A convex version of multivariate adaptive regression splines

Multivariate adaptive regression splines (MARS) provide a flexible statistical modeling method that employs forward and backward search algorithms to identify the combination of basis functions that best fits the data and simultaneously conduct variable selection. In optimization, MARS has been used successfully to estimate the unknown functions in stochastic dynamic programming (SDP), stochastic programming, and a Markov decision process, and MARS could be potentially useful in many real world optimization problems where objective (or other) functions need to be estimated from data, such as in surrogate optimization. Many optimization methods depend on convexity, but a non-convex MARS approximation is inherently possible because interaction terms are products of univariate terms. In this paper a convex MARS modeling algorithm is described. In order to ensure MARS convexity, two major modifications are made: (1) coefficients are constrained, such that pairs of basis functions are guaranteed to jointly form convex functions and (2) the form of interaction terms is altered to eliminate the inherent non-convexity. Finally, MARS convexity can be achieved by the fact that the sum of convex functions is convex. Convex-MARS is applied to inventory forecasting SDP problems with four and nine dimensions and to an air quality ground-level ozone problem.     

Phase Coded Waveform Design for Sonar Sensor Network

Since it is known that interference with each sonar sensor could be effectively reduced when waveforms are appropriately designed for a Sonar Sensor Network (SSN), we provide a set of new ternary codes called optimized punctured Zero Correlation Zone sequence-pair set (ZCZPS) and provide a method to construct such codes. We study the codes’ properties especially using the ambiguity function to analyze the nature of the output of the matched filter. We apply our provided ternary codes to the SSN as pulse compression codes for narrowband pulse signals and simulate the target detection performance of the system. Comparing with the classical periodic Gold sequences, our codes could improve the system detection performance.     

Temperature dependence of the photoluminescence energy of CdTe/ZnTe heterostructures

The photoluminescence (PL) energy was calculated for the heterostructure CdTe/ZnTe, where the emission is based on quantum confinement. This system was studied because it yields a high-intensity emission, covering a wide range of the visible spectrum wavelengths. Our calculations obtained the energy of the fundamental state for electrons and holes, as well as the PL energy using both the Varshni and Manoogian–Woolley formalisms taking into consideration the effects of strain. The Varshni and M–W formalisms for the study of the band gap energy as a function of the temperature differ for high- and low-temperature values. This is because the M–W model takes into account the thermal dilatation and electron–phonon interaction through the U and V parameters. The calculations used quantum wells based on the CdTe/ZnTe heterostructure with different thicknesses (number of monolayers) and involved analyzing the dependence on the thickness and barrier height. The aims of this study were analyzing how the shift occurs in the emission color while increasing of the quantum-well width, and studying the temperature dependence of two different formalisms. The PL energy as a function of the quantum well thickness presents an inflection point at around five monolayers, which is in agreement with reports regarding the critical thickness of heterostructure CdTe/ZnTe. This inflection point is modified when the strain in the heterostructure is taken into account.     

Embedded planar surface segmentation system for stereo images

An embedded system is developed to segment stereo images using disparity. The recent developments in the embedded system architecture have allowed real time implementation of low-level vision tasks such as stereo disparity computation. At the same time, an intermediate level task such as segmentation is rarely attempted in an embedded system. To solve the planar surface segmentation problem, which is iterative in nature, our system implements a Segmentation–Estimation framework. In the segmentation phase, segmentation labels are assigned based on the underlying plane parameters. Connected component analysis is carried out on the segmentation result to select the largest spatially connected area for each plane. From the largest areas, the parameters for each plane are reestimated. This iterative process was implemented on TMS320DM642 based embedded system that operates at 3–5 frames per second on images of size 320 × 240.     

Butyrate Feeding Reverses CypD-Related Mitoflash Phenotypes in Mouse Myofibers

Mitoflashes are spontaneous transients of the biosensor mt-cpYFP. In cardiomyocytes, mitoflashes are associated with the cyclophilin D (CypD) mediated opening of mitochondrial permeability transition pore (mPTP), while in skeletal muscle they are considered hallmarks of mitochondrial respiration burst under physiological conditions. Here, we evaluated the potential association between mitoflashes and the mPTP opening at different CypD levels and phosphorylation status by generating three CypD derived fusion constructs with a red shifted, pH stable Ca²⁺ sensor jRCaMP1b. We observed perinuclear mitochondrial Ca²⁺ efflux accompanying mitoflashes in CypD and CypDS42A (a phosphor-resistant mutation at Serine 42) overexpressed myofibers but not the control myofibers expressing the mitochondria-targeting sequence of CypD (CypDN30). Assisted by a newly developed analysis program, we identified shorter, more frequent mitoflash activities occurring over larger areas in CypD and CypDS42A overexpressed myofibers than the control CypDN30 myofibers. These observations provide an association between the elevated CypD expression and increased mitoflash activities in hindlimb muscles in an amyotrophic lateral sclerosis (ALS) mouse model previously observed. More importantly, feeding the mice with sodium butyrate reversed the CypD-associated mitoflash phenotypes and protected against ectopic upregulation of CypD, unveiling a novel molecular mechanism underlying butyrate mediated alleviation of ALS progression in the mouse model.     

Distributed and Energy Efficient Self-Organization for On–Off Wireless Sensor Networks

In this paper, we utilize clustering to achieve energy efficiency for the on–off wireless sensor network, whose member nodes alternate between active and inactive states. In the proposed Distributed and Energy Efficient Self Organization (DEESO) scheme, the head election is adjusted adaptively to the remaining battery levels of local active nodes, which is a completely distributed approach compared to LEACH that relying on other routing schemes to access global information. Furthermore, we apply the Adaptive Channel Assignment (ACA) to address the on-off topology changes. Simulation results show that DEESO delivers 184% amount of data to the base station as LEACH for the same amount of energy consumption and the effective network lifetime is extended by around 50%.     

Information Management Challenges and the Adverse Consequences of Using Reverse Auctions

Using reverse auctions (RAs) requires managing complex information. Yet, such challenges are not well understood and are often even underestimated. To address this knowledge gap, I conducted a Delphi study and follow-up interviews to identify information management challenges and the adverse consequences. Drawing on the agency theory and the garbage can theory, I developed three dimensions of information management challenges, advanced a theoretical model, and formulated propositions to suggest the adverse consequences. My findings suggest that deficiency, violation, and anarchy of information can result in procedural failure, financial and competitive loss to suppliers, and damage to buyer–supplier relationships.     

Modular discriminant analysis and its applications

We propose a multi-linear algebra based subspace learning approach for finding linear projection which preserves some implicit structural or locally-spatial information among the original feature space. Our method uses a new tensor data representation model, in which, each group of data points are partitioned into several equal-sized sub-groups with its neighbors affiliated to them, and all sub-groups are concatenated to represent as the tensor space product of the original feature space. Then, a new optimization algorithm called Lagrangian multiplier mode (L-mode) is presented for computing the optimal linear projections. We show that our method has three ways for resolving the Small Sample Size problem: by applying the fuzzy matrix model to avoid the disturbance from non-interested determinant, by a quadratic sample correlation model, and by projecting the samples into a manifold using linear programming. Extensive experimental results conducted on two benchmark face biometrics datasets i.e. Yale-B and CMU-PIE, and a nutrition surveillance dataset demonstrate that our method is effective and robust than the state-of-the-arts such as Principal Component Analysis, Linear Discriminant Analysis, Locality Preserving Projections and their variations on both classification accuracies and computational expenses.