Time-varying Weights Research Articles

Deep Learning-Based Average Consensus

In this study, we analyzed the problem of accelerating the linear average consensus algorithm for complex networks. We propose a data-driven approach to tuning the weights of temporal (i.e., time-varying) networks using deep learning techniques. Given a finite-time window, the proposed approach first unfolds the linear average consensus protocol to obtain a feedforward signal-flow graph, which is regarded as a neural network. The edge weights of the obtained neural network are then trained using standard deep learning techniques to minimize consensus error over a given finite-time window. Through this training process, we obtain a set of optimized time-varying weights, which yield faster consensus for a complex network. We also demonstrate that the proposed approach can be extended for infinite-time window problems. Numerical experiments revealed that our approach can achieve a significantly smaller consensus error compared to baseline strategies.

Convergence on non-autonomous inertial neural networks with unbounded distributed delays

ABSTRACTIn this paper, without using reduced-order method, a class of non-autonomous inertial neural networks with time-varying connection weights and unbounded continuously distributed delays are considered. Based upon Lyapunov function method and differential inequality technique, some sufficient conditions are established to reveal that all solutions of the addressed model and their derivatives converge to zero vector, which extend and complement earlier ones in the literature. Moreover, a numerical example is provided to illustrate the correctness of the theoretical results.

Korean exchange rate forecasts using Bayesian variable selection

ABSTRACT Using Bayesian variable selection, we demonstrate that economic variables forecast Korea-US exchange rates better than random walk or random walk with drift model at a short horizon. It implies that the failure of out-of-sample exchange rate forecasts is due to the uncertainties associated with selecting proper predictors, rather than the lack of relationship between the exchange rate and its theoretical determinants. Our results also suggest that time-variant and asymmetric weights on predictors should be taken into account to understand exchange rates dynamics. (JEL classification: C11, C53, F31)

Automatic flight control design considering objective and subjective risks during carrier landing

In this article, a design scheme of automatic carrier landing system control law based on combination of the objective risk and the subjective risk is proposed, in order to improve the safety and flying quality of the landing. The nonlinear longitudinal mathematical model is constructed in the air wake turbulence condition during carrier landing, which is transformed into a linear perturbed model by the state-space equations with deviation state variables. The concepts of the objective risk and the subjective risk in the recovery of an aircraft aboard a carrier are addressed. A principle of predicting the future states based on the current ones is put forward so that a mathematic model for the objective risk is established, synthetically considering the current and future landing state deviations. For the other risk, the corresponding model is obtained by the subjective experiences of the pilots in the flight simulation tests. Furthermore, a novel model predictive control algorithm, which contains the additional subjective risk and the time-varying weights of the state terms, is proposed. Automatic carrier landing control law is built by introducing the objective risk, the subjective risk, and the effect of carrier air wake disturbance. In the rolling optimization progress, these time-varying weights are dynamically tuned according to the constantly changing objective risk to control the state deviations and suppress this risk, while the subjective risk is handled by the additional risk terms. Besides, the action of carrier air wake disturbance is considered and compensated in the derivation of the linear matrix inequalities. Test results based on a semi-physical simulation platform indicate that the new automatic carrier landing system control algorithm proposed by this article brings about an excellent carrier landing performance as well as an improved flying quality.

An evolution-dependent multi-objective ensemble model of vanishing moment with adversarial auto-encoder for short-term wind speed forecasting in Xinjiang wind farm, China

Wind speed forecasting can enhance the safety and economy of wind energy integration and conversion. The characteristic of the wind speed evolves over time. In this paper, a novel evolution-dependent multi-objective ensemble model of vanishing moment is proposed to solve the above problem. The proposed model can assign time-varying ensemble weights to base predictors with the different vanishing moments, so as to achieve better forecasting performance. In this model, the deep Adversarial Auto-Encoder (AAE) is firstly utilized to convert the wind speed into a two-dimension Gaussian distribution feature space. The Bat Algorithm (BA) is applied to partition the feature space into several sectors. Each sector represents a cluster of wind speed. Outlier Robust Extreme Learning Machines (ORELMs) improved by Maximal Overlap Discrete Wavelet Packet Transform (MODWPT) with the different vanishing moments are applied as base predictors, and the Multi-Objective Feasible Enhanced Particle Swarm Optimization (MOFEPSO) is used to obtain the best ensemble weights for each cluster. At last, the suitable ensemble model is applied for forecasting at each time. Four actual wind speed series collected from Xinjiang wind farm are used to verify the effectiveness of the proposed model. The experimental results indicate the proposed model outperforms other benchmark models.

An Optimal Early Warning System for Currency Crises Under Model Uncertainty

This paper assesses several early warning (EWS) models of financial crises to propose a model that can predict the incidence of a currency crisis in developing countries. For this purpose, we employ the equal weighting (EW) and dynamic model averaging (DMA) approaches to combine forecast from individual models allowing for time-varying weights. Taking Egypt as a case study and focusing only on currency crises, our findings show that combined forecast (EW- and DMA-based EWS) to account for uncertainty perform better than other competing models in both in-sample and out-of-sample forecasts.

Forecasting Realized Volatility in Financial Markets Based on a Time-Varying Non-Parametric Model

This paper proposes a new non-parametric adaptive combination model for the prediction of realized volatility on the basis of applying and extending the time-varying probability density function theory. We initially construct an adaptive time-varying weight mechanism for a combination forecast. To compare the predictive power of the models, we take the SPA test, which uses bootstrap as the evaluation criterion and employs the rolling window strategy for out-of-sample forecasting. The empirical study shows that the non-parametric TVF model forecasts more accurately than the HAR-RV model. In addition, the average combination forecast model does not have a significant advantage over any single model while our adaptive combination model does.

Multi-Target State Estimation Using Interactive Kalman Filter for Multi-Vehicle Tracking

In this paper, an interactive Kalman filter (IKF) is proposed to demonstrate the interaction between targets as to how the behavior of a desired target is affected by the behavior of its neighbors. The IKF utilizes two types of interactions available in multi-agent systems, namely, cooperative and competitive. The IKF is similar to the distributed Kalman filter (DKF) in terms of architecture, method of representation of equations, and use of neighborhood weight matrix while IKF appears to be a general form of DKF. In this method, a network of IKF nodes is constructed such that each node is associated with every target. There are edges between nodes for which the corresponding targets have effect on each other. Time-varying weights are used to control the interaction information exchanged among IKF nodes. The method of calculating interaction weights in the weight matrix plays a key role on the estimation results. The calculation of optimal IKF gain and evaluations on MOTP, MOTA, and MSE metrics illustrate the effectiveness of the proposed filter in vehicle tracking.

Unified inference for nonlinear factor models from panels with fixed and large time span

We provide unifying inference theory for parametric nonlinear factor models based on a panel of noisy observations. The panel has a large cross-section and a time span that may be either small or large. Moreover, we incorporate an additional source of information, provided by noisy observations on some known functions of the factor realizations. The estimation is carried out via penalized least squares, i.e., by minimizing the L2 distance between observations from the panel and their model-implied counterparts, augmented by a penalty for the deviation of the extracted factors from the noisy signals of them. When the time dimension is fixed, the limit distribution of the parameter vector is mixed Gaussian with conditional variance depending on the path of the factor realizations. On the other hand, when the time span is large, the convergence rate is faster and the limit distribution is Gaussian with a constant variance. In this case, however, we incur an incidental parameter problem since, at each point in time, we need to recover the concurrent factor realizations. This leads to an asymptotic bias that is absent in the setting with a fixed time span. In either scenario, the limit distribution of the estimates for the factor realizations is mixed Gaussian, but is related to the limiting distribution of the parameter vector only in the scenario with a fixed time horizon. Although the limit behavior is very different for the small versus large time span, we develop a feasible inference theory that applies, without modification, in either case. Hence, the user need not take a stand on the relative size of the time dimension of the panel. Similarly, we propose a time-varying data-driven weighting of the penalty in the objective function, which enhances efficiency by adapting to the relative quality of the signal for the factor realizations.

The Contribution of Jump Signs and Activity to Forecasting Stock Price Volatility

We propose a novel approach to decompose realized jump measures by type of activity (finite/infinite) and sign, and also provide noise-robust versions of the ABD jump test (Andersen et al., 2007b) and realized semivariance measures. We find that infinite (finite) jumps improve the forecasts at shorter (longer) horizons; but the contribution of signed jumps is limited. As expected, noise-robust measures deliver substantial forecast improvements at higher sampling frequencies, although standard volatility measures at the 300-second frequency generate the smallest MSPEs. Since no single model dominates across sampling frequency and forecasting horizon, we show that model-averaged volatility forecasts --using time-varying weights and models from the model confidence set-- generally outperform forecasts from both the benchmark and single best extended HAR model. Finally, forecasts using volatility and jump measures based on transaction sampling are inferior to the forecasts from clock-based sampling.

Forecasting Bond Risk Premia with Unspanned Macroeconomic Information

I provide evidence on the existence of unspanned macro risk. I investigate the usefulness of unspanned macro information for forecasting bond risk premia in a macro-finance term structure model from the perspective of a bond investor. I account for model uncertainty by combining forecasts with and without unspanned output and inflation risks optimally from the forecaster’s objective. Incorporating macro information generates significant gains in forecasting bond risk premia relative to yield curve information at long forecast horizons, especially when allowing for time-varying combination weight. These gains in predictive accuracy significantly improve investor utility.

Forecasting Bond Risk Premia with Unspanned Macroeconomic Information

I provide evidence on the existence of unspanned macro risk. I investigate the usefulness of unspanned macro information for forecasting bond risk premia in a macro-finance term structure model from the perspective of a bond investor. I account for model uncertainty by combining forecasts with and without unspanned output and inflation risks optimally from the forecaster’s objective. Incorporating macro information generates significant gains in forecasting bond risk premia relative to yield curve information at long forecast horizons, especially when allowing for time-varying combination weight. These gains in predictive accuracy significantly improve investor utility.

Multi-objective optimization strategy of adaptive cruise control considering regenerative energy

A multi-objective optimization strategy considering regenerative braking was proposed. Taking into account the impact of relative velocity, driver style, and road adhesion conditions, a variable time headway strategy was first proposed. Then, a multi-objective optimization strategy in adaptive cruise system was designed under the model predictive control framework. The incremental adaptive model predictive control with time-varying weights was constructed to be used as the upper controller, and slack variable was used to process the constraints. The constraints of regenerative braking were analyzed, and a new brake force distribution strategy based on multi-source information fusion was proposed to further optimize the economy. On the AMESim & Simulink co-simulation platform, a battery electric vehicle model was built and the proposed strategy was simulated. The results showed that, comparing to the constant-weight strategy, the proposed strategy had better robustness, which could rapidly and timely adjust the control target and guarantee the safety, comfort, economy, and following. The multi-source information braking force distribution strategy can guarantee several goals of the system while improving the economy. It can regenerate more braking energy, and the braking regenerative energy contribution increased by 5.68%.

Effect of Weight Change on Economic Outcomes Among Persons with Type 2 Diabetes Mellitus in the United States: Beyond Glycemic Control.

Previous studies report weight loss to be associated with significantly lower total health care costs among patients with type 2 diabetes mellitus (T2DM). The effect of weight change on health care costs, independent of glycemic control and after controlling for time-varying covariates among T2DM patients, remains unknown. To evaluate the effect of weight change, independent of glycemic control, on all-cause and T2DM-related health care resource utilization (HCRU) and costs among T2DM patients in the United States. A retrospective cohort study was conducted using a linked data extract composed of IQVIA's RWI Data Adjudicated Claims-US and Ambulatory Electronic Medical Record data. Adults (aged ≥ 18 years) with T2DM receiving ≥ 1 oral antidiabetic drug (OAD) medication, glucagon-like peptide-1 receptor agonist (GLP-1RA), and/or short- or long-acting insulin between January 1, 2010, and December 31, 2014 were included (the date of the first observed medical claim with a diagnosis code or medication prescription claim was the index date). Baseline characteristics were evaluated in the 6-month pre-index period. Weight loss (3%, 5%, or 7% from baseline) was evaluated over two 6-month periods (months 1-6 and 7-12) following the index date. Covariates included time-varying weight, hemoglobin A1c (A1c), costs, and HCRU within each 6-month period. Outcomes of interest (all-cause and T2DM-related HCRU and costs) were evaluated in the 6-month (months 13-18) and 12-month (months 13-24) periods following the initial 1- to 6-month and 7- to 12-month post-index periods. Structural nested mean models were used to evaluate the effect of weight change on these outcomes, independent of glycemic control. 1,407 patients were included (mean age = 55 years; 55% male), with a mean baseline weight of 102.2 kg (median = 99.7 kg) and a mean baseline A1c of 7.4% (median = 6.9%). In adjusted analysis, weight loss was associated with significantly lower all-cause and T2DM-related annual total health care costs. Compared with those showing no weight change, a 3%, 5%, and 7% weight loss resulted in approximately $500, $800, and $1,100 in savings, respectively, in all-cause annual total health care costs per patient in the year following the weight loss. Similarly, compared with those with no weight change, a 3%, 5%, and 7% weight loss resulted in approximately $200, $300, and $400 in savings, respectively, in T2DM-related annual total health care costs per patient in the following year. Even greater savings (up to ~$2,000 and ~$800 in all-cause and T2DM-related annual costs per patient, respectively) were experienced by those who lost weight compared with those who gained weight. After accounting for glycemic control, this study found that weight loss was associated with additional significant reductions in all-cause and T2DM-related annual total health care costs. Understanding the role of weight loss in T2DM may provide useful evidence for decision makers as they evaluate therapy options for T2DM. This study was funded by Novo Nordisk. Dang-Tan, Smolarz, and Iyer are employees of Novo Nordisk. Karkare and DeKoven (employees of IQVIA) and Fridman (employed by AMF Consulting) were contracted by Novo Nordisk to conduct this study. Fridman also reports personal fees from Shire, GSK, and CSL Behring, outside of the submitted work. Lu, an employee of IQVIA, accessed the database and conducted the statistical analysis for this study.

The KOF Globalisation Index \u2013 revisited

We introduce the revised version of the KOF Globalisation Index, a composite index measuring globalization for every country in the world along the economic, social and political dimension. The original index was introduced by Dreher (Applied Economics, 38(10):1091–1110, 2006) and updated in Dreher et al. (2008). This second revision of the index distinguishes between de facto and de jure measures along the different dimensions of globalization. We also disentangle trade and financial globalization within the economic dimension of globalization and use time-varying weighting of the variables. The new index is based on 43 instead of 23 variables in the previous version. Following Dreher (Applied Economics, 38(10):1091–1110, 2006), we use the new index to examine the effect of globalization on economic growth. The results suggest that de facto and de jure globalization influence economic growth differently. Future research should use the new KOF Globalisation Index to re-examine other important consequences of globalization and why globalization was proceeding rapidly in some countries, such as South Korea, but less so in others. The KOF Globalisation Index can be downloaded from http://www.kof.ethz.ch/globalisation/.

Commodity Terms of Trade

This paper presents a comprehensive database of country-specific commodity price indices for 182 economies covering the period 1962-2018. For each country, the change in the international price of up to 45 individual commodities is weighted using commodity-level trade data. The database includes a commodity terms-of-trade index—which proxies the windfall gains and losses of income associated with changes in world prices—as well as additional country-specific series, including commodity export and import price indices. We provide indices that are constructed using, alternatively, fixed weights (based on average trade flows over several decades) and time-varying weights (which can account for time variation in the mix of commodities traded and the overall importance of commodities in economic activity). The paper also discusses the dynamics of commodity terms of trade across country groups and their influence on key macroeconomic aggregates.

Social learning with multiple true states

In order to investigate social learning with multiple true states, a social learning model with time-varying topology and reliance weight is proposed. In this model, a time-varying topology mechanism for social networks is constructed since people always tend to communicate with those who have similar opinions with them. Simultaneously, the adaptive time-varying reliance weight mechanism is designed according to the closeness degree of agents’ neighbors. The simulation results show that asymptotic learning can be achieved and communities emerge under certain parameter values. Finally, how the parameters influence the belief evolution is analyzed, and a first order phase transition phenomenon is discovered.

Visual and Vestibular Inputs Affect Muscle Synergies Responsible for Body Extension and Stabilization in Sit-to-Stand Motion.

The sit-to-stand motion is a common movement in daily life and understanding the mechanism of the sit-to-stand motion is important. Our previous study shows that four muscle synergies can characterize the sit-to-stand motion, and they have specific roles, such as upper body flexion, rising from a chair, body extension, and posture stabilization. The time-varying weight of these synergies are changed to achieve adaptive movement. However, the relationship between sensory input and the activation of the muscle synergies is not completely understood. In this paper, we aim to clarify how vestibular and visual inputs affect the muscle synergy in sit-to-stand motion. To address this, we conducted experiments as follows. Muscle activity, body kinematics, and ground reaction force were measured for the sit-to-stand motion under three different conditions: control, visual-disturbance, and vestibular-disturbance conditions. Under the control condition, the participants stood without any intervention. Under the visual-disturbance condition, the participants wore convex lens glasses and performed the sit-to-stand motion in a dark room. Under the vestibular-disturbance condition, a caloric test was performed. Muscle synergies were calculated for these three conditions using non-negative matrix factorization. We examined whether the same four muscle synergies were employed under each condition, and the changes in the time-varying coefficients were determined. These experiments were conducted on seven healthy, young participants. It was found that four muscle synergies could explain the muscle activity in the sit-to-stand motion under the three conditions. However, there were significant differences in the time-varying weight coefficients. When the visual input was disturbed, a larger amplitude was found for the muscle synergy that activated mostly in the final posture stabilization phase of the sit-to-stand motion. Under vestibular-disturbance condition, a longer activation was observed for the synergies that extended the entire body and led to posture stabilization. The results implied that during human sit-to-stand motion, visual input has less contribution to alter or correct activation of muscle synergies until the last phase. On the other hand, duration of muscle synergies after the buttocks leave are prolonged in order to adapt to the unstable condition in which sense of verticality is decreased under vestibular-disturbance.

Commodity Terms of Trade: A New Database

This paper presents a comprehensive database of country-specific commodity price indices for 182 economies covering the period 1962-2018. For each country, the change in the international price of up to 45 individual commodities is weighted using commodity-level trade data. The database includes a commodity terms-of-trade index-which proxies the windfall gains and losses of income associated with changes in world prices-as well as additional country-specific series, including commodity export and import price indices. We provide indices that are constructed using, alternatively, fixed weights (based on average trade flows over several decades) and time-varying weights (which can account for time variation in the mix of commodities traded and the overall importance of commodities in economic activity). The paper also discusses the dynamics of commodity terms of trade across country groups and their influence on key macroeconomic aggregates.

A Novel Hybrid Model Based on TVIW-PSO-GSA Algorithm and Support Vector Machine for Classification Problems

The increasingly serious haze problem in China has brought about a growing public awareness of air quality. Precise air quality index (AQI) forecasts play an important role in both controlling air pollution and promoting the sustainable development of human society. However, the randomness, non-stationarity, and irregularity of the AQI series make its classifications very difficult. This paper introduces a time-varying inertia weighting (TVIW) strategy based on a combination of gravitation search algorithm (GSA) and particle swarm optimization (PSO) called the TVIW-PSO-GSA. The TVIW-PSO-GSA is utilized to optimize the penalty parameter C and kernel function parameter γ of a support vector machine (SVM) to create a hybrid TVIW-PSO-GSA-SVM algorithm. Twenty-three benchmark functions, five UCI datasets, and an AQI hierarchical classification example are tested to find that the convergence speed and performance of the TVI-PSO-GSA exceed those of other algorithms, and the TVIW-PSO-GSA-SVM algorithm also achieves higher classification accuracy and efficiency than the PSO-GSA-SVM, GSA-SVM, GA-SVM, or PSO-SVM, which indicates that the TVIW-PSO-GSA-SVM reliably and accurately classifies AQI and UCI datasets.