Finite-size effects in on-line learning of multilayer neural networks

We complement recent advances in thermodynamic limit analyses of mean on-line gradient descent learning dynamics in multi-layer networks by calculating fluctuations possessed by finite dimensional systems. Fluctuations from the mean dynamics are largest at the onset of specialisation as student hidden unit weight vectors begin to imitate specific teacher vectors, increasing with the degree of symmetry of the initial conditions. In light of this, we include a term to stimulate asymmetry in the learning process, which typically also leads to a significant decrease in training time.

Concreteness effects in different tasks: implications for models of short-term memory

This study investigates concreteness effects in tasks requiring short-term retention. Concreteness effects were assessed in serial recall, matching span, order reconstruction, and free recall. Each task was carried out both in a control condition and under articulatory suppression. Our results show no dissociation between tasks that do and do not require spoken output. This argues against the redintegration hypothesis according to which lexical-semantic effects in short-term memory arise only at the point of production. In contrast, concreteness effects were modulated by task demands that stressed retention of item versus order information. Concreteness effects were stronger in free recall than in serial recall. Suppression, which weakens phonological representations, enhanced the concreteness effect with item scoring. In a matching task, positive effects of concreteness occurred with open sets but not with closed sets of words. Finally, concreteness effects reversed when the task asked only for recall of word positions (as in the matching task), when phonological representations were weak (because of suppression), and when lexical semantic representations overactivated (because of closed sets). We interpret these results as consistent with a model where phonological representations are crucial for the retention of order, while lexical-semantic representations support maintenance of item identity in both input and output buffers.

Crystal orientation effects in scratch testing with a spherical indenter

Spherical scratch tests were conducted in individual grains of a randomly oriented polycrystalline body-centered-cubic (bcc) Ti-Nb alloy. For each grain, scratch tests were conducted at four different levels of normal load, which resulted in varying amounts of plastic strain during indentation. The results show a dependence of the horizontal load component on the crystallographic orientation and on the amount of plastic strain. The component of the horizontal force that resulted from plastic deformation was found to correlate with the active slip systems for the particular grain orientation. © 2010 Materials Research Society.

Ownership effects in consumers' brand extension evaluations

In this paper, we examine how brand ownership status affects consumers’ evaluation of brand extensions, using an experiment. In evaluating both brand extension and parent brands, brand owners differ from both nonowners and nonusers of a brand's product category in important ways. While the functional similarity between a brand and its extension impacts on all three groups’ brand extensions, its effects on nonowners and nonusers are more significant than those on brand owners. For brand owners, the most important consideration in their evaluation of brand extensions seems to be the image consistency between a brand and its extensions. Furthermore, there is an interaction effect between brand image consistency and product similarity for brand owners, whereas this effect is nonexistent for nonowners and nonusers.

Contextual effects in speed perception may occur at an early stage of processing

How does nearby motion affect the perceived speed of a target region? When a central drifting Gabor patch is surrounded by translating noise, its speed can be misperceived over a fourfold range. Typically, when a surround moves in the same direction, perceived centre speed is reduced; for opposite-direction surrounds it increases. Measuring this illusion for a variety of surround properties reveals that the motion context effects are a saturating function of surround speed (Experiment I) and contrast (Experiment II). Our analyses indicate that the effects are consistent with a subtractive process, rather than with speed being averaged over area. In Experiment III we exploit known properties of the motion system to ask where these surround effects impact. Using 2D plaid stimuli, we find that surround-induced shifts in perceived speed of one plaid component produce substantial shifts in perceived plaid direction. This indicates that surrounds exert their influence early in processing, before pattern motion direction is computed. These findings relate to ongoing investigations of surround suppression for direction discrimination, and are consistent with single-cell findings of direction-tuned suppressive and facilitatory interactions in primary visual cortex (V1).

Radiation effects in metal-oxide-semiconductor capacitors

An investigation has been undertaken into the effects of various radiations on commercially made Al-SiO2-Si Capacitors (MOSCs). Detailed studies of the electrical and physical nature of such devices have been used to characterise both virgin and irradiated devices. In particular, an investigation of the nature and causes of dielectric breakdown in MOSCs has revealed that intrinsic breakdown is a two-stage process dominated by charge injection in a pre-breakdown stage; this is associated with localised high-field injection of carriers from the semiconductor substrate to interfacial and bulk charge traps which, it is proposed, leads to the formation of conducting channels through the dielectric with breakdown occurring as a result of the dissipation of the conduction band energy. A study of radiation-induced dielectric breakdown has revealed the possibility of anomalous hot-electron injection to an excess of bulk oxide traps in the ionization channel produced by very heavily ionizing radiation, which leads to intrinsic breakdown in high-field stressed devices. These findings are interpreted in terms of a modification to the model for radiation-induced dielectric breakdown based upon the primary dependence of breakdown on charge injection rather than high-field mechanisms. The results of a detailed investigation of charge trapping and interface state generation in such MOSCs due to various radiations has revealed evidence of neutron induced interface states, and of the generation of positive oxide charge in devices due to all of the radiations tested. In particular, the greater the linear energy transfer of the radiation, the greater the magnitude of charge trapped in the oxide and the greater the number of interface states generated. These findings are interpreted in terms of Si-H and Si-OH bond-breaking at the Si-SiO2 interface which is enhanced by charge carrier transfer to the interface and by anomalous charge injection to compensate for the excess of charge carriers created by the radiation.

A study of magnetic non-linearity and finite length effects in solid iron subjected to a travelling MMF wave

This thesis describes an experimental and analytic study of the effects of magnetic non-linearity and finite length on the loss and field distribution in solid iron due to a travelling mmf wave. In the first half of the thesis, a two-dimensional solution is developed which accounts for the effects of both magnetic non-linearity and eddy-current reaction; this solution is extended, in the second half, to a three-dimensional model. In the two-dimensional solution, new equations for loss and flux/pole are given; these equations contain the primary excitation, the machine parameters and factors describing the shape of the normal B-H curve. The solution applies to machines of any air-gap length. The conditions for maximum loss are defined, and generalised torque/frequency curves are obtained. A relationship between the peripheral component of magnetic field on the surface of the iron and the primary excitation is given. The effects of magnetic non-linearity and finite length are combined analytically by introducing an equivalent constant permeability into a linear three-dimensional analysis. The equivalent constant permeability is defined from the non-linear solution for the two-dimensional magnetic field at the axial centre of the machine to avoid iterative solutions. In the linear three-dimensional analysis, the primary excitation in the passive end-regions of the machine is set equal to zero and the secondary end faces are developed onto the air-gap surface. The analyses, and the assumptions on which they are based, were verified on an experimental machine which consists of a three-phase rotor and alternative solid iron stators, one with copper end rings, and one without copper end rings j the main dimensions of the two stators are identical. Measurements of torque, flux /pole, surface current density and radial power flow were obtained for both stators over a range of frequencies and excitations. Comparison of the measurements on the two stators enabled the individual effects of finite length and saturation to be identified, and the definition of constant equivalent permeability to be verified. The penetration of the peripheral flux into the stator with copper end rings was measured and compared with theoretical penetration curves. Agreement between measured and theoretical results was generally good.

Phase transitions and memory effects in the dynamics of Boolean networks

The generating functional method is employed to investigate the synchronous dynamics of Boolean networks, providing an exact result for the system dynamics via a set of macroscopic order parameters. The topology of the networks studied and its constituent Boolean functions represent the system's quenched disorder and are sampled from a given distribution. The framework accommodates a variety of topologies and Boolean function distributions and can be used to study both the noisy and noiseless regimes; it enables one to calculate correlation functions at different times that are inaccessible via commonly used approximations. It is also used to determine conditions for the annealed approximation to be valid, explore phases of the system under different levels of noise and obtain results for models with strong memory effects, where existing approximations break down. Links between Boolean networks and general Boolean formulas are identified and results common to both system types are highlighted. © 2012 Copyright Taylor and Francis Group, LLC.

Experiments and analysis of combined diffraction and self diffraction effects in a nematic liquid crystal cell

The potential for nonlinear optical processes in nematic-liquid-crystal cells is great due to the large phase changes resulting from reorientation of the nematic-liquid-crystal director. Here the combination of diffraction and self-diffraction effects are studied simultaneously by the use of a pair of focused laser beams which are coincident on a homeotropically aligned liquid-crystal cell. The result is a complicated diffraction pattern in the far field. This is analyzed in terms of the continuum theory for liquid crystals, using a one-elastic-constant approximation to solve the reorientation profile. Very good comparison between theory and experiment is obtained. An interesting transient grating, existing due to the viscosity of the liquid-crystal material, is observed in theory and practice for large cell-tilt angles.

Memory effects in a non-equilibrium growth model

We study memory effects in a kinetic roughening model. For d=1, a different dynamic scaling is uncovered in the memory dominated phases; the Kardar-Parisi-Zhang scaling is restored in the absence of noise. dc=2 represents the critical dimension where memory is shown to smoothen the roughening front (a=0). Studies on a discrete atomistic model in the same universality class reconfirm the analytical results in the large time limit, while a different scaling behavior shows up for ting the memory characteristic of the atomistic model. Results can be generalized for other nonconservative systems.