Analysis of a complexity-based pruning scheme for classification trees

A complexity-based pruning procedure for classification trees is described, and bounds on its finite sample performance are established. The procedure selects a subtree of a (possibly random) initial tree in order to minimize a complexity penalized measure of empirical risk. The complexity assigned to a subtree is proportional to the square root of its size. Two cases are considered. In the first, the growing and pruning data sets are identical, and in the second, they are independent Using the performance bound, the Bayes risk consistency of pruned trees obtained via the procedure is established when the sequence of initial trees satisfies suitable geometric and structural constraints. The pruning method and its analysis are motivated by work on adaptive model selection using complexity regularization.     

On the existence of <Emphasis Type="Bold">1, 3</Emphasis>P <Emphasis Type="Bold">o</Emphasis> resonances in H<Superscript>-</Superscript> between n = 2 and 3 H thresholds

We examine the influence of relativistic and QED effects on the existence of the ^1,3P ^o H^- resonances between n = 2 and 3 hydrogen thresholds, the relativistic and QED corrections and the coupling effects between the high singlet and triplet states are considered as first-order perturbations. We firstly obtain accurate non-relativistic resonant energies and widths of fifteen ¹P ^o resonances, and fifteen ³P ^o resonances. The fifteen ¹P ^o resonances are classified to be ₃ (2, 0) ^- _n ( 4 ? n ? 12) and ₃ (1, 1) ⁺ _n ( 3 ? n ? 8). The fifteen ³P ^o resonances are classified to be ₃ (2, 0) ⁺ _n ( 3 ? n ? 12) and ₃ (1, 1) ^- _n ( 4 ? n ? 8). We found there exist six Feshbach resonances for ₃ (2, 0) ^- _n (¹P ^o ) series, four Feshbach resonances for ₃ (1, 1) ⁺ _n (¹P ^o ) series, seven Feshbach resonances for ₃ (2, 0) ⁺ _n (³P ^o ) series, and three Feshbach resonances for ₃ (1, 1) ^- _n (¹P ^o ) series. Received 22 February 2002 Published online 24 September 2002     

Prediction error method for second-order blind identification

Blind channel identification methods based on the oversampled channel output are a problem of current theoretical and practical interest. In this paper, we introduce a second-order blind identification technique based on a linear prediction approach. In contrast to eigenstructure-based methods, it will be shown that the linear prediction error method is “robust” to order overdetermination. An asymptotic performance analysis of the proposed estimation method is carried out, consistency and asymptotic normality of the estimates is established. A closed-form expression for the asymptotic covariance of the estimates is given. Numerical simulations and investigations are finally presented to demonstrate the potential and the “robustness” of the proposed method     

Mikrotechnik für Raumfahrtanwendungen

This contribution describes the preparation, based upon a chemically amplified novolak resist (CAR), electron beam lithography, and ECR plasma etching, of structures with a high aspect ratio (10∶1) and lateral dimensions in the sub-micrometer range (150nm–300nm) which may serve as collector surfaces for sub-μm dust particles in a space experiment.     

Determination of diffusing capacity by modeling the dynamics of C18O uptake in newborns

Dynamic modeling of lung C¹⁸O diffusion is used to measure the C¹⁸O transfer factor (TL_co) of 14 newborns aged 1-4 mo. The model equation is based on the alveolar fractions of C¹⁸O and on changing alveolar ventilation induced by the rebreathing conditions. The model does not involve the volume of the rebreathing bag which is usually needed when applying rebreathing technique add which is a source of error. The equation is discretized and solved for recorded data obtained with equipment adapted to use in newborns. A least-square parameter calculation technique is applied to estimate TL_co. Results show a strong relationship between this index and the biometrical ones and confirm those found in the literature featuring that the measurement duration can be considerably shortened     

Fields and currents in the organs of the human body when exposed topower lines and VLF transmitters

A study is made of the electric fields and currents induced in the organs of the human body when exposed to high-voltage 50-60-Hz transmission lines and 10-30-kHz high-power transmitters. Relevant analyses previously carried out are summarized and supplemented with detailed investigations that complete the picture. Incomplete, misleading, and incorrect statements and methods in the related literature are pointed out, completed, and corrected. The major contribution is to provide quantitatively accurate, relatively simple analytic formulas that relate the incident electric field to the induced field in the organs of the body. The formulation and solution of the underlying integral equation are carried out in an Appendix     

Long-wavelength 640×486 GaAs-AlGaAs quantum well infraredphotodetector snap-shot camera

A 9-μm cutoff 640×486 snap-shot quantum well infrared photodetector (QWIP) camera has been demonstrated. The performance of this QWIP camera is reported including indoor and outdoor imaging. The noise equivalent differential temperature (NEΔT) of 36 mK has been achieved at 300 K background with f/2 optics. This is in good agreement with expected focal plane array sensitivity due to the practical limitations on charge handling capacity of the multiplexer, read noise, bias voltage, and operating temperature     

Method for error recovery of line spectral frequencies

The authors have exploited the ideas used in vector quantisation for error recovery of scalar quantised LSFs. The good performance of this method has provided high resistance of the LSFs to channel errors, outperforming other schemes by, possibly, a considerable margin. Better objective and subjective performances were obtained with this new method which obviates the need for more powerful FEC schemes for transmission over noisy channels.<>