Building Support Vector Machines in the Context of Regularised Least Squares

This paper formulates a linear kernel support vector machine (SVM) as a regularized least-squares (RLS) problem. By defining a set of indicator variables of the errors, the solution to the RLS problem is represented as an equation that relates the error vector to the indicator variables. Through partitioning the training set, the SVM weights and bias are expressed analytically using the support vectors. It is also shown how this approach naturally extends to Sums with nonlinear kernels whilst avoiding the need to make use of Lagrange multipliers and duality theory. A fast iterative solution algorithm based on Cholesky decomposition with permutation of the support vectors is suggested as a solution method. The properties of our SVM formulation are analyzed and compared with standard SVMs using a simple example that can be illustrated graphically. The correctness and behavior of our solution (merely derived in the primal context of RLS) is demonstrated using a set of public benchmarking problems for both linear and nonlinear SVMs.

Dynamics of an axisymmetric electromagnetic 'crucible' melting

Different industrial induction melting processes involve free surface and melt-solid interface of the liquid metal subject to dynamic change during the technological operation. Simulation of the liquid metal dynamics requires to solve the non-linear, coupled hydrodynamic-electromagnetic-heat transfer problem accounting for the time development of the liquid metal free boundary with a suitable turbulent viscosity model. The present paper describes a numerical solution method applicable for various axisymmetric induction melting processes, such as, crucible with free top surface, levitation, semi-levitation, cold crucible and similar melting techniques. The presented results in the cases of semi-levitation and crucible with free top surface meltings demonstrate oscillating transient behaviour of the free metal surface indicating the presence of gravity-inertial-electromagnetic waves which are coupled to the internal fluid flow generated by both the rotational and potential parts of the electromagnetic force.

JOINT ENERGY BEAMFORMING AND TIME ASSIGNMENT FOR COOPERATIVE WIRELESS POWERED NETWORKS

Wireless power transfer (WPT) and radio frequency (RF)-based energy har- vesting arouses a new wireless network paradigm termed as wireless powered com- munication network (WPCN), where some energy-constrained nodes are enabled to harvest energy from the RF signals transferred by other energy-sufficient nodes to support the communication operations in the network, which brings a promising approach for future energy-constrained wireless network design. In this paper, we focus on the optimal WPCN design. We consider a net- work composed of two communication groups, where the first group has sufficient power supply but no available bandwidth, and the second group has licensed band- width but very limited power to perform required information transmission. For such a system, we introduce the power and bandwidth cooperation between the two groups so that both group can accomplish their expected information delivering tasks. Multiple antennas are employed at the hybrid access point (H-AP) to en- hance both energy and information transfer efficiency and the cooperative relaying is employed to help the power-limited group to enhance its information transmission throughput. Compared with existing works, cooperative relaying, time assignment, power allocation, and energy beamforming are jointly designed in a single system. Firstly, we propose a cooperative transmission protocol for the considered system, where group 1 transmits some power to group 2 to help group 2 with information transmission and then group 2 gives some bandwidth to group 1 in return. Sec- ondly, to explore the information transmission performance limit of the system, we formulate two optimization problems to maximize the system weighted sum rate by jointly optimizing the time assignment, power allocation, and energy beamforming under two different power constraints, i.e., the fixed power constraint and the aver- age power constraint, respectively. In order to make the cooperation between the two groups meaningful and guarantee the quality of service (QoS) requirements of both groups, the minimal required data rates of the two groups are considered as constraints for the optimal system design. As both problems are non-convex and have no known solutions, we solve it by using proper variable substitutions and the semi-definite relaxation (SDR). We theoretically prove that our proposed solution method can guarantee to find the global optimal solution. Thirdly, consider that the WPCN has promising application potentials in future energy-constrained net- works, e.g., wireless sensor network (WSN), wireless body area network (WBAN) and Internet of Things (IoT), where the power consumption is very critical. We investigate the minimal power consumption optimal design for the considered co- operation WPCN. For this, we formulate an optimization problem to minimize the total consumed power by jointly optimizing the time assignment, power allocation, and energy beamforming under required data rate constraints. As the problem is also non-convex and has no known solutions, we solve it by using some variable substitutions and the SDR method. We also theoretically prove that our proposed solution method for the minimal power consumption design guarantees the global optimal solution. Extensive experimental results are provided to discuss the system performance behaviors, which provide some useful insights for future WPCN design. It shows that the average power constrained system achieves higher weighted sum rate than the fixed power constrained system. Besides, it also shows that in such a WPCN, relay should be placed closer to the multi-antenna H-AP to achieve higher weighted sum rate and consume lower total power.

An investigation of phosphate adsorption from aqueous solution onto hydrous niobium oxide prepared by co-precipitation method

The synthetic hydrous niobium oxide has been used for phosphate removal from the aqueous solutions. The kinetic data correspond very well to the pseudo second-order equation The phosphate removal tended. to increase with a decrease of pH. The equilibrium data describe very well the Langmuir isotherm. The peak appearing at 1050 cm(-1) in IR spectra after adsorption was attributed to the bending vibration of adsorbed phosphate. The adsorption capacities are high, and increased with increasing temperature. The evaluated Delta G degrees and Delta H degrees indicate the spontaneous and endothermic nature of the reactions. The adsorptions occur with increase in entropy (Delta S positive) value suggest increase in randomness at the solid-liquid interface during the adsorption. A phosphate desorbability of approximately 60% was observed with water at pH 12, which indicated a relatively strong bonding between the adsorbed phosphate and the sorptive sites on the surface of the adsorbent. (C) 2008 Elsevier B.V. All rights reserved.

Solution of a heterogeneous modeling of a horizontal-flow anaerobic immobilized biomass (HAIB) reactor by the sequencing method

A modeling study was completed to develop a methodology that combines the sequencing and finite difference methods for the simulation of a heterogeneous model of a tubular reactor applied in the treatment of wastewater. The system included a liquid phase (convection diffusion transport) and a solid phase (diffusion reaction) that was obtained by completing a mass balance in the reactor and in the particle, respectively. The model was solved using a pilot-scale horizontal-flow anaerobic immobilized biomass (HAIB) reactor to treat domestic sewage, with the concentration results compared with the experimental data. A comparison of the behavior of the liquid phase concentration profile and the experimental results indicated that both the numerical methods offer a good description of the behavior of the concentration along the reactor. The advantage of the sequencing method over the finite difference method is that it is easier to apply and requires less computational time to model the dynamic simulation of outlet response of HAIB.

Seedling responses of four Australian tree species to toxic concentrations of manganese in solution culture

Little is known about the responses of Australian plants to excess metal, including Mn. It is important to remedy this lack of information so that knowledgeable decisions can be made about managing Mn contaminated sites where inhabited by Australian vegetation. Acacia holosericea, Melaleuca leucadendra, Eucalyptus crebra and Eucalyptus camaldulensis were grown in dilute solution culture for 10 weeks. The seedlings ( 42 days old) were exposed to six Mn treatments viz., 1, 8, 32, 128, 512 and 2048 muM. The order of tolerance to toxic concentrations of Mn was A. holosericea congruent to = E. crebra < M. leucadendra < E. camaldulensis, the critical external concentrations being approximately 5.1, 5.0, 21 and 330 muM, respectively. The critical tissue Mn concentrations for the youngest fully expanded leaf and total shoots were, respectively, 265 and 215 mug g(-1) DM for A. holosericea, 445 and 495 mug g(-1) DM for M. leucadendra, 495 and 710 mug g(-1) DM for E. crebra and 7230 and 6510 mug g(-1) DM for E. camaldulensis. The high tolerance of E. camaldulensis ( as opposed to the sensitivity of E. crebra) to excess Mn raises concern about fauna feeding on the plant and is consistent with hypotheses suggesting the Eucalyptus subgenus Symphomyrtus is particularly tolerant of stress, including excess Mn. The results from this paper provide the first comprehensive combination of growth responses, critical external concentrations, critical tissue concentrations and plant toxicity symptoms for three important Australian genera, viz., Eucalyptus, Acacia and Melaleuca, for use in the management of Mn toxic sites.