force exerted on them from the surface of the paper. Our brushes provide naturalness in computer painting, and drastically decrease the designing complexity inherent in the conventional 2D outline-based painting.

Grants

  1. Lee, J., National fund, The Science and Technology Agency of Japan. Research on automatic translation of Japanese text into Japanese sign language, assistive engineering for participation of aged and disabled persons, Dec. 1996.

Others

  1. Asai, K., Bachelor Thesis: A study on neural network based character recognition. Univ. of Aizu, 1996. Thesis Advisor: Q. F. Zhao.

  2. Kimura, M., Bachelor Thesis: A study on preprocessing for handwritten character recognition. Univ. of Aizu, 1996. Thesis Advisor: Q. F. Zhao.

  3. Suzuki, T., Bachelor Thesis: A study on automatic design of a simple OCR system for handwritten digits. Univ. of Aizu, 1996. Thesis Advisor: Q. F. Zhao.

  4. Shirasaka, M., Bachelor Thesis: A study on invariant character recognition. Univ. of Aizu, 1996, Thesis Advisor: Q. F. Zhao.

  5. Tetsuka, M., Bachelor Thesis: A study on visualization of IEA based concept learning. Univ. of Aizu, 1996, Thesis Advisor: Q. F. Zhao.

  6. Fukuda, M., Bachelor Thesis: Kinematics and Inverse Kinematics of the Human Arm. Univ. of Aizu, 1996, Thesis Advisor: J. T. Lee.

  7. Itakura, H., Bachelor Thesis: Mouth Shape Simulation. Univ. of Aizu. 1996, Thesis Advisor: J. T. Lee.

  8. Kanomata, S., Bachelor Thesis: The Grammatical Structure of Japanese Sign Language. Univ. of Aizu, 1996, Thesis Advisor: J. T. Lee.

  9. Moriya, Y., Bachelor Thesis: A Study on Hand Surface Deformation, Univ. of Aizu, 1996. Thesis Advisor: J. T. Lee.

Academic Activities

  1. Qiangfu Zhao, IEEE, May 1996, Organizer of Special session in IEEE ICEC'96, Nagoya.

  2. Qiangfu Zhao, IEEE, June 1996. Organizer and chairman of Special sessions in IEEE ICNN'96, Washintong, D.C.

  3. Qiangfu Zhao, Oct. 1996. Invited talk in Beijing Institute of Technology, Beijing, China.

  4. Jintae Lee, Aug. 1996. Invited talk in Systems Engineering Research Center, Taejon, Korea.



Next: Computer Education Laboratory Up: Department of Computer Previous: Computer Communications Laboratory


www@u-aizu.ac.jp
October 1997