Wyniki wyszukiwania - Biblioteka Nauki

1

Fuzzy inference neural networks with fuzzy parameters

100%

Rutkowska D. , Hayashi Y.

TASK Quarterly : scientific bulletin of Academic Computer Centre in Gdansk

|

2003

|

tom Vol. 7, No 1

7-22

EN

This paper concerns fuzzy neural networks and fuzzy inference neural networks, which are two different approaches to neuro-fuzzy combinations. The former is a direct fuzzification of artificial neural networks by introducing fuzzy signals and fuzzy weights. The latter is a representation of fuzzy systems in the form of multi-layer connectionist networks, similar to neural networks. Parameters of membership functions (centers and widths) play the role of neural network weights. In this paper, fuzzy inference neural networks with fuzzy parameters are considered. Neuro-fuzzy systems of this kind utilize both approaches: fuzzy neural networks and fuzzy inference neural networks. They also pertain to fuzzy systems of type 2 since membership functions with fuzzy parameters characterize type 2 fuzzy sets. Various architectures of these networks have been obtained for fuzzy systems based on different fuzzy implications. By analogy with fuzzy inference neural networks with crisp parameters, methods of learning fuzzy parameters and rule generation can be derived for neuro-fuzzy systems with fuzzy parameters. Fuzzy inference neural networks are studied in the framework of fuzzy granulation. In particular, fuzzy clustering as fuzzy information granulation is proposed to be applied in order to generate fuzzy IF-THEN rules. Applications of fuzzy inference neural networks are also outlined.

2

Characterization of symbolic rules embedded in deep DIMLP networks : a challenge to transparency of deep learning

100%

Bologna G. , Hayashi Y.

Journal of Artificial Intelligence and Soft Computing Research

|

2017

|

tom Vol. 7, No. 4

265--286

EN

Rule extraction from neural networks is a fervent research topic. In the last 20 years many authors presented a number of techniques showing how to extract symbolic rules from Multi Layer Perceptrons (MLPs). Nevertheless, very few were related to ensembles of neural networks and even less for networks trained by deep learning. On several datasets we performed rule extraction from ensembles of Discretized Interpretable Multi Layer Perceptrons (DIMLP), and DIMLPs trained by deep learning. The results obtained on the Thyroid dataset and the Wisconsin Breast Cancer dataset show that the predictive accuracy of the extracted rules compare very favorably with respect to state of the art results. Finally, in the last classification problem on digit recognition, generated rules from the MNIST dataset can be viewed as discriminatory features in particular digit areas. Qualitatively, with respect to rule complexity in terms of number of generated rules and number of antecedents per rule, deep DIMLPs and DIMLPs trained by arcing give similar results on a binary classification problem involving digits 5 and 8. On the whole MNIST problem we showed that it is possible to determine the feature detectors created by neural networks and also that the complexity of the extracted rulesets can be well balanced between accuracy and interpretability.

3

Fundamental Study of Evaluation at Berthing Training for Pilot Trainees Using a Ship Maneuvering Simulator

80%

Inoue K. , Okazaki T. , Murai K. , Hayashi Y.

TransNav : International Journal on Marine Navigation and Safety of Sea Transportation

|

2013

|

tom Vol. 7, no. 1

135--141

EN

Use of the ship maneuvering simulator (SMS) is at the core of pilot trainees education and training, so it is desirable to have an evaluation method that can be completed shortly after each berthing training session. There are basically two methods of docking maneuvering that pilot trainees learn: one in which the ship enters from outside the port and is berthed directly at the target quay, and a second method in which the vessel carries out a turn in front of the target quay before berthing. The authors suggested an evaluation index in a previous study concerning the first docking method. In the present study, the authors propose an evaluation method for the case of berthing the vessel using the turning maneuver. Since the index obtained by this method offers a single numerical benchmark, it is an easy–to‐understand result of the training exercise. The authors carried out experiments using a SMS and confirmed that the proposed evaluation method is effective and helpful to improve the effectiveness of SMS training.

4

Control of a redundant 7DOF upper-limb power-assist exoskeleton robot

80%

Hayashi Y. , Kiguchi K. , Dubey R.

Journal of Artificial Intelligence and Soft Computing Research

|

2011

|

tom Vol. 1, No. 3

207--214

EN

Many kinds of power-assist robots have been developed in order to assist daily activities or rehabilitation of the elderly or physically weak persons. Upper-limb power-assist robots are important to assist in many daily life activities such as eating, drinking, etc.. A human upper-limb has 7 degrees of freedom to achieve various tasks dexterously. Therefore, to assist all upper-limb joint motions of a human, the upper-limb power-assist robot is required to have 7DOF. To achieve a desired task, a person moves the hand to the desired position and orientation and/or applies certain amount of force to the target. Therefore, it is important that the upper-limb power-assist robots help control the hand position/ orientation or hand force of the user. However, the hand position/orientation or hand force is 6-dimensional vector, so the 7DOF upper-limb power-assist robot has a redundancy and in general, a pseudo-inverse matrix is used in the control. In this paper, an optimal control method is proposed to deal with the redundancy of the upper-limb power-assist exoskeleton robot considering comfortable motion of the user. The motion intention and the comfort of the user are taken into account in the pro-posed method. The effectiveness of the proposed method was evaluated by the experiments.

5

Proposal for Global Standard Maneuvering Orders for Tugboats

80%

Ishikura A. , Sugita K. , Hayashi Y. , Murai K.

TransNav : International Journal on Marine Navigation and Safety of Sea Transportation

|

2013

|

tom Vol. 7, no. 4

509--513

EN

The use of “Standard Maneuvering Orders” for tugboats, vocabulary and phrases mutually preagreed between ships and tugboats, is essential for the former to provide clear direction for the latter when berthing or un‐berthing safely. Tugboats will need time to change their posture before they take actions in response to orders from persons responsible for ships’ maneuvering. Therefore, when giving directions to change tugboats’ posture, persons who handle their ships are required to send out tug orders, with regard to “delay time,” a gap be‐tween the orders from ships and the actions taken by tugboats. “Tug Orders” standardized and used in Japan are composed of the following three factors concerning towage work: tugboat’s motion, direction and engine power, but the author’s research shows that there are “Non‐standard” special maneuvering orders other than those “standardized,” which causes such problems as a gap in perception between pilots and tugboat’s opera‐tors, etc. The purpose of this paper is to research the delay time between orders for and actions by tugboats and consider the appropriate and safe timing of providing instructions to them, and then to propose globally‐authorized “Standard Maneuvering Orders for tugboats”, discussing a problem involved in the use of the special orders used in Japan, and the way in which tug orders are used in other countries.

6

Involvement of apoptotic protease cascade for tissue destruction in Sjogren's syndrome

80%

Hayashi Y. , Yayagi K. , Haneji N.

Archivum Immunologiae et Therapiae Experimentalis

|

2000

|

tom 48

|

nr 5

7

Multimodal neurosurgery force feedback system based on mesh fusion modeling

70%

Kuroda Y. , Kamada K. , Hayashi Y. , Imura M. , Oshiro O.

Biocybernetics and Biomedical Engineering

|

2011

|

tom Vol. 31, no. 2

33-50

EN

Virtual reality based force feedback system is spotlighted as a safe and efficient training environment to obtain surgical skills. Neurosurgery utilizes multimodal patient images for visualization of a variety of functions in head. The aim of this study is to establish a concept of multimodal neurosurgery force feedback system based on mesh fusion modeling. In the model of mesh fusion, we developed an algorithm to detect overlapped region between the multiple meshes that are obtained from multimodal images, and to determine a new boundary between the meshes. Then, the method solved interaction between the newly defined mesh boundaries using the interaction model based on a finite element method. The proposed method was implemented, and applied to both simple and patient datasets for evaluating its applicability. As a result, the method succeeded to be applied to both simple and patient datasets. Finally, we demonstrated the early stage of the surgical approach in neurosurgery. Simulation results showed a real-time simulation of brain tissue deformation with force feedback.

8

Roles of endogenous prostaglandins and cyclooxygenase isozymes in mucosal defense of inflamed rat stomach

70%

Takeeda M. , Hayashi Y. , Yamato M. , Murakami M. , Takeuchi K.

Journal of Physiology and Pharmacology

|

2004

|

tom 55

|

nr 1,2

EN

Endogenous prostaglandins (PGs) are involved in adaptive gastric protection against acute injury, and cyclooxygenase (COX)-1 is responsible for the production of PGs in this phenomenon. In the present study, we examined the effect of various COX inhibitors on gastric ulcerogenic and acid secretory responses following daily exposure of the stomach to iodoacetamide (IA) and investigated the role for COX isozyme in gastric protection under subchronic mucosal irritation. Gastric mucosal irritation was induced by addition of 0.1% IA to drinking water, and the gastric mucosa was examined on the 6th day. Indomethacin (5 mg/kg) or SC-560 (selective COX-1 inhibitor, 5 mg/kg) or rofecoxib (selective COX-2 inhibitor, 5 mg/kg) was given p.o. twice 24 hr and 3 hr before the termination of IA treatment. Giving IA in drinking water for 5 days produced minimal damage in the stomach. The damage was significantly worsened by indomethacin, resulting in hemorrhagic lesions. Both SC-560 and rofecoxib also aggravated such lesions, although the effect of rofecoxib was more pronounced. Treatment with IA decreased acid secretion in pylorus-ligated stomachs, and this change was significantly reverted by indomethacin as well as SC-560 and rofecoxib. Mucosal PGE2 content was increased following IA treatment, with apparent expression of COX-2 mRNA in the stomach, and the increased PGE2 production was significantly suppressed by SC-560 and rofecoxib as well as indomethacin. These results suggest that endogenous PGs derived from both COX-1 and COX-2 are involved in the mucosal defense of the inflamed stomach, partly by decreasing acid secretion and contribute to maintaining the mucosal integrity under such conditions.