Feature-subspace aggregating: ensembles for stable and unstable learners

This paper introduces a new ensemble approach, Feature-Subspace Aggregating (Feating), which builds local models instead of global models. Feating is a generic ensemble approach that can enhance the predictive performance of both stable and unstable learners. In contrast, most existing ensemble approaches can improve the predictive performance of unstable learners only. Our analysis shows that the new approach reduces the execution time to generate a model in an ensemble through an increased level of localisation in Feating. Our empirical evaluation shows that Feating performs significantly better than Boosting, Random Subspace and Bagging in terms of predictive accuracy, when a stable learner SVM is used as the base learner. The speed up achieved by Feating makes feasible SVM ensembles that would otherwise be infeasible for large data sets. When SVM is the preferred base learner, we show that Feating SVM performs better than Boosting decision trees and Random Forests. We further demonstrate that Feating also substantially reduces the error of another stable learner, k-nearest neighbour, and an unstable learner, decision tree.     

Neural mass activity, bifurcations, and epilepsy

In this letter, we propose a general framework for studying neural mass models defined by ordinary differential equations. By studying the bifurcations of the solutions to these equations and their sensitivity to noise, we establish an important relation, similar to a dictionary, between their behaviors and normal and pathological, especially epileptic, cortical patterns of activity. We then apply this framework to the analysis of two models that feature most phenomena of interest, the Jansen and Rit model, and the slightly more complex model recently proposed by Wendling and Chauvel. This model-based approach allows us to test various neurophysiological hypotheses on the origin of pathological cortical behaviors and investigate the effect of medication. We also study the effects of the stochastic nature of the inputs, which gives us clues about the origins of such important phenomena as interictal spikes, interictal bursts, and fast onset activity that are of particular relevance in epilepsy.     

On the simulation of nonlinear bidimensional spiking neuron models

Bidimensional spiking models are garnering a lot of attention for their simplicity and their ability to reproduce various spiking patterns of cortical neurons and are used particularly for large network simulations. These models describe the dynamics of the membrane potential by a nonlinear differential equation that blows up in finite time, coupled to a second equation for adaptation. Spikes are emitted when the membrane potential blows up or reaches a cutoff θ. The precise simulation of the spike times and of the adaptation variable is critical, for it governs the spike pattern produced and is hard to compute accurately because of the exploding nature of the system at the spike times. We thoroughly study the precision of fixed time-step integration schemes for this type of model and demonstrate that these methods produce systematic errors that are unbounded, as the cutoff value is increased, in the evaluation of the two crucial quantities: the spike time and the value of the adaptation variable at this time. Precise evaluation of these quantities therefore involves very small time steps and long simulation times. In order to achieve a fixed absolute precision in a reasonable computational time, we propose here a new algorithm to simulate these systems based on a variable integration step method that either integrates the original ordinary differential equation or the equation of the orbits in the phase plane, and compare this algorithm with fixed time-step Euler scheme and other more accurate simulation algorithms.     

Child head injury criteria investigation through numerical simulation of real world trauma

Finite element modelling has been used for decades in the study of adult head injury biomechanics and determination of injury criteria. Interest is recently growing in investigation on pediatric head injury which requires elaboration of biofidelic models that take into account child's head particularities in terms of size, geometry, and mechanical properties. In this study, a finite element model of a 3-year-old child head is proposed. The model is reconstructed from real CT scan images and mechanical properties are extracted from available data in the literature. A large number of real accidents (25 falls) are reconstructed with proposed model using different brain constitutive relationships in order to investigate their influence on model response. Mechanical output parameters (HIC, pressure, shearing stress) are calculated from these simulations. Statistical analysis was performed in order to evaluate predictive capability of the parameters. Von Mises stress appears to be clearly the most predictive parameters, allowing clear distinction between injured and non-injured cases. To the authors' knowledge, this study proposes for the first time a statistically based neurological injury criterion for a pediatric population using finite element modelling.     

Utilising psychophysical techniques to investigate the effects of age,typeface design,size and display polarity on glance legibility

Psychophysical research on text legibility has historically investigated factors such as size, colour and contrast, but there has been relatively little direct empirical evaluation of typographic design itself, particularly in the emerging context of glance reading. In the present study, participants performed a lexical decision task controlled by an adaptive staircase method. Two typefaces, a ‘humanist’ and ‘square grotesque’ style, were tested. Study I examined positive and negative polarities, while Study II examined two text sizes. Stimulus duration thresholds were sensitive to differences between typefaces, polarities and sizes. Typeface also interacted significantly with age, particularly for conditions with higher legibility thresholds. These results are consistent with previous research assessing the impact of the same typefaces on interface demand in a simulated driving environment. This simplified methodology of assessing legibility differences can be adapted to investigate a wide array of questions relevant to typographic and interface designs.

Practitioner Summary: A method is described for rapidly investigating relative legibility of different typographical features. Results indicate that during glance-like reading induced by the psychophysical technique and under the lighting conditions considered, humanist-style type is significantly more legible than a square grotesque style, and that black-on-white text is significantly more legible than white-on-black.     

The impact of false warnings on partial and full lane departure warnings effectiveness and acceptance in car driving

In the past, lane departure warnings (LDWs) were demonstrated to improve driving behaviours during lane departures but little is known about the effects of unreliable warnings. This experiment focused on the influence of false warnings alone or in combination with missed warnings and warning onset on assistance effectiveness and acceptance. Two assistance unreliability levels (33 and 17%) and two warning onsets (partial and full lane departure) were manipulated in order to investigate interaction. Results showed that assistance, regardless unreliability levels and warning onsets, improved driving behaviours during lane departure episodes and outside of these episodes by favouring better lane-keeping performances. Full lane departure and highly unreliable warnings, however, reduced assistance efficiency. Drivers’ assistance acceptance was better for the most reliable warnings and for the subsequent warnings. The data indicate that imperfect LDWs (false warnings or false and missed warnings) further improve driving behaviours compared to no assistance.

Practitioner Summary: This study revealed that imperfect lane departure warnings are able to significantly improve driving performances and that warning onset is a key element for assistance effectiveness and acceptance. The conclusion may be of particular interest for lane departure warning designers.     

Quantum dot light emitting devices for photomedical applications

While OLEDs have struggled to find a niche lighting application that can fully take advantage of their unique form factors as thin, flexible, lightweight and uniformly large‐area luminaire, photomedical researchers have been in search of low‐cost, effective illumination devices with such form factors that could facilitate widespread clinical applications of photodynamic therapy (PDT) or photobiomodulation (PBM). Although existing OLEDs with either fluorescent or phosphorescent emitters cannot achieve the required high power density at the right wavelength windows for photomedicine, the recently developed ultrabright and efficient deep red quantum dot light emitting devices (QLEDs) can nicely fit into this niche. Here, we report for the first time the in‐vitro study to demonstrate that this QLED‐based photomedical approach could increase cell metabolism over control systems for PBM and kill cancerous cells efficiently for PDT. The perspective of developing wavelength‐specific, flexible QLEDs for two critical photomedical fields (wound repair and cancer treatment) will be presented with their potential impacts summarized. The work promises to generate flexible QLED‐based light sources that could enable the widespread use and clinical acceptance of photomedical strategies including PDT and PBM.     

Optimal digital dynamical decoupling for general decoherence via Walsh modulation

We provide a general framework for constructing digital dynamical decoupling sequences based on Walsh modulation—applicable to arbitrary qubit decoherence scenarios. By establishing equivalence between decoupling design based on Walsh functions and on concatenated projections, we identify a family of optimal Walsh sequences, which can be exponentially more efficient, in terms of the required total pulse number, for fixed cancellation order, than known digital sequences based on concatenated design. Optimal sequences for a given cancellation order are highly non-unique—their performance depending sensitively on the control path. We provide an analytic upper bound to the achievable decoupling error and show how sequences within the optimal Walsh family can substantially outperform concatenated decoupling in principle, while respecting realistic timing constraints.     

Gradient based hyper-parameter optimisation for well conditioned kriging metamodels

In this work a two step approach to efficiently carrying out hyper parameter optimisation, required for building kriging and gradient enhanced kriging metamodels, is presented. The suggested approach makes use of an initial line search along the hyper-diagonal of the design space in order to find a suitable starting point for a subsequent gradient based optimisation algorithm. During the optimisation an upper bound constraint is imposed on the condition number of the correlation matrix in order to keep it from being ill conditioned. Partial derivatives of both the condensed log likelihood function and the condition number are obtained using the adjoint method, the latter has been derived in this work. The approach is tested on a number of analytical examples and comparisons are made to other optimisation approaches. Finally the approach is used to construct metamodels for a finite element model of an aircraft wing box comprising of 126 thickness design variables and is then compared with a sub-set of the other optimisation approaches.     

Inconsistent heuristics in theory and practice

In the field of heuristic search it is usually assumed that admissible heuristics are consistent, implying that consistency is a desirable attribute. The term “inconsistent heuristic” has, at times, been portrayed negatively, as something to be avoided. Part of this is historical: early research discovered that inconsistency can lead to poor performance for A^? (nodes might be re-expanded many times). However, the issue has never been fully investigated, and was not re-considered after the invention of IDA^?.This paper shows that many of the preconceived notions about inconsistent heuristics are outdated. The worst-case exponential time of inconsistent heuristics is shown to only occur on contrived graphs with edge weights that are exponential in the size of the graph. Furthermore, the paper shows that rather than being something to be avoided, inconsistent heuristics often add a diversity of heuristic values into a search which can lead to a reduction in the number of node expansions. Inconsistent heuristics are easy to create, contrary to the common perception in the AI literature. To demonstrate this, a number of methods for achieving effective inconsistent heuristics are presented.Pathmax is a way of propagating inconsistent heuristic values in the search from parent to children. This technique is generalized into bidirectional pathmax (BPMX) which propagates values from a parent to a child node, and vice versa. BPMX can be integrated into IDA^? and A^?. When inconsistent heuristics are used with BPMX, experimental results show a large reduction in the search effort required by IDA^?. Positive results are also presented for A^? searches.