Dynamic cluster generation for a fuzzy classifier with ellipsoidalregions

In this paper, we discuss a fuzzy classifier with ellipsoidal regions that dynamically generates clusters. First, for the data belonging to a class we define a fuzzy rule with an ellipsoidal region. Namely, using the training data for each class, we calculate the center and the covariance matrix of the ellipsoidal region for the class. Then we tune the fuzzy rules, i.e., the slopes of the membership functions, successively until there is no improvement in the recognition rate of the training data. Then if the number of the data belonging to a class that are misclassified into another class exceeds a prescribed number, we define a new cluster to which those data belong and the associated fuzzy rule. Then we tune the newly defined fuzzy rules in the similar way as stated above, fixing the already obtained fuzzy rules. We iterate generation of clusters and tuning of the newly generated fuzzy rules until the number of the data belonging to a class that are misclassified into another class does not exceed the prescribed number. We evaluate our method using thyroid data, Japanese Hiragana data of vehicle license plates, and blood cell data. By dynamic cluster generation, the generalization ability of the classifier is improved and the recognition rate of the fuzzy classifier for the test data is the best among the neural network classifiers and other fuzzy classifiers if there are no discrete input variables     

Efficient distribution-free population learning of simple concepts

We consider a variant of the ‘population learning model’ proposed by Kearns and Seung [8], in which the learner is required to be ‘distribution-free’ as well as computationally efficient. A population learner receives as input hypotheses from a large population of agents and produces as output its final hypothesis. Each agent is assumed to independently obtain labeled sample for the target concept and output a hypothesis. A polynomial time population learner is said to PAC-learn a concept class, if its hypothesis is probably approximately correct whenever the population size exceeds a certain bound which is polynomial, even if the sample size for each agent is fixed at some constant. We exhibit some general population learning strategies, and some simple concept classes that can be learned by them. These strategies include the ‘supremum hypothesis finder’, the ‘minimum superset finder’ (a special case of the ‘supremum hypothesis finder’), and various voting schemes. When coupled with appropriate agent algorithms, these strategies can learn a variety of simple concept classes, such as the ‘high–low game’, conjunctions, axis-parallel rectangles and others. We give upper bounds on the required population size for each of these cases, and show that these systems can be used to obtain a speed up from the ordinary PAC-learning model [11], with appropriate choices of sample and population sizes. With the population learner restricted to be a voting scheme, what we have is effectively a model of ‘population prediction’, in which the learner is to predict the value of the target concept at an arbitrarily drawn point, as a threshold function of the predictions made by its agents on the same point. We show that the population learning model is strictly more powerful than the population prediction model. Finally, we consider a variant of this model with classification noise, and exhibit a population learner for the class of conjunctions in this model. This revised version was published online in June 2006 with corrections to the Cover Date.     

Voltage-balancing method for IGBTs connected in series

This paper presents a new method for balancing voltages of series-connected insulated gate bipolar transistors (IGBTs). This method can be implemented only by adding simple circuits to the gate drive system of the IGBTs, and its effect of balancing the IGBT's collector-emitter voltages during the switching transients is remarkable. This principal strategy and experimental results with series-connected IGBTs are first described. After that, further experimental results are shown from the switching tests of four 2.5-kV flat-packaged IGBTs connected in series. Through the switching tests, superior characteristics of the proposed method have been confirmed.     

Reduction of defects and improvement in the performance of blue phosphor for thick‐dielectric electroluminescent displays using Color‐by‐Blue

Abstract— Defect‐free large‐area inorganic thick‐dielectric EL (TDEL) displays using Color by Blue (CBB) technology have been successfully developed. We have achieved the world's highest blue‐phosphor luminance of 900 cd/m² for a single‐pixel device by using CBB and by optimizing the e‐beam gun configuration and the flow rate of H₂S in the vacuum chamber. By analyzing the defects on panels with triple‐pattern phosphors and CBB panels, we also found that the number of defects on CBB panels can be drastically reduced compared with those on triple‐pattern panels. The defect‐free 17‐in. VGA CBB panels show better characteristics, a high peak luminance of 600 cd/m² and a high contrast ratio of 1000:1, compared with those of triple‐pattern panels.     

Left ventricular diastolic properties of hypertensive patients measured by pulsed tissue Doppler imaging

Examination of left ventricular (LV) diastolic dysfunction in hypertensive patients has been based on parameters obtained from the transmitral flow velocity during pulsed Doppler echocardiography. However, these parameters are affected by loading conditions. We evaluated LV diastolic function along the longitudinal and transverse axes by pulsed tissue Doppler imaging (TDI) in 50 hypertensive (HT) patients and 36 age-matched healthy volunteers (N). Transmitral flow velocity was recorded by pulsed Doppler echocardiography. LV posterior wall motion velocity along the longitudinal and transverse axes also was recorded by pulsed TDI. In both groups, peak early diastolic velocity of the LV posterior wall (Ew) along the transverse axis (N: 15.8+/-5.2 cm/s, HT: 12.2+/-4.4 cm/s) was higher than that along the longitudinal axis (N: 12.7+/-3.1 cm/s, HT: 9.5+/-3.3 cm/s). Peak atrial systolic velocity of the LV posterior wall (Aw) along the longitudinal axis (N: 9.1+/-1.8 cm/s, HT: 9.7 +/-2.6 cm/s) significantly exceeded that along the transverse axis (N: 8.0+/-2.2 cm/s, HT: 8.4+/-2.4 cm/s) in both groups. The Ews were lower and the Aws were higher along both axes in the patient group than in the control group. The time intervals from the aortic component of the second heart sound to the peak of the early diastolic wave (IIA-Ews) along both the transverse (N: 142+/-18 ms, HT: 154+/-19 ms) and longitudinal (N: 151 16 ms, HT: 162+/-20 ms) axes were longer in the patient group. In 29 patients, Ews along both axes correlated negatively (transverse: r = -0.80, P < .0001; longitudinal: r = -0.71, P < .0001) and IIA-Ews correlated positively (transverse: r = 0.81, P < .0001; longitudinal: r = 0.74, P < .001) with the time constant of the LV pressure decay during isovolumic diastole. The Aws along both axes in the 24 patients without pseudonormalization in transmitral flow velocity correlated positively (transverse: r = 0.60, P < .001; longitudinal: r = 0.74, P < .0001) with the LV end-diastolic pressure. In conclusion, LV relaxation and filling along the longitudinal and transverse axes were impaired in many patients with hypertension. Pulsed TDI was useful for evaluating LV diastolic dynamics in this disease.     

Clinically applicable bulk isolation of blood CD34+ cells for autografting in children

The quality of population-based cancer registries has been measured by the indices of the proportion of total incident cases (DCO%) registered by death certificate only (DCO), and the ratio of incidence to mortality (I/D ratio). Recently it has been recommended that DCO% should be used as an index for the reliability of diagnosing cancers and that the proportion of cases first notified via death certificate (DCN, DCN%) be used as an index for the completeness of registration. Parkin introduced a method to estimate the registration rate, the estimated proportion of the "true incidence" that are registered in population-based registries. We recommend a modified method for estimating the registration rate for cancer registries where DCN% is relatively high, as it is in Japan, as Parkin's method may overestimate the registration rate. The method is as follows: the registration rate = (1-DCN% x 1/D ratio)/(1-DCN%). The registration rates at the Osaka Cancer Registry between 1966 and 1992 were estimated using our method. During this period, the yearly registration rate was 74.6-78.4% for males and 69.1-73.3% for females. When the cancer cases were looked at according to site, the yearly registration rate was 74.2-81.6% for stomach cancer, 81.2-89.3% for lung cancer, and 71.3-76.9% for uterine cancer. These results show that the registration rate is high for cancers that have an unfavorable prognosis and low for cancers that have a favorable prognosis. We recommend that all cancer registries in Japan calculate the completeness of registration by utilizing DCN defined as the sum of DCO plus cases not reported as cancer but with supportive clinical information of such obtained through survey of the registry for DCN.     

Propagation properties of longitudinal leaky surface waves on lithium tetraborate

Theoretical and experimental results of longitudinal leaky surface waves with a higher phase velocity than that of ordinary leaky surface waves and a low propagation loss on lithium tetraborate (LBO) are investigated in detail. They propagate along the surface with a phase velocity close to that of longitudinal bulk wave, slightly radiating two kinds of shear bulk waves (or one shear bulk wave in the case that one of two shear wave terms is uncoupled) into the solid. Most surface components of the mode consist of a longitudinal wave term and an electromagnetic wave term. The detailed propagation properties of the longitudinal leaky surface waves on LBO with the Euler angles (phi, theta, 90 degrees ) are investigated theoretically and experimentally. The (011) cut of LBO was found to be desirable for higher frequency SAW devices. One of the reasons why that mode on LBO has a low propagation loss is also discussed.     

Effect of mean stress on fatigue properties of 1800 MPa-class spring steels

Fatigue tests were conducted for 1800 MPa-class spring steels at various stress ratios. For comparison, similar fatigue tests were conducted for conventional steels whose tensile strength was lower than 1200 MPa. The spring steels exhibited fish-eye fractures, and the origins of these fractures were oxide, TiN and the matrix itself. In contrast, the conventional steels never exhibited fish-eye fractures. The fatigue strength of these steels decreased monotonously as the stress ratio increased, when the fatigue strength was evaluated in terms of stress amplitude. However, the fatigue strength degradation was less than that expected from a modified-Goodman line, and the best fit line was obtained by connecting the fatigue limit at zero mean stress to true fracture strength instead of tensile strength. This research also reviewed application of a power low to the stress ratio effect evaluation. In these results, the difference between the spring and conventional steels was negligible.     

Bile Acid–Drug Interaction via Organic Anion-Transporting Polypeptide 4C1 Is a Potential Mechanism of Altered Pharmacokinetics of Renally Excreted Drugs

Patients with liver diseases not only experience the adverse effects of liver-metabolized drugs, but also the unexpected adverse effects of renally excreted drugs. Bile acids alter the expression of renal drug transporters, however, the direct effects of bile acids on drug transport remain unknown. Renal drug transporter organic anion-transporting polypeptide 4C1 (OATP4C1) was reported to be inhibited by chenodeoxycholic acid. Therefore, we predicted that the inhibition of OATP4C1-mediated transport by bile acids might be a potential mechanism for the altered pharmacokinetics of renally excreted drugs. We screened 45 types of bile acids and calculated the IC₅₀, K_i values, and bile acid–drug interaction (BDI) indices of bile acids whose inhibitory effect on OATP4C1 was >50%. From the screening results, lithocholic acid (LCA), glycine-conjugated lithocholic acid (GLCA), and taurine-conjugated lithocholic acid (TLCA) were newly identified as inhibitors of OATP4C1. Since the BDI index of LCA was 0.278, LCA is likely to inhibit OATP4C1-mediated transport in clinical settings. Our findings suggest that dose adjustment of renally excreted drugs may be required in patients with renal failure as well as in patients with hepatic failure. We believe that our findings provide essential information for drug development and safe drug treatment in clinics.