Cost-sensitive learning of top-down modulation for attentional control

A biologically-inspired model of visual attention known as basic saliency model is biased for object detection. It is possible to make this model faster by inhibiting computation of features or scales, which are less important for detection of an object. To this end, we revise this model by implementing a new scale-wise surround inhibition. Each feature channel and scale is associated with a weight and a processing cost. Then a global optimization algorithm is used to find a weight vector with maximum detection rate and minimum processing cost. This allows achieving maximum object detection rate for real time tasks when maximum processing time is limited. A heuristic is also proposed for learning top-down spatial attention control to further limit the saliency computation. Comparing over five objects, our approach has 85.4 and 92.2% average detection rates with and without cost, respectively, which are above 80% of the basic saliency model. Our approach has 33.3 average processing cost compared with 52 processing cost of the basic model. We achieved lower average hit numbers compared with NVT but slightly higher than VOCUS attentional systems.     

Design of proportional-derivative-type state feedback controllers for congestion control of transmission control protocol networks

A new proportional-derivative-type state feedback controller is proposed for congestion control of transmission control protocol (TCP) networks. An analytical TCP model is adopted. In the proposed control scheme, it is possible to efficiently control the TCP traffic using only the queue length at the router without the need to know the TCP window size which is not available locally. The results are presented in terms of delay-dependent linear matrix inequality. The proposed method is verified by simulation examples using NS software, and the effectiveness and superiority of our method over other control schemes, such as the proportional-integral, random early detection and generalised minimum variancemethods, are also shown.     

Joint replenishment policy with backordering and special sale

In this paper, an inventory control model with a joint replenishment policy and a temporary discount is developed. We assume that shortage is allowed and buyer uses an economic order quantity inventory control model. Different cases based on ordering policies for the first joint replenishment if the special order is not taken, and coincidence of a special period length with a positive or negative inventory level of the last regular period length, are investigated. Furthermore, several theorems are proved through which closed-form solutions are obtained. At the end, two numerical examples illustrate the different situations that the buyer may face and sensitivity analyses for both examples are reported.     

Permanent magnet configuration in design of an eddy current damper

This paper discusses the design of an eddy current passive damper using different configurations of permanent magnets. Motional eddy current damping effect is used for the development of a passive damper. Eddy currents are generated in a conductor in a time-varying magnetic field. They are induced either by movement of the conductor in a static field or by changing the strength of the magnetic field, initiating motional and transformer electromotive forces, respectively. The conceived eddy current damper consists of a conductor as an outer tube, and an array of axially magnetized, ring-shaped permanent magnets (PMs), separated by iron pole pieces as a mover. The relative movement of the magnets and the conductor causes the conductor to undergo motional eddy currents. Using this concept, damping characteristics of the new damper is obtained through analytical modeling, and verified by experimental analysis. The optimum PMs’ size and configuration are also derived using analytical and finite element analysis, respectively. A damping coefficient as high as 53 kg/s is achievable with the proposed design specifications.     

Uncertainty handling in convolutional neural networks

Neural Computing and Applications - The performance of convolutional neural networks is degraded by noisy data, especially in the test phase. To address this challenge, a new convolutional neural...     

Synthesis and Stability of New Nitrophenol Complexes of <Superscript>99<Emphasis Type="Italic">m</Emphasis></Superscript>Tc as Possible Hypoxia Imaging Radiopharmaceuticals

Seven new nitrophenol derivatives were synthesized; their ^99mTc complexes (N₂OS chelates), which are possible hypoxia tumor imaging agents, were prepared by the reaction with [^99mTc]NaTcO₄ in the presence of SnCl₂ at pH 10. The in vitro biostabilty of the complexes was evaluated. The purity and stability (in human and rat serum) of the complexes were evaluated by chromatographic methods (TLC, HPLC). The most stable (for more than 6 h) is the complex of ^99mTc with 3-[3′-N-(2″-hydroxy-5″-nitrobenzylamino)-2′-hydroxypropyl]-1-methylthiourea.     

A framework for diagnosing cervical cancer disease based on feedforward MLP neural network and ThinPrep histopathological cell image features

In this paper, Levenberg–Marquardt feedforward MLP neural network (LMFFNN) was proposed to classify cervical cell images obtained from 100 patients including healthy, low-grade intraepithelial squamous lesion and high-grade intraepithelial squamous lesion cases. This neural network along with extracted cell image features is a new model for cervical cell image classification. The semiautomated cervical cancer diagnosis system is composed of two phases: image preprocessing/processing and feedforward MLP neural network. In the first stage, image preprocessing is done to reduce the existing noises without lowering the resolution. After that, image processing algorithms were applied to manually cropped cell images to achieve a linear plot which includes real components, were used as LMFFNN inputs for classification of cervical cell images. Based on the results, cervical cell images were classified successfully with 100 % correct classification rate using the proposed method. Moreover, the rates of sensitivity and specificity were calculated as 100 % using LMFFNN method. It was shown there was a good agreement between the expert decision and values gained from the ANN model.     

Nonlinear secondary resonance of FG porous silicon nanobeams under periodic hard excitations based on surface elasticity theory

To impart desirable material properties, functionally graded (FG) porous silicon has been produced in which the porosity changes gradually across the material volume. The prime objective of this work is to predict the influence of the surface free energy on the nonlinear secondary resonance of FG porous silicon nanobeams under external hard excitations. On the basis of the closed-cell Gaussian-random field scheme, the mechanical properties of the FG porous material are achieved corresponding to the uniform and three different FG patterns of porosity dispersion. The Gurtin–Murdoch theory of elasticity is implemented into the classical beam theory to construct a surface elastic beam model. Thereafter, with the aid of the method of multiple time-scales together with the Galerkin technique, the size-dependent nonlinear differential equations of motion are solved corresponding to various immovable boundary conditions and porosity dispersion patterns. The frequency response and amplitude response associated with the both subharmonic and superharmonic hard excitations are obtained including multiple vibration modes and interactions between them. It is found that for the subharmonic excitation, the nanobeam is excited within a specific range of the excitation amplitude, and this range shifts to higher excitation amplitude by incorporating the surface free energy effects. For the superharmonic excitation, by taking surface stress effect into account, the excitation amplitude associated with the peak of the vibration amplitude enhances. Moreover, in the subharmonic case, it is demonstrated that by increasing the porosity coefficient, the value of the excitation frequency at the joint point of the two branches of the frequency-response curve reduces. In the superharmonic case, it is revealed that an increment in the value of porosity coefficient leads to decrease the peak of the oscillation amplitude and the associated excitation frequency.

     

A swift technique for damage detection of determinate truss structures

In this study, a novel two-stage approach for damage detection of determinate truss structures is proposed. The method lies in the group of vibration-based methods but it just needs the first natural frequency and mode shape vector of these structures for identifying the location and severity of damage. In the first stage, the modal residual force vector for different modes of a structure is introduced and the one associated with the first mode is applied to the structure as an external nodal force vector. Then, the residual local nodal force vector can be computed for all elements of the structure. Next, the elements with non-zero residual internal force are considered as damaged elements. In the second stage, the damage severity of each damaged element is determined using a new relation which can be categorized as a force–displacement relation. To show the efficiency and simplicity of the proposed method, three truss structures including a 13-bar planar truss, a 29-bar planar truss, and a 77-bar planar truss under different damage scenarios are studied; the results of which indicate that the method is innovatively capable of suitably detecting, for determinate truss structures, not only damaged members but also their individual damage severity by carrying solely one analysis.