A Team of Continuous-Action Learning Automata for Noise-Tolerant Learning of Half-Spaces

Learning automata are adaptive decision making devices that are found useful in a variety of machine learning and pattern recognition applications. Although most learning automata methods deal with the case of finitely many actions for the automaton, there are also models of continuous-action-set learning automata (CALA). A team of such CALA can be useful in stochastic optimization problems where one has access only to noise-corrupted values of the objective function. In this paper, we present a novel formulation for noise-tolerant learning of linear classifiers using a CALA team. We consider the general case of nonuniform noise, where the probability that the class label of an example is wrong may be a function of the feature vector of the example. The objective is to learn the underlying separating hyperplane given only such noisy examples. We present an algorithm employing a team of CALA and prove, under some conditions on the class conditional densities, that the algorithm achieves noise-tolerant learning as long as the probability of wrong label for any example is less than 0.5. We also present some empirical results to illustrate the effectiveness of the algorithm.

Kinetics of Pressureless Infiltration of Al-Mg Melts into Porous Alumina Preforms

Effective ���hydrodynamic��� radii governing infiltration kinetics of reactive Al-Mg melts into alumina preforms were found to be three orders of magnitude smaller than the average pore size of the packed bed and also smaller compared with the kinetics for a nonreactive system. A sinusoidal capillary model was developed to predict flow kinetics within the packed bed. For the reactive system, two factors were ascribed for additional melt retardation: (1) different intrinsic wettabilities of the two liquids on alumina, thereby leading to significantly different ���effective��� local contact angles; and (2) local solute depletion from the meniscus, which was incorporated as a time-dependent contact angle.

Temperature deformation behaviour of a Mg-0.8Al alloy

There is considerable interest currently in developing magnesium based alloys as replacements for aluminum alloys in automobile applications, due to their high specific strength as compared to aluminum alloys. However, the poor formability of magnesium alloys has restricted their applications; superplasticity can be utilized to form components with complex shapes. In the present study, the compressive deformation characteristics of a Mg-0.8 wt% Al alloy with an initial grain size of 19 +/- 1.0 mum have been studied in the temperature range of 623-673 K and at strain rates ranging from 10(-7) to 10(-3) s(-1). The stress exponent was observed to decrease with a decrease in stress. The results are analyzed in terms of the existing theoretical models for high temperature deformation. Furthermore, the potential for superplasticity in this alloy is explored, based on the mechanical and microstructural characteristics of the alloy.

Using heat treatment effects and EBSD analysis to tailor microstructure of hybrid Mg nanocomposite for enhanced overall mechanical response

In this study, a detailed investigation on the effect of heat treatment on the microstructural characteristics, texture evolution and mechanical properties of Mg-(5.6Ti+2.5B(4)C)(BM) hybrid nanocomposite is presented. Optimised heat treatment parameters, namely, heat treatment temperature and heat treatment time, were first identified through grain size and microhardness measurements. Initially, heat treatment of composites was conducted at temperature range between 100 and 300 degrees C for 1 h. Based on optical microscopic analysis and microhardness measurements, it was evident that significant grain growth and reduction in microhardness occurred for temperatures > 200 degrees C. The cutoff temperature that caused significant grain growth/matrix softening was thus identified. Second, at constant temperature (200 degrees C), the effect of variation of heat treatment time was carried out (ranging between 1 and 5 h) so as to identify the range wherein increase in average grain size and reduction in microhardness occurred. Furthering the study, the effect of optimised heat treatment parameters (200 degrees C, 5 h) on the microstructural texture evolution and hence, on the tensile and compressive properties of the Mg-(5.6Ti+2.5B(4)C)(BM) hybrid nanocomposite was carried out. From electron backscattered diffraction (EBSD) analysis, it was identified that the optimised heat treatment resulted in recrystallisation and residual stress relaxation, as evident from the presence of similar to 87% strain free grains, when compared to that observed in the non-heat treated/as extruded condition (i.e. 2.2 times greater than in the as extruded condition). For the heat treated composite, under both tensile and compressive loads, a significant improvement in fracture strain values (similar to 60% increase) was observed when compared to that of the non-heat treated counterpart, with similar to 20% reduction in yield strength. Based on structure-property correlation, the change in mechanical characteristics is identified to be due to: (1) the presence of less stressed matrix/reinforcement interface due to the relief of residual stresses and (2) texture weakening due to matrix recrystallisation effects, both arising due to heat treatment.

Correlation of microstructure and creep behaviour of MRI230D Mg alloy developed by two different casting technologies

The relationship between the as-cast microstructure and creep behaviour of the heat-resistant MRI230D Mg alloy produced by two different casting technologies is investigated. The alloy in both ingot-casting (IC) and high pressure die-casting (HPDC) conditions consists of alpha-Mg, 06 ((Mg,AI)(2)Ca), Al-Mn and Sn-Mg-Ca rich phases. However, the HPDC alloy resulted in relatively finer grain size and higher volume fraction of finer, denser network of eutectic C36 phase in the as-cast microstructure as compared to that of the IC alloy. The superior creep resistance exhibited by the HPDC alloy at all the stress levels and temperatures employed in the present investigation was attributed to the more effective dispersion strengthening effect caused by the presence of finer and denser network of the C36 phase. The increased amount of the eutectic C36 phase was the only change observed in the microstructures of both alloys following creep tests. (C) 2015 Elsevier B.V. All rights reserved.

Effect of recrystallization and grain growth on the mechanical properties of an extruded AZ21 Mg alloy

An experimental investigation into the effect of microstructural changes, which occur during post-extrusion annealing of a Mg based AZ21 alloy, on tensile and fatigue properties is conducted. Mechanical properties in the as-cast, as-extruded, and microstructural states that correspond to recovery, recrystallization and grain growth stages of annealing are compared. Results show that these microstructural changes do not alter the yield strength of the alloy markedly whereas significant differences were noted in the ultimate tensile strength as well as ductility. The initiation of abnormal grain growth (or secondary recrystallization) renders the tensile stress-strain response elastic perfectly plastic and results in a large drop in ductility, as high as similar to 60% during intermediate stages of abnormal grain growth, vis-A-vis the ductility of the as-extruded alloy. While the fatigue performance of all the wrought alloys is far superior to as expected, abnormal grain growth leads to a marked decrease in the endurance that of the as-cast alloy, limit. Possible microscopic origins of these are discussed. (C) 2009 Elsevier B.V. All rights reserved.

Voltage delay during constant-current or constant-resistance discharge of Mg-MnO2 dry cells: A comparative study

A theoretical approach has been developed to relate the voltage delay transients of the Mg-MnO2 dry cell observed during discharge by two commonly employed modes, viz., (1) at constant current, and (2) across a constant resistance. The approach has been verified by comparison of experimentally obtained transients with those generated from theory. The method may be used to predict the delay parameters of the Mg-MnO2 dry cell under the two modes of discharge and can, in principle, be extended to lithium batteries.

Equivalent circuit parameters of the film-covered magnesium/electrolyte interface in Mg/MnO2 dry cells from transient and ac impedance measurements

The electrochemical properties of the film-covered anode/solution interface in the magnesium/ manganese dioxide dry cell have been evaluated. The most plausible electrical equivalent circuit description of the Mg/solution interface with the passive film intact, has been identified. These results are based on the analysis of ac impedance and voltage transient measurements made on the dry cell under conditions which cause no damage to the protective passive film on the anode. The study demonstrates the complementary character of impedance and transient measurements when widely different frequency ranges are sampled in each type of investigation. The values and temperature dependence of the anode-film resistance, film capacitance, double-layer capacitance and charge-transfer resistance of the film-covered magnesium/solution interface have been determined. The magnitude of these values and its implications in understanding the important performance aspects of the magnesium/manganese dioxide dry cell are discussed. The study may be extended, in principle, to Li, Al and Ca batteries.

Equivalent circuit parameters of the film-covered magnesium/electrolyte interface in Mg/MnO2 dry cells from transient and ac impedance measurements

The electrochemical properties of the film-covered anode/solution interface in the magnesium/ manganese dioxide dry cell have been evaluated. The most plausible electrical equivalent circuit description of the Mg/solution interface with the passive film intact, has been identified. These results are based on the analysis of ac impedance and voltage transient measurements made on the dry cell under conditions which cause no damage to the protective passive film on the anode. The study demonstrates the complementary character of impedance and transient measurements when widely different frequency ranges are sampled in each type of investigation. The values and temperature dependence of the anode-film resistance, film capacitance, double-layer capacitance and charge-transfer resistance of the film-covered magnesium/solution interface have been determined. The magnitude of these values and its implications in understanding the important performance aspects of the magnesium/manganese dioxide dry cell are discussed. The study may be extended, in principle, to Li, Al and Ca batteries.

Reversibility of Mg/Mg2+ couple in a gel polymer electrolyte

Investigations on solid state rechargeable magnesium batteries are considered important similar to lithium batteries. In view of negligible hazards and less reactivity of the magnesium, in comparison with lithium, studies on rechargeable magnesium batteries are expected to have a wide scope in future. Solid polymer electrolytes, which conduct Mg2+ ions and reversibility of a Mg/Mg2+ couple are essential components of the studies. In the present investigations, the existence of reversibility of a Mg/Mg2+ couple in a gel polymer electrolyte (GPE) medium is established for the first time in literature. Results obtained by electrochemical impedance spectroscopy and cyclic voltammetry on Mg/GPE/Mg, SS/GPE/SS symmetrical cells show evidence for the reversibility. (C) 1999 Elsevier Science Ltd. All rights reserved.

Achieving the D-MG and DMD Tradeoffs of MIMO fading channels

An overview of our recent results relating to the explicit construction of space-time block codes achieving the DMG tradeoff of the quasi-static fading channel is presented. The results include the explicit construction of D-MG optimal codes,generalization of perfect codes to any number of transmit antennas as well as optimal diversity-multiplexing-delay constructions for the MIMO ARQ Channel.

Effect of hybridizing micron-sized Ti with nano-sized SiC on the microstructural evolution and mechanical response of Mg-5.6Ti composite

In this work, the effect of hybridizing micro-Ti with nano-SiC particulates on the microstructural and the mechanical behaviour of Mg-5.6Ti composite were investigated. Mg materials containing micron-sized Ti particulates hybridized with different amounts of nano-size SiC particulates were synthesized using the disintegrated melt deposition method followed by hot extrusion. The microstructural and mechanical behaviour of the developed Mg hybrid composites were studied in comparison with Mg-5.6Ti. Microstructural characterization revealed grain refinement attributed to the presence of uniformly distributed micro-Ti particles embedded with nano-SiC particulates. Electron back scattered diffraction (EBSD) analyses of Mg-(5.6Ti + 1.0SiC)(BM) hybrid composite showed relatively more localized recrystallized grains and lesser tensile twin fraction, when compared to Mg-5.6Ti. The evaluation of mechanical properties indicated that the best combination of strength and ductility was observed in the Mg-(5.6Ti + 1.0SiC)(BM) hybrid composites. The superior strength properties of the Mg-(5.6Ti + x-SiC)(BM) hybrid composites when compared to Mg-5.6Ti is attributed to the presence of nano-reinforcements, the uniform distribution of the hybridized particles and the better interfacial bonding between the matrix and the reinforcement particles, achieved by nano-SiC addition.

Mechanism design for cost optimal PAC learning in the presence of strategic noisy annotators

We consider the problem of Probably Ap-proximate Correct (PAC) learning of a bi-nary classifier from noisy labeled exam-ples acquired from multiple annotators(each characterized by a respective clas-sification noise rate). First, we consider the complete information scenario, where the learner knows the noise rates of all the annotators. For this scenario, we derive sample complexity bound for the Mini-mum Disagreement Algorithm (MDA) on the number of labeled examples to be ob-tained from each annotator. Next, we consider the incomplete information sce-nario, where each annotator is strategic and holds the respective noise rate as a private information. For this scenario, we design a cost optimal procurement auc-tion mechanism along the lines of Myer-son���s optimal auction design framework in a non-trivial manner. This mechanism satisfies incentive compatibility property,thereby facilitating the learner to elicit true noise rates of all the annotators.

Microstructural evolution and mechanical properties of Mg composites containing nano-B4C hybridized micro-Ti particulates

In this work, the microstructural evolution and mechanical properties of extruded Mg composites containing micro-Ti particulates hybridized with varying contents of nano-B4C are investigated, and compared with Mg-5.6Ti. Microstructural characterization showed the presence of uniformly distributed micro-Ti particles embedded with nano-B4C particulates that resulted in significant grain refinement. Electron back scattered diffraction (EBSD) analyses of Mg-(5.6Ti + x-B4C)(BM) hybrid composites showed that the addition of hybridized particle resulted in relatively more recrystallized grains, realignment of basal planes and extension of weak basal fibre texture when compared to Mg-5.6Ti. The evaluation of mechanical properties indicated improved strength with ductility retention in Mg-(5.6Ti + x-B4C)(BM) hybrid composites. When compared to Mg-5.6Ti, the superior strength properties of the Mg-(5.6Ti + xB(4)C)(BM) hybrid composites are attributed to the presence of nano-reinforcements, the uniform distribution of the hybridized particles, better interfacial bonding between the matrix and the reinforcement particles and the matrix grain refinement achieved by nano-B4C addition. The ductility enhancement obtained in hybrid composites can be attributed to the fibre texture spread and favourable basal plane orientation achieved due to nano B4C addition. (C) 2013 Elsevier B.V. All rights reserved.

Effect Of Directionality Of Grinding Marks On Friction And Formation Of Transfer Layer

Surface texture influences friction and transfer layer formation during sliding. In the present investigation, basic studies were conducted using inclined pin-on-plate sliding tester to understand the effect of directionality of surface grinding marks of hard material on friction and transfer layer formation during sliding against soft materials. 080 M40 steel plates were ground to attain different surface roughness with unidirectional grinding marks. Then pins made of soft materials such as pure Al, pure Mg and Al-Mg alloy were slid against the prepared steel plates. Grinding angle (i.e., the angle between direction of sliding and grinding marks) was varied between 0 degrees and 90 degrees in the tests. Experiments were conducted under both dry and lubricated conditions on each plate in ambient environment. It was observed that the transfer layer formation and the coefficient of friction, which has two components adhesion and plowing - depend primarily on the directionality of grinding marks of the harder mating surface, and independent of surface roughness of the harder mating surface. For the case of pure Mg, stick-slip phenomenon was observed under dry condition for all grinding angles and it was absent upto 20 degrees grinding angles under lubricated condition. However, for the case of Al, it was observed only under lubricated conditions for angles exceeding 20 degrees. As regards the alloy, namely, Al-Mg alloy, it, was absent in both conditions. For the case of pure Mg and Al, it was observed that the amplitude of stick-slip motion primarily depends on plowing component of friction. The grinding angle effect on coefficient of friction was attributed to the variation of plowing component of friction with grinding angle.