Search

Multitask learning (MTL) can improve performance on one task by sharing representations with one or more related auxiliary tasks. Usually, MTL networks are trained on a composite loss function formed by a fixed weighted combination of separate task losses. In practice, however, static loss weights lead to poor results for two reasons. First, the relevance of the auxiliary tasks gradually drifts throughout the learning process. Second, for minibatch-based optimization, the optimal task weights vary significantly from one update to the next depending on the minibatch sample composition. Here, we introduce HydaLearn, an intelligent weighting algorithm that connects the main-task gain to the individual task gradients, to inform dynamic loss weighting at the minibatch level, addressing the two above shortcomings. We demonstrate significant performance increases on synthetic data and two real-world data sets.

Background: Compassionate Communities have been put forward as a new model for community-based palliative care to positively impact the health and wellbeing of those experiencing challenges of serious illness, death, dying, and loss. Despite the growing international movement to develop these public health initiatives to end-of-life care, only a handful of initiatives have undergone some form of evaluation. Aim: To provide guidance on designing evaluation research by identifying theoretical frameworks to understand the development, implementation, and underlying mechanisms of Compassionate Communities. Methods: To identify suitable theoretical frameworks for the study of Compassionate Communities, we applied two steps. The first step examined the characteristics of Compassionate Communities and translated them into assessment criteria for the selection of theoretical frameworks. The second step consisted of applying the identified assessment criteria to a list of widely used and highly cited theoretical frameworks. Results: Three well-established theoretical frameworks were identified as being most suitable to study the development, implementation, and underlying mechanisms of Compassionate Communities: The Consolidated Framework for Implementation Research (CFIR), the integrated-Promoting Action on Research Implementation in Health Services framework (i-PARIHS), and the Extended Normalization Process Theory (ENPT). Conclusions: The article supports and encourages the use of theoretical frameworks to evaluate the complex processes behind public health palliative care initiatives. The complementary use of two determinant frameworks and an implementation theory provides theoretical grounding to gain rich insights into the emergent and shifting interplays between agency, social processes, and contextual factors that shape the development and implementation of Compassionate Communities.

Compassionate Communities are places and environments in which people, networks, and institutions actively work together and are empowered to improve the circumstances, health, and well-being of those facing serious illness, death, dying, and loss. The study of their development, implementation, and evaluation requires an interdisciplinary research approach that has hitherto been lacking. In 2020, 8 research groups from 4 faculties at Vrije Universiteit Brussel united in the interdisciplinary Compassionate Communities Center of Expertise (COCO) to investigate Compassionate Communities. This article describes the first results of COCO: (a) an interdisciplinary mode of collaboration, (b) a shared conceptual understanding and definition of Compassionate Communities, and (c) a shared research agenda on Compassionate Communities.

In a Markov model the transition probabilities between states do not depend on the time spent in the current state. The present paper explores two ways of selecting the states of a discrete-time Markov model for a system partitioned into categories where the duration of stay in a category affects the probability of transition to another category. For a set of panel data, we compare the likelihood fits of the Markov models with states based on duration intervals and with states defined by duration values. For hierarchical systems, we show that the model with states based on duration values has a better maximum likelihood fit than the baseline Markov model where the states are the categories. We also prove that this is not the case for the duration-interval model, under conditions on the data that seem realistic in practice. Furthermore, we use the Akaike and Bayesian information criteria to compare these alternative Markov models. The theoretical findings are illustrated by an analysis of a real-world personnel data set.

The ability to solve differential equations represents a key step in the modeling and understanding of complex systems. There exist several analytical and numerical methods for solving differential equations, each with their own advantages and limitations. Physics-informed neural networks (PINNs) offer an alternative perspective. Although PINNs deliver promising results, many stones remain unturned about this method. In this paper, we introduce a method that improves the efficiency of PINNs in solving differential equations. Our method is related to the formulation of the problem: Instead of training a network to solve an nth order differential equation, we propose transforming the problem into the equivalent system of n first-order equations in phase space. The target of the network is to solve all equations of the system simultaneously, effectively introducing a multitask optimization problem. We compare both approaches empirically on various problems, ranging from second-order differential equations with constant coefficients to higher-order and nonlinear problems. We also show that our approach is suited for solving partial differential equations. Our results show that the system approach performs equal or better in most experiments performed. We analyze the learning process for the few runs that did not perform well and show that the problem stems from conflicting gradients during training, effectively obstructing multitask learning. The result of this paper is a straightforward heuristic that can be incorporated into any subsequent research that builds on PINNs solving differential equations. Moreover, it also shows how to make PINNs even more efficient by implementing techniques from multitask learning literature.

Causal classification concerns the estimation of the net effect of a treatment on an outcome of interest at the instance level, i.e., of the individual treatment effect (ITE). For binary treatment and outcome variables, causal classification models produce ITE estimates that essentially allow one to rank instances from a large positive effect to a large negative effect. Often, as in uplift modeling (UM), one is merely interested in this ranking, rather than in the ITE estimates themselves. In this regard, we investigate the potential of learning to rank (L2R) techniques to learn a ranking of the instances directly. We propose a unified formalization of different binary causal classification performance measures from the UM literature and explore how these can be integrated into the L2R framework. Additionally, we introduce a new metric for UM with L2R called the promoted cumulative gain (PCG). We employ the L2R technique LambdaMART to optimize the ranking according to PCG and show improved results over the use of standard L2R metrics and equal to improved results when compared with state-of-the-art UM. Finally, we show how L2R techniques can be used to specifically optimize for the top-k fraction of the ranking in a UM context, however, these results do not generalize to the test set. ispartof: IEEE Transactions on Knowledge and Data Engineering vol:34 issue:10 pages:4888-4904 status: published

In Markov chain theory, stochastic matrices are used to describe inter-state transitions. Powers of such transition matrices are computed to determine the behaviour within a Markov system. For this, diagonalizable matrices are preferred because of their useful properties. The non-diagonalizable matrices are therefore undesirable. The aim is to determine a nearby diagonalizable matrix A, starting from a non-diagonalizable matrix (A) over tilde. Previous studies tackled this problem, limited to 3 x 3 stochastic matrices. In this paper, these results are generalized for n x n stochastic matrices. Spectral properties of A are preserved in this process, such that A and (A) over tilde have coinciding semisimple eigenvalues and coinciding corresponding eigenvectors. This problem is examined and solved in this study and an algorithm is presented to find such a diagonalizable matrix (A) over tilde.

Machine learning offers the potential to revolutionize the inverse design of complex nanophotonic components. Here, we propose a novel variant of this formalism specifically suited for the design of resonant nanophotonic components. Typically, the first step of an inverse design process based on machine learning is training a neural network to approximate the non-linear mapping from a set of input parameters to a given optical system’s features. The second step starts from the desired features, e.g. a transmission spectrum, and propagates back through the trained network to find the optimal input parameters. For resonant systems, this second step corresponds to a gradient descent in a highly oscillatory loss landscape. As a result, the algorithm often converges into a local minimum. We significantly improve this method’s efficiency by adding the Fourier transform of the desired spectrum to the optimization procedure. We demonstrate our method by retrieving the optimal design parameters for desired transmission and reflection spectra of Fabry–Pérot resonators and Bragg reflectors, two canonical optical components whose functionality is based on wave interference. Our results can be extended to the optimization of more complex nanophotonic components interacting with structured incident fields.

We consider two covering variants of the network design problem. We are given a set of origin/destination pairs, called O/D pairs, and each such O/D pair is covered if there exists a path in the network from the origin to the destination whose length is not larger than a given threshold. In the first problem, called the Maximal Covering Network Design problem, one must determine a network that maximizes the total fulfilled demand of the covered O/D pairs subject to a budget constraint on the design costs of the network. In the second problem, called the Partial Covering Network Design problem, the design cost is minimized while a lower bound is set on the total demand covered. After presenting formulations, we develop a Benders decomposition approach to solve the problems. Further, we consider several stabilization methods to determine Benders cuts as well as the addition of cut-set inequalities to the master problem. We also consider the impact of adding an initial solution to our methods. Computational experiments show the efficiency of these different aspects., SCOPUS: ar.j, info:eu-repo/semantics/published

Uplift modeling is the subfield of causal inference that focuses on the ranking of individuals by their treatment effects. Uplift models are typically evaluated using Qini curves or Qini scores. While intuitive, the theoretical grounding for Qini in the literature is limited, and the mathematical connection to the well-understood Receiver Operating Characteristic ROC is unclear. In this paper, we first introduce the ROCini, an uplift evaluation metric similar in intuition to Qini but derived from the well understood ROC. Using Ordinal Dominance Graph theory, the ROCini is extended to the pROCini, a mathematically better behaved metric that facilitates theoretical analysis. Exploiting the theoretical properties of pROCini, confidence bounds are derived. Finally, the empirical performance of ROCini and pROCini is validated in a simulation study.

Recent developments in causal machine learning open perspectives for new approaches that support decision-making in healthcare processes using causal models. In particular, Heterogeneous Treatment Effect (HTE) inference enables the estimation of causal treatment effects for individual cases, offering great potential in a process mining context. At the same time, HTE literature typically focuses on clinical outcome measures, disregarding process efficiency. This paper shows the potential of jointly considering the clinical and operational effects of treatments in the context of healthcare processes. Moreover, we present a simple pipeline that makes existing HTE machine learning techniques directly applicable to event logs. Besides these conceptual contributions, a proof-of-concept application starting from the publicly available sepsis event log is outlined, forming the basis for a critical reflection regarding HTE estimation in a process mining context.

Traditional gender beliefs and gender prejudice (sexism) are thought to play a key role in the reproduction of gender inequality. It is known that such traditional gender beliefs are more common among strongly religious people. Europe is facing a growing population of religious, often immigrant youth from Islamic countries such as Turkey and Morocco. Against that background, we investigate how ambivalent sexism is related to adolescents’ religious affiliation (being Christian or Muslim), religiosity (how important religion is in one's life), perceived pressure for religious conformity and ethnic background (native, Turkish, Moroccan) by performing multivariate multilevel regression analyses. We rely on data from two independent samples gathered in 2013 and 2018 by the Youth Research Platform among Dutch-speaking boys (N = 1637) and girls (N = 2058) between 14 and 18 years old. The results show the gendered ways in which religion and ambivalent sexism towards women are related. For girls, ambivalent sexism wasrelated to perceived pressure for religious conformity and ethnic background (more group-level aspects). Boys’ ambivalent sexism was related to more individual-level aspects such as higher religiosity. No differences were found between Muslim or Christian youth., Traditional gender beliefs and gender prejudice (sexism) are thought to play a key role in the reproduction of gender inequality. It is known that such traditional gender beliefs are more common among strongly religious people. Europe is facing a growing population of religious, often immigrant youth from Islamic countries such as Turkey and Morocco. Against that background, we investigate how ambivalent sexism is related to adolescents’ religious affiliation (being Christian or Muslim), religiosity (how important religion is in one's life), perceived pressure for religious conformity and ethnic background (native, Turkish, Moroccan) by performing multivariate multilevel regression analyses. We rely on data from two independent samples gathered in 2013 and 2018 by the Youth Research Platform among Dutch-speaking boys (N = 1637) and girls (N = 2058) between 14 and 18 years old. The results show the gendered ways in which religion and ambivalent sexism towards women are related. For girls, ambivalent sexism wasrelated to perceived pressure for religious conformity and ethnic background (more group-level aspects). Boys’ ambivalent sexism was related to more individual-level aspects such as higher religiosity. No differences were found between Muslim or Christian youth. © 2021 European Sociological Association.

Reciprocal copy number variations (CNVs) of 16p11.2 are associated with a wide spectrum of neuropsychiatric and neurodevelopmental disorders. Here, we use human induced pluripotent stem cells (iPSCs)-derived dopaminergic (DA) neurons carrying CNVs of 16p11.2 duplication (16pdup) and 16p11.2 deletion (16pdel), engineered using CRISPR-Cas9. We show that 16pdel iPSC-derived DA neurons have increased soma size and synaptic marker expression compared to isogenic control lines, while 16pdup iPSC-derived DA neurons show deficits in neuronal differentiation and reduced synaptic marker expression. The 16pdel iPSC-derived DA neurons have impaired neurophysiological properties. The 16pdel iPSC-derived DA neuronal networks are hyperactive and have increased bursting in culture compared to controls. We also show that the expression of RHOA is increased in the 16pdel iPSC-derived DA neurons and that treatment with a specific RHOA-inhibitor, Rhosin, rescues the network activity of the 16pdel iPSC-derived DA neurons. Our data suggest that 16p11.2 deletion-associated iPSC-derived DA neuron hyperactivation can be rescued by RHOA inhibition., 16p11.2 CNVs are associated with neurodevelopmental disorders. Here, the authors show that 16p11.2 deletion is associated with hyperactivation of human iPSC-derived dopaminergic neuron networks and is rescued by RHOA inhibition in vitro.

This paper proposes a polynomial-time algorithm to construct the monotone stepwise curve that minimizes the sum of squared errors with respect to a given cloud of data points. The fitted curve is also constrained on the maximum number of steps it can be composed of and on the minimum step length. Our algorithm relies on dynamic programming and is built on the basis that said curve-fitting task can be tackled as a shortest-path type of problem. Numerical results on synthetic and realistic data sets reveal that our algorithm is able to provide the globally optimal monotone stepwise curve fit for samples with thousands of data points in less than a few hours. Furthermore, the algorithm gives a certificate on the optimality gap of any incumbent solution it generates. From a practical standpoint, this piece of research is motivated by the roll-out of smart grids and the increasing role played by the small flexible consumption of electricity in the large-scale integration of renewable energy sources into current power systems. Within this context, our algorithm constitutes an useful tool to generate bidding curves for a pool of small flexible consumers to partake in wholesale electricity markets., SCOPUS: ar.j, info:eu-repo/semantics/published

In the spring of 2021, Amazon hosted the Last Mile Routing Research Challenge with support from the MIT Center for Transportation & Logistics. The goal of the challenge was to encourage participants to develop innovative approaches to produce solutions to a real-world route sequencing problem, leveraging a large set of real operational data released by Amazon. All finalists were asked to submit short technical papers to document their methodological approach and its expected performance. This proceedings document provides a consolidated overview of the papers submitted by the participants.

Automated decisions based on trained algorithms influence human life in an increasingly far-reaching way. In recent years, it has become clear that these decisions are often accompanied by bias and unfair treatment of different subpopulations.Meanwhile, several notions of fairness circulate in the scientific literature, with trade-offs between profit and fairness and between fairness metrics among themselves. Based on both analytical calculations and numerical simulations, we show in this study that some profit-fairness trade-offs and fairness-fairness trade-offs depend substantially on the underlying score distributions given to subpopulations and we present two complementary perspectives to visualize this influence. We further show that higher symmetry in scores of subpopulations can significantly reduce the trade-offs between fairness notions within a given acceptable strictness, even when sacrificing expressiveness. Our exploratory study may help to understand how to overcome the strict mathematical statements about the statistical incompatibility of certain fairness notions.

Discrete time Markov models are used in a wide variety of social sciences. However, these models possess the memoryless property, which makes them less suitable for certain applications. Semi-Markov models allow for more flexible sojourn time distributions, which can accommodate for duration of stay effects. An overview of differences and possible obstacles regarding the use of Markov and semi-Markov models in manpower planning was first given by Valliant and Milkovich (1977). We further elaborate on their insights and introduce hybrid semi-Markov models for open systems with transition-dependent sojourn time distributions. Hybrid semi-Markov models aim to reduce model complexity in terms of the number of parameters to be estimated by only taking into account duration of stay effects for those transitions for which it is useful. Prediction equations for the stock vector are derived and discussed. Furthermore, the insights are illustrated and discussed based on a real world personnel dataset. The hybrid semi-Markov model is compared with the Markov and the semi-Markov models by diverse model selection criteria.

Student dropout is a major concern in higher education, as it leads to direct economic losses and substantial social costs. Public and private institutions spend considerable resources to prevent student dropout. The efficiency and effectiveness of these investments, however, may be improved by adopting a profit-driven perspective. In this paper, we propose a novel approach for implementing student dropout prediction using data-driven methods. Extending upon profit metrics as used in business analytics, we design a novel performance measure for evaluating predictive models that is tailored to the student dropout problem and that quantifies the net savings of a retention campaign. This metric supports the identification and selection of students to optimally allocate the limited resources for preventing student dropout and to maximize the resulting savings. Experiments were performed using data from three bachelor's programs of a higher education institution containing information on dropouts and participation in a retention program, i.e., tutorials. The proposed metric allows for a better choice of prediction model and classification threshold than conventional approaches and, as a result, yields tangible savings for the institution. Finally, the presented approach and experimental results highlight pathways to design tailored student retention programs.

Operational forecasting in supply chain management supports a variety of short-term planning decisions, such as production scheduling and inventory management. In this respect, improving short-term forecast accuracy is a way to build a more agile supply chain for manufacturing companies. Demand forecasting often relies on well-established univariate forecasting methods to extrapolate historical demand. Collaboration across the supply chain, including information sharing, is suggested in the literature to improve upon the forecast accuracy of such traditional methods. In this paper, we review empirical studies considering the use of downstream information in demand forecasting and investigate different modeling approaches and forecasting methods to incorporate such data. Where empirical findings on information sharing mainly focus on point-of-sale data in two-level supply chains, this research empirically investigates the added value of using sell-through data originating from intermediaries, next to historical demand figures, in a multi-echelon supply chain. In a case study concerning a US drug manufacturer, we evaluate different methods to incorporate this data and consider both time series methods and machine learning techniques to produce multi-step ahead weekly forecasts. The results show that the manufacturer can effectively improve its short-term forecast accuracy by integrating sell-through data into the forecasting process and provide useful insights as to the different modeling approaches used. The conclusion holds for all forecast horizons considered, though it is most pronounced for one-step ahead forecasts. Therefore, our research provides a clear incentive for manufacturers to assess the forecast accuracy that can be achieved by using sell-through data.

People do not change easily. They seem irrationally fond of the status quo and need a great deal of convincing before changing habits, even when the change appears beneficial. For instance, to get people to change modes of transportation, policymakers and researchers have thought up a great deal of incentives. Unfortunately, these incentives do not always have the desired effect or sometimes even have the adverse impact. Withing the novel framework of Ergodicity Economics, it is possible to explains some of these failures by highlighting our overdependence on ensemble averages when designing incentives and judging the benefits of a modal shift. Indeed, when modelling human decision-making, ergodicity is often implicitly assumed. Consequently, ensemble averages judge the (ir)rationality of human decision-making. However, when we step away from this assumption, i.e., ensemble averages and time averages are no longer assumed to be equal, we can rationalize much of this seemingly irrational behaviour by looking at the time average of an individual in a dynamic and stochastic environment, instead of relying on the expected value of the process. Loss aversion, e.g., rather than being a compulsory drive to cling to what you already have, becomes a way of mitigating downside risk and eventual ruin. The status quo bias or the irrational hesitance towards change also makes sense in a non-ergodic world as uncertainty looms greater in a non-ergodic world than in an ergodic one. In this contriution, we will use transportation data to demonstrate that, we can shift the dialogue regarding strategies for inducing behavioural change from one that focuses on idiosyncratic preferences and irrationalities to one that enhances the objective drivers of change. Instead of looking into personal reasons for not opting for sustainable mobility, we look at the determining factors objectively keeping us from change in a non-ergodic world.

In the majority of executive domains, a notion of normality is involved in most strategic decisions. However, few data-driven tools that support strategic decision-making are available. We introduce and extend the use of autoencoders to provide strategically relevant granular feedback. A first experiment indicates that experts are inconsistent in their decision making, highlighting the need for strategic decision support. Furthermore, using two large industry-provided human resources datasets, the proposed solution is evaluated in terms of ranking accuracy, synergy with human experts, and dimension-level feedback. This three-point scheme is validated using (a) synthetic data, (b) the perspective of data quality, (c) blind expert validation, and (d) transparent expert evaluation. Our study confirms several principal weaknesses of human decision-making and stresses the importance of synergy between a model and humans. Moreover, unsupervised learning and in particular the autoencoder are shown to be valuable tools for strategic decision-making. ispartof: Decision Support Systems vol:150 status: published

The traditional Capacitated Vehicle Routing Problem (CVRP) minimizes the total distance of the routes under the capacity constraints of the vehicles. But more often, the objective involves multiple criteria including not only the total distance of the tour but also other factors such as travel costs, travel time, and fuel consumption. Moreover, in reality, there are numerous implicit preferences ingrained in the minds of the route planners and the drivers. Drivers, for instance, have familiarity with certain neighborhoods and knowledge of the state of roads, and often consider the best places for rest and lunch breaks. This knowledge is difficult to formulate and balance when operational routing decisions have to be made. This motivates us to learn the implicit preferences from past solutions and to incorporate these learned preferences in the optimization process. These preferences are in the form of arc probabilities, i.e., the more preferred a route is, the higher is the joint probability. The novelty of this work is the use of a neural network model to estimate the arc probabilities, which allows for additional features and automatic parameter estimation. This first requires identifying suitable features, neural architectures and loss functions, taking into account that there is typically few data available. We investigate the difference with a prior weighted Markov counting approach, and study the applicability of neural networks in this setting., LIPIcs, Vol. 210, 27th International Conference on Principles and Practice of Constraint Programming (CP 2021), pages 42:1-42:17

This article presents a modeling approach to tackle the multidimensional military manpower planning problem. Requirements relevant to military manpower planning include the need to maintain distributions of manpower characteristics (such as rank and age) with respect to career advancement within acceptable proximity to desired norms. Moreover, they include ensuring that the personnel allocated to the various roles of the defence force will be adequate throughout the planning period. Addressing these tasks simultaneously represents a significant challenge for human resources managers. This paper proposes a model that assigns personnel to career paths (CPs) pre-defined as feasible prior to optimization. Adequate solutions through this model are generated by mixed integer goal programming. This solving approach is applied to a case study of the Belgian Defence’s manpower and results in valuable insights for military human resources managers. For the illustration of our approach, we focus on a military organization, but it could be used for any hierarchical organization, such as a police force or a university.

In an inventory system, the lead time is the time between the moment when a decision is made about the quantity to order (or produce) and the moment when this amount is available to satisfy demand. This lead time has an impact on inventory (holding and shortage) costs as well as on the purchase or production costs. Usually, shorter lead times result in lower inventory costs together with higher purchase or production costs. The lower inventory costs are caused by a reduced exposure to demand uncertainty, while the higher purchase or production costs are due to expediting the production and/or delivery process which typically comes with additional expenses. In this paper, we consider a periodic review dynamic order-up-to inventory system in which shortages are backordered. Such a dynamic system is appropriate when demand is non-stationary and demand forecast distributions are a crucial input to the inventory system. We propose a framework to quantify the effect of reducing (or extending) the lead time on the exposure to forecast uncertainty in terms of a justified increase (or required decrease) in unit purchase or production cost. This difference in unit purchase or production cost quantifies the offset to compensate for the change in exposure to forecast uncertainty, and hence the change in inventory costs. The framework is based on the work of de Treville et al. (2014) which deals with valuing lead time in the single-period newsvendor problem. We apply the framework to a case study and illustrate its use as decision support tool for deciding what lead time to use. In the case study, sales history and open order data are available to make demand forecasts. As more and more orders for a specific period are collected when approaching this period, naturally, the open order data strongly affects the evolution of forecast uncertainty in lead time, making the framework of particular interest.

ispartof: CRITICAL CARE vol:24 issue:1 ispartof: location:England status: published

The electromagnetic near field enables subwavelength applications such as near-field microscopy and nanoparticle manipulation. Present methods to structure the near field rely on optical antenna theory, involving nanostructures that locally convert propagating waves into confined near-field patterns. We developed a theory of remote rather than local near-field shaping, based on cascaded mode conversion and interference of counterpropagating guided waves with different propagation constants. We demonstrate how to structure at will the longitudinal and transverse variation of the near field, allowing for distributions beyond the conventional monotonic decay of the evanescent field. We provide an experimental realization that confirms our theory. Our method applies to fields with arbitrary polarization states and mode profiles, providing a path toward three-dimensional control of the near field.

Uplift modeling is an approach for estimating the incremental effect of an action or treatment at the individual level. It has gained attention in the marketing and analytics communities due to its ability to adequately model the effect of direct marketing actions via predictive analytics. The main contribution of our study is the implementation of the uplift modeling framework to maximize the effectiveness of retention efforts in higher education institutions i.e., improvement of academic performance by offering tutorials. The objective is to improve the design of retention programs by tailoring them to students who are more likely to be retained if targeted. Data from three different bachelor programs from a Chilean university were collected. Students who participated in the tutorials are considered the treatment group, otherwise, they are assigned to the nontreatment group. Our results demonstrate the virtues of uplift modeling in tailoring retention efforts in higher education over conventional predictive modeling approaches.

Background Ventilator sharing is one option to emergently increase ventilator capacity during a crisis but has been criticized for its inability to adjust for individual patient needs. Newer ventilator sharing designs use valves and restrictors to control pressures for each patient. A key component of these designs is an inline Positive End Expiratory Pressure (PEEP) Valve but these are not readily available. Creating an inline PEEP valve by converting a standard bag-valve-mask PEEP valve is possible with the addition of a 3D printed collar. Methods This was a feasibility study assessing the performance and safety of a method for converting a standard PEEP valve into an inline PEEP valve. A collar was designed and printed that covers the exhaust ports of the valve and returns exhaled gases to the ventilator. Results The collar piece was simple to print and easily assembled with the standard PEEP valve. In bench testing it successfully created differential pressures in 2 simulated expiratory limbs without leaking to the atmosphere at pressures greater than 60 cm of H2O. Conclusion Our novel inline PEEP valve design shows promise as an option for building a safer ventilator sharing system.

A particular class of focused, pulsed light beams can propagate self-similarly in free space at a fixed group velocity. Now, scientists present a law of refraction that determines how the group velocity of these beams changes as they refract at an interface between two materials.

PurposeThe main goal of employee retention is to prevent competent employees from leaving the company. When analysing the main reasons why employees leave and determining their turnover probability, the question arises: Which retention strategies have an actual effect on turnover and for which profile of employees do these strategies work?Design/methodology/approachTo determine the effectiveness of different retention strategies, an overview is given of retention strategies that can be found in the literature. Next, the paper presents a procedure to build an uplift model for testing the effectiveness of the different strategies on HR data. The uplift model is based on random forest estimation and applies personal treatment learning estimation.FindingsThrough a data-driven approach, the actual effect of retention strategies on employee turnover is investigated. The retention strategies compensation and recognition are found to have a positive average treatment effect on the entire population, while training and flexibility do not. However, with personalised treatment learning, the treatment effect on the individual level can be estimated. This results in an ability to profile employees with the highest estimated treatment effect.Practical implicationsThe results yield useful information for human resources practitioners. The personalised treatment analysis results in detailed retention information for these practitioners, which allows them to target the right employees with the right strategies.Originality/valueEven though the uplift modelling approach is becoming increasingly popular within marketing, this approach has not been taken within human resources analytics. This research opens the door for further research and for practical implementation.

We show that it is possible for every non-diagonalizable stochastic 3 × 3 matrix to be perturbed into a diagonalizable stochastic matrix with the eigenvalues, arbitrarily close to the eigenvalues of the original matrix, with the same principal eigenspaces. An algorithm is presented to determine a perturbation matrix, which preserves these spectral properties. Additionally, a relation is proved between the eigenvectors and generalized eigenvectors of the original matrix and the perturbed matrix.

The current business environment requires companies to design strategies that maintain customers loyal and engaged. Retention programs are popular schemes that aim to retain existing customers through incentives. Nonetheless, retention programs do not necessarily lead to returns when they target individuals who will not respond favorably. Business analytics has become in recent years a supporting field for decision-making, since it combines notions from machine learning and management science to extract helpful insights from business data. An extensive set of predictive modeling techniques and performance evaluation metrics have been developed to support targeting decisions. Conventional evaluation metrics, however, do not necessarily lead to the selection of the best model, i.e., the model that maximizes the profit of retention campaigns. Hence, the need for profit-driven evaluation approaches. This study extends the existing literature on profit-driven churn prediction evaluation by taking into account customer heterogeneity and by assessing the framework from the perspective of mutual funds. The intuition is that certain customers generate more value than others, and hence targeting on the basis of churn propensities is suboptimal. Moreover, attrition in mutual funds is highly dependent on the state of the portfolio. The multithreshold multisegment profit-driven churn prediction evaluation framework for mutual funds defines customer heterogeneity on the basis of customer lifetime value (CLV) levels. The customer base is segmented according to the different CLV levels, and the optimal fraction of customers to target is determined within each segment according to: the churn propensities, the segment-wise average CLV, and the costs and success rate of the incentive. Six classifiers are trained based on information regarding customer demographics, customer-firm interactions and financial indicators. Subsequently, current profit-driven churn prediction evaluation approaches are contrasted against our framework. Theresults consistently indicate that our framework leads to the largest average profit when compared with conventional profit metrics. In addition, we perform a sensitivity analysis to determine the impact of increased retention campaign costs on the average profit and the number of targeted individuals. Our approach prioritizes targeting valuable customers, i.e., segments with a large CLV, as the retention campaign becomes more expensive.

Current day high speed optical communication systems employ photonic circuits using platforms such as silicon photonics. In these systems, the polarization state of light drifts due to effects such as polarization mode dispersion and nonlinear phenomena generated by photonic circuit building blocks. As the complexity, the number, and the variety of these building blocks grows, the demand increases for an in-situ polarization determination strategy. Here, we show that the transfer of the Belinfante momentum to particles in the evanescent field of waveguides depends in a non-trivial way on the polarization state of light within that waveguide. Surprisingly, we find that the maxima and minima of the lateral force are not produced with circularly polarized light, corresponding to the north and south poles of the Poincaré sphere. Instead, the maxima are shifted along the great circle of the sphere due to the phase differences between the scattered TE and TM components of light. This effect allows for an unambiguous reconstruction of the local polarization state of light inside a waveguide. Importantly, this technique depends on interaction with only the evanescent tails of the fields, allowing for a minimally invasive method to probe the polarization within a photonic chip.

International audience; In their ongoing search for competitive advantage, firms increasingly leverage online innovation communities (ICs). The viability of these ICs may be jeopardized by big data environments and inferior member participation. Therefore, community managers must address poor member participation, together with the data-rich environment. This study examines the viability of a proactive motivational email campaign to reduce inferior member participation using uplift modeling; it also explores optimal treatment characteristics, including message scope (untargeted versus targeted), message content (hedonic, cognitive, and social message), and member profile (self-interest–oriented and positive emotional writing style). The findings indicate that marketing decision makers should use proactive, targeted emails with cognitive motivational elements to mitigate inferior levels of member participation. These findings have important implications for innovation scholars and community managers.

Metasurfaces allow for agile manipulation of incoming light using a single layer of resonators. Despite recent progress, it remains difficult to generate new spectral components using nonlinear surfaces, because of the limited interaction length of a pulse traveling through a single surface. Time-dependent surfaces offer an exciting alternative to overcome this limitation, but a self-consistent framework to describe this mechanism is lacking. Based on an analytical model and finite-difference time-domain numerical simulations, we obtain physical insight into the frequency-shifting process that occurs when broadband electromagnetic pulses interact with time-varying surfaces. In particular, we find that there is an intriguing relationship between the bandwidth of the incident pulse, the targeted frequency shift, and the number of Lorentzian resonators that need to be implemented. We also demonstrate that in certain parameter regimes, pulse distortion and a deviation of pulse amplitude cannot be avoided. These results are independent of the mechanism that generates the time dependence. They are also independent of the frequency, the geometry, size, and material of the unit cell, but they are rather a direct result of the subtle interplay between time-dependent Lorentzian resonances.

The goal of this paper is to investigate a decision support system for vehicle routing, where the routing engine learns from the subjective decisions that human planners have made in the past, rather than optimizing a distance-based objective criterion. This is an alternative to the practice of formulating a custom VRP for every company with its own routing requirements. Instead, we assume the presence of past vehicle routing solutions over similar sets of customers, and learn to make similar choices. The approach is based on the concept of learning a first-order Markov model, which corresponds to a probabilistic transition matrix, rather than a deterministic distance matrix. This nevertheless allows us to use existing arc routing VRP software in creating the actual route plans. For the learning, we explore different schemes to construct the probabilistic transition matrix. Our results on a use-case with a small transportation company show that our method is able to generate results that are close to the manually created solutions, without needing to characterize all constraints and sub-objectives explicitly. Even in the case of changes in the client sets, our method is able to find solutions that are closer to the actual route plans than when using distances, and hence, solutions that would require fewer manual changes to transform into the actual route plan.

We study the predict+optimise problem, where machine learning and combinatorial optimisation must interact to achieve a common goal. These problems are important when optimisation needs to be performed on input parameters that are not fully observed but must instead be estimated using machine learning. Our contributions are two-fold: 1) we provide theoretical insight into the properties and computational complexity of predict+optimise problems in general, and 2) develop a novel framework that, in contrast to related work, guarantees to compute the optimal parameters for a linear learning function given any ranking optimisation problem. We illustrate the applicability of our framework for the particular case of the unit-weighted knapsack predict+optimise problem and evaluate on benchmarks from the literature.

We study a prediction�+�optimisation formulation of the knapsack problem. The goal is to predict the profits of knapsack items based on historical data, and afterwards use these predictions to solve the knapsack. The key is that the item profits are not known beforehand and thus must be estimated, but the quality of the solution is evaluated with respect to the true profits. We formalise the problem, the goal of minimising expected regret and the learning problem, and investigate different machine learning approaches that are suitable for the optimisation problem. Recent methods for linear programs have incorporated the linear relaxation directly into the loss function. In contrast, we consider less intrusive techniques of changing the loss function, such as standard and multi-output regression, and learning-to-rank methods. We empirically compare the approaches on real-life energy price data and synthetic benchmarks, and investigate the merits of the different approaches.

Much of the power of CP comes from the ability to create complex hybrid search algorithms specific to an application. Unfortu- nately there is no widely accepted standard for specifying search, and each solver typically requires detailed knowledge in order to build com- plex searches. This makes the barrier to entry for exploring different search methods quite high. Furthermore, search is a core part of the solver and usually highly optimised. Any imposition on the solver writer to change this part of their system is significant. In this paper we investigate how powerful we can make a uniform lan- guage for meta-search without placing any burden on the solver writer. The key to this is to only interact with the solver when a solution is found. We present MiniSearch, a meta-search language that can directly use any FlatZinc solver. Optionally, it can interact with solvers through an efficient C++ API. We illustrate the expressiveness of the language and performance using different solvers on a number of examples. ispartof: pages:376-392 ispartof: Proceedings of the 21st International Conference on Principles and Practice of Constraint Programming (CP 2015) vol:9255 pages:376-392 ispartof: Principles and Practice of Constraint Programming location:Cork, Ireland date:31 Aug - 4 Sep 2015 status: published

We introduce the problem of k-pattern set mining, concerned with finding a set of k related patterns under constraints. This contrasts to regular pattern mining, where one searches for many individual patterns. The k-pattern set mining problem is a very general problem that can be instantiated to a wide variety of well-known mining tasks including concept-learning, rule-learning, redescription mining, conceptual clustering and tiling. To this end, we formulate a large number of constraints for use in k-pattern set mining, both at the local level, that is, on individual patterns, and on the global level, that is, on the overall pattern set. Building general solvers for the pattern set mining problem remains a challenge. Here, we investigate to what extent constraint programming (CP) can be used as a general solution strategy. We present a mapping of pattern set constraints to constraints currently available in CP. This allows us to investigate a large number of settings within a unified framework and to gain insight in the possibilities and limitations of these solvers. This is important as it allows us to create guidelines in how to model new problems successfully and how to model existing problems more efficiently. It also opens up the way for other solver technologies. ispartof: IEEE Transactions on Knowledge and Data Engineering vol:25 issue:2 pages:402-418 status: published

Finding small sets of interesting patterns is an important challenge in pattern mining. In this paper, we argue that several well-known approaches that address this challenge are based on performing pairwise comparisons between patterns. Examples include finding closed patterns, free patterns, relevant subgroups and skyline patterns. Although progress has been made on each of these individual problems, a generic approach for solving these problems (and more) is still lacking. This paper tackles this challenge. It proposes a novel, generic approach for handling pattern mining problems that involve pairwise compar- isons between patterns. Our key contributions are the following. First, we propose a novel algebra for programming pattern mining problems. This algebra extends relational algebras in a novel way towards pattern mining. It allows for the generic combination of constraints on individual patterns with dominance relations between patterns. Second, we introduce a modified generic constraint satisfaction system to evaluate these algebraic expressions. Experiments show that this generic approach can indeed effectively identify patterns expressed in the algebra. ispartof: pages:557-566 ispartof: 13th IEEE International Conference on Data Mining pages:557-566 ispartof: IEEE International Conference on Data Mining location:Dallas, Texas, USA date:7 Dec - 10 Dec 2013 status: published

Most traditional biclustering algorithms identify biclusters with no or little overlap. In this paper, we introduce the problem of identifying staircases of biclusters. Such staircases may be indicative for causal relationships between columns and can not easily be identified by existing biclustering algorithms. Our formalization relies on a scoring function based on the Minimum Description Length principle. Furthermore, we propose a first algorithm for identifying staircase biclusters, based on a combination of local search and constraint programming. Experiments show that the approach is promising. ispartof: pages:139-146 ispartof: 12th IEEE International Conference on Data Mining Workshops pages:139-146 ispartof: International workshop on Co-Clustering and Applications (CoClus'12) in conjunction with IEEE ICDM 2012 location:Brussels date:10 Dec - 10 Dec 2012 status: published

The pattern mining community has shifted its attention from local pattern mining to pattern set mining. The task of pattern set mining is concerned with finding a set of patterns that satisfies a set of constraints and often also scores best w.r.t. an optimisation criteria. Furthermore, while in local pattern mining the constraints are imposed at the level of individual patterns, in pattern set mining they are also concerned with the overall set of patterns. A wide variety of different pattern set mining techniques is available in literature. The key contribution of this paper is that it studies, compares and evaluates such search strategies for pattern set mining. The investigation employs concept-learning as a benchmark for pattern set mining and employs a constraint programming framework in which key components of pattern set mining are formulated and implemented. The study leads to novel insights into the strong and weak points of different pattern set mining strategies. Book subtitle: 15TH PACIFIC-ASIA CONFERENCE, PAKDD 2011 ispartof: pages:382-394 ispartof: Advances in Knowledge Discovery and Data Mining, 15th Pacific-Asia Conference, PAKDD 2011 vol:6635 issue:PART 2 pages:382-394 ispartof: Pacific-Asia Conference on Knowledge Discovery and Data Mining location:Shenzhen, China date:24 May - 27 May 2011 status: published