Prediction error property of the Lasso Estimator and its generalization

The lasso procedure is an estimator-shrinkage and variable selection method. This paper shows that there always exists an interval of tuning parameter values such that the corresponding mean squared prediction error for the lasso estimator is smaller than for the ordinary least squares estimator. For an estimator satisfying some condition such as unbiasedness, the paper defines the corresponding generalized lasso estimator. Its mean squared prediction error is shown to be smaller than that of the estimator for values of the tuning parameter in some interval. This implies that all unbiased estimators are not admissible. Simulation results for five models support the theoretical results.

Reaching out to promote physical activity in Australia: a statewide randomized controlled trial of a stage-targeted intervention

Purpose. This study examined the broader use of a print-media intervention, which was previously shown to be effective at promoting physical activity to participants recruited from a regional Australian community, as a strategy suitable for a more diverse statewide population sample.
Methods. Participants were randomly selected adults who responded to a telephone interview conducted by the New South Wales Health Department and consented to Participate in a randomized controlled trial. Consenters were allocated to either intervention (n = 361) or control (n = 358) conditions. The intervention, a personalized letter plus stage-targeted booklets, was sent 1 week postbaseline. Data were collected via telephone interview at baseline and 2 and 8 months and were analyzed using repeated measures analysis of variance (ANOVA) and mean squared statistics.
Results. The groups were similar at baseline (mean age 43 +/- 3 years; 64 % women). Process evaluation showed high intervention recall (76 % at 2 months) and high follow-up response rates (> 85 % at 8 months) were achieved. Nonsignificant increases in physical activity were observed (Fl,719 = 2.18, p = .14).
Discussion. A single mailing of stage-targeted print materials was not effective in promoting increases in physical activity among participants selected from the statewide population. Future research could examine how the effectiveness of print media might be enhanced, possibly by using supplementary media, community-based prompts, or other incentives.

Forecasting inflation, industrial output and exchange rates : a template study for India

This paper reports results from a forecasting study for inflation, industrial output and exchange rates for India. We cannot reject the null hypothesis for linearity for all series used except for the growth rate of the foreign exchange series and our analysis is based on linear models, ARIMA and bivariate transfer functions and restricted VAR. Forecasting performance is evaluated using the models’ root mean-squared error differences and Theil’s inequality coefficients from recursive origin static, fixed origin dynamic and rolling origin dynamic forecasts. For models based on weekly data, based on RMSEs, we find that the bivariate models improve upon the forecasts of the ARIMA model while for models based on monthly data the ARIMA model has almost always better performance. In choosing between the two bivariate models on the basis of RMSEs, our overall results tend to support the use of a restricted VAR, as this model had the best forecasting performance more frequently than the transfer function model.

Performances of three minimally invasive cardiac output monitoring systems

We evaluated cardiac output (CO) using three new methods – the auto-calibrated FloTrac–Vigileo (CO_ed), the non-calibrated Modelflow (CO_mf ) pulse contour method and the ultra-sound HemoSonic system (CO_hs) – with thermodilution (CO_td) as the reference. In 13 postoperative cardiac surgical patients, 104 paired CO values were assessed before, during and after four interventions: (i) an increase of tidal volume by 50%; (ii) a 10 cm H₂O increase in positive end-expiratory pressure; (iii) passive leg raising and (iv) head up position. With the pooled data the difference (bias (2SD)) between CO_ed and CO_td, CO_mf and CO_td and CO_hs and CO_td was 0.33 (0.90), 0.30 (0.69) and −0.41 (1.11) l.min⁻¹, respectively. Thus, Modelflow had the lowest mean squared error, suggesting that it had the best performance. CO_ed significantly overestimates changes in cardiac output while CO_mf and CO_hs values are not significantly different from those of CO_td. Directional changes in cardiac output by thermodilution were detected with a high score by all three methods.

Estimating relative disulfide energies : an accurate ab initio potential energy surface

Disulfide torsional energy, a good predictor of disulfide redox potential in proteins, may be estimated by interpolation on a potential energy surface (PES) describing the twisting of diethyl disulfide through its three central dihedral angles. Here we update PES calculations at the M05-2X level of theory with the 6-31G(d) basis set. Although the surface shows no qualitative differences from an earlier MP2(full) PES, energy differences greater than 1 kJ mol^–1 were seen for conformations with χ₂ between –60° and 30°, or with χ₃ below 60° or above 130°. This is particularly significant for highly strained disulfides that are likely to be spontaneously reduced by mechanical means. In benchmarking against the high-level G3X method, M05-2X showed significantly reduced root mean squared deviation compared with MP2(full) (1.0 versus 2.0 kJ mol^–1 respectively). Results are incorporated into a web application that calculates relative torsional energies from disulfide dihedral angles (http://www.sbinf.org/applications/pes.html).

Data-driven computation of contact dynamics during two-point manipulation of deformable objects

This study represents a preliminary step towards data-driven computation of contact dynamics during manipulation of deformable objects at two points of contact. A modeling approach is proposed that characterizes the individual interaction at both points and the mutual effects of the two interactions on each other via a set of parameters. Both global as well as local coordinate systems are tested for encoding the contact mechanics. Artificial neural networks are trained on simulated data to capture the object behavior. A comparison of test data with the output of the trained system reveals a mean squared error percentage between 1% and 3% for simple interactions.

Multi-label classification with clustering for image and text categorization

This paper explores effective multi-label classification methods for multi-semantic image and text categorization. We perform an experimental study of clustering based multi-label classification (CBMLC) for the target problem. Experimental evaluation is conducted for identifying the impact of different clustering algorithms and base classifiers on the predictive performance and efficiency of CBMLC. In the experimental setting, three widely used clustering algorithms and six popular multi-label classification algorithms are used and evaluated on multi-label image and text datasets. A multi-label classification evaluation metrics, micro F1-measure, is used for presenting predictive performances of the classifications. Experimental evaluation results reveal that clustering based multi-label learning algorithms are more effective compared to their non-clustering counterparts.

An accelerated particle swarm optimization based levenberg marquardt back propagation algorithm

The Levenberg Marquardt (LM) algorithm is one of the most effective algorithms in speeding up the convergence rate of the Artificial Neural Networks (ANN) with Multilayer Perceptron (MLP) architectures. However, the LM algorithm suffers the problem of local minimum entrapment. Therefore, we introduce several improvements to the Levenberg Marquardt algorithm by training the ANNs with meta-heuristic nature inspired algorithm. This paper proposes a hybrid technique Accelerated Particle Swarm Optimization using Levenberg Marquardt (APSO_LM) to achieve faster convergence rate and to avoid local minima problem. These techniques are chosen since they provide faster training for solving pattern recognition problems using the numerical optimization technique.The performances of the proposed algorithm is evaluated using some bench mark of classification’s datasets. The results are compared with Artificial Bee Colony (ABC) Algorithm using Back Propagation Neural Network (BPNN) algorithm and other hybrid variants. Based on the experimental result, the proposed algorithms APSO_LM successfully demonstrated better performance as compared to other existing algorithms in terms of convergence speed and Mean Squared Error (MSE) by introducing the error and accuracy in network convergence.

An adaptive complementary filter for inertial sensor based data fusion to track upper body motion

  Remote human activity monitoring is critical and essential in physiotherapy with respect to the skyrocketing healthcare expenditure and the fast aging population. One of frequently used method to monitor human activity is wearing inertial sensors since it is low-cost and accurate. However, the measurements of those sensors are able only to estimate the orientation and rotation angles with respect to actual movement angles, because of differences in the body’s co-ordination system and the sensor’s co-ordination system. There were numerous studies being conducted to improve the accuracy of estimation, though there is potential for further discussions on improving accuracy by replacing heavy algorithms to less complexity. This research is an attempt to propose an adaptive complementary filter for identifying human upper arm movements. Further, this article discusses a feasibility of upper arm rehabilitation using the proposed adaptive complementary filter and inertial measurement sensors. The proposed algorithm is tested with four healthy subjects wearing an inertial sensor against gold standard, which is the VICON system. It demonstrated root mean squared error of 8.77◦ for upper body limb orientation estimation when compared to gold standard VICON optical motion capture system.

Feasible estimation in cointegrated panels

In this paper we propose a simple procedure for data dependent determination of the number of lags and leads to use in feasible estimation of cointegrated panel regressions. Results from Monte Carlo simulations suggests that the feasible estimators considered enjoys excellent precision in terms of root mean squared error and reasonable power with effective size hovering close to the nominal level. The good performance of the feasible estimators is verified empirically through an application to the long run money demand.

Channel estimation for time-varying MIMO relay systems

In this paper, we investigate the channel estimation problem for multiple-input multiple-output (MIMO) relay communication systems with time-varying channels. The time-varying characteristic of the channels is described by the complex-exponential basis expansion model (CE-BEM). We propose a superimposed channel training algorithm to estimate the individual first-hop and second-hop time-varying channel matrices for MIMO relay systems. In particular, the estimation of the second-hop time-varying channel matrix is performed by exploiting the superimposed training sequence at the relay node, while the first-hop time-varying channel matrix is estimated through the source node training sequence and the estimated second-hop channel. To improve the performance of channel estimation, we derive the optimal structure of the source and relay training sequences that minimize the mean-squared error (MSE) of channel estimation. We also optimize the relay amplification factor that governs the power allocation between the source and relay training sequences. Numerical simulations demonstrate that the proposed superimposed channel training algorithm for MIMO relay systems with time-varying channels outperforms the conventional two-stage channel estimation scheme.

Channel estimation for two-way MIMO relay systems in frequency-selective fading environments

© 2002-2012 IEEE. In this paper, we investigate the channel estimation problem for two-way multiple-input multiple-output (MIMO) relay communication systems in frequency-selective fading environments. We apply the method of superimposed channel training to estimate the individual channel state information (CSI) of the first-hop and second-hop links for two-way MIMO relay systems with frequency-selective fading channels. In this algorithm, a relay training sequence is superimposed on the received signals at the relay node to assist the estimation of the second-hop channel matrices. The optimal structure of the source and relay training sequences is derived to minimize the mean-squared error (MSE) of channel estimation. Moreover, the optimal power allocation between the source and relay training sequences is derived to improve the performance of channel estimation. Numerical examples are shown to demonstrate the performance of the proposed superimposed channel training algorithm for two-way MIMO relay systems in frequency-selective fading environments.

A panel data approach to economic forecasting: the bias-corrected average forecast

In this paper, we propose a novel approach to econometric forecasting of stationary and ergodic time series within a panel-data framework. Our key element is to employ the (feasible) bias-corrected average forecast. Using panel-data sequential asymptotics we show that it is potentially superior to other techniques in several contexts. In particular, it is asymptotically equivalent to the conditional expectation, i.e., has an optimal limiting mean-squared error. We also develop a zeromean test for the average bias and discuss the forecast-combination puzzle in small and large samples. Monte-Carlo simulations are conducted to evaluate the performance of the feasible bias-corrected average forecast in finite samples. An empirical exercise based upon data from a well known survey is also presented. Overall, theoretical and empirical results show promise for the feasible bias-corrected average forecast.

A panel data approach to economic forecasting: the bias-corrected average forecast

In this paper, we propose a novel approach to econometric forecasting of stationary and ergodic time series within a panel-data framework. Our key element is to employ the bias-corrected average forecast. Using panel-data sequential asymptotics we show that it is potentially superior to other techniques in several contexts. In particular it delivers a zero-limiting mean-squared error if the number of forecasts and the number of post-sample time periods is sufficiently large. We also develop a zero-mean test for the average bias. Monte-Carlo simulations are conducted to evaluate the performance of this new technique in finite samples. An empirical exercise, based upon data from well known surveys is also presented. Overall, these results show promise for the bias-corrected average forecast.

A panel data approach to economic forecasting: the bias-corrected average forecast

In this paper, we propose a novel approach to econometric forecasting of stationary and ergodic time series within a panel-data framework. Our key element is to employ the (feasible) bias-corrected average forecast. Using panel-data sequential asymptotics we show that it is potentially superior to other techniques in several contexts. In particular, it is asymptotically equivalent to the conditional expectation, i.e., has an optimal limiting mean-squared error. We also develop a zeromean test for the average bias and discuss the forecast-combination puzzle in small and large samples. Monte-Carlo simulations are conducted to evaluate the performance of the feasible bias-corrected average forecast in finite samples. An empirical exercise, based upon data from a well known survey is also presented. Overall, these results show promise for the feasible bias-corrected average forecast.