Beyond Relevance Ranking: A General Graph Matching Framework for Utility-Oriented Learning to Rank

Learning to rank from logged user feedback, such as clicks or purchases, is a central component of many real-world information systems. Different from human-annotated relevance labels, the user feedback is always noisy and biased. Many existing learning to rank methods infer the underlying relevance of query–item pairs based on different assumptions of examination, and still optimize a relevance based objective. Such methods rely heavily on the correct estimation of examination, which is often difficult to achieve in practice. In this work, we propose a general framework U-rank+ for learning to rank with logged user feedback from the perspective of graph matching. We systematically analyze the biases in user feedback, including examination bias and selection bias. Then, we take both biases into consideration for unbiased utility estimation that directly based on user feedback, instead of relevance. In order to maximize the estimated utility in an efficient manner, we design two different solvers based on Sinkhorn and LambdaLoss for U-rank+ . The former is based on a standard graph matching algorithm, and the latter is inspired by the traditional method of learning to rank. Both of the algorithms have good theoretical properties to optimize the unbiased utility objective while the latter is proved to be empirically more effective and efficient in practice. Our framework U-rank+ can deal with a general utility function and can be used in a widespread of applications including web search, recommendation, and online advertising. Semi-synthetic experiments on three benchmark learning to rank datasets demonstrate the effectiveness of U-rank+ . Furthermore, our proposed framework has been deployed on two different scenarios of a mainstream App store, where the online A/B testing shows that U-rank+ achieves an average improvement of 19.2% on click-through rate and 20.8% improvement on conversion rate in recommendation scenario, and 5.12% on platform revenue in online advertising scenario over the production baselines.

An Active Fault-Tolerant Control Strategy for Current Sensors Failure for Induction Motor Drives Using a Single Observer for Currents Estimation and Axes Transformation

This paper proposes a fault-tolerant control technique against current sensors failure in direct torque controlled induction motors drives, based on a new modification of Luenberger observer for currents estimation and axes transformation for vector rotation. Several important aspects are covered in the proposed algorithm, such as the detection of sensors failure, the isolation of faulty sensors, and the reconfiguration of the control system by a correct estimation. A logic circuit ensures fault detection by analyzing the residual signal between the measured and estimated quantities, while a single observer performs the task of estimating the line currents. In addition, a decision logic circuit isolates the erroneous signal and simultaneously selects the appropriate estimated current signal. An axes transformation ensures rotation from (a,b) to (α,β), which keeps a low-cost control using only two current sensors. The proposed scheme is tested on MATLAB/Simulink environment and experimentally validated in a laboratory prototype mainly containing a dS1104 card and 4 kW induction motor.

Measurements that matter with hybrid model of estimation

The biggest challenge of delivering projects on time, especially in the field of Telecommunication, is proper planning and correct estimation of the effort. Such estimations while different phases of the development could be extremely thorny and easily can jeopardize final goals and handover on time. Thus, in this paper, the proposed approach of Hybrid model estimation is to reduce such unlike stakes and give a clear definition of “how needs to be done” with regular estimations transferred to Story Points.

Naturally Ventilated Double Skin Façades: Comparisons Between Different CFD Models

Double Skin Façades (DSFs) have become widespread solutions commonly employed in new and existing buildings in the last decades. Since its introduction, the multi-layered façade has improved profoundly, assuming more articulate and complex shapes for better energy performances and combining advanced technologies as innovative materials or systems. However, the effectiveness and the thermal behaviour of DSFs should be carefully evaluated since the design phase by selecting proper methodologies, thus avoiding inaccurate results. In fact, the correct estimation of the airflows inside DSF channels is heavily influenced by the simulation settings. Furthermore, the lack of measurements or empirical validations in the field is the primary source of concern for researchers. Considering the available numerical methods for investigating DSFs, Computational Fluid Dynamics (CFD) simulations have proven to be the most appropriate option. The present work compares multiple Double Skin Façade configurations by performing CFD analyses and adopting different turbulence models in bi- and three-dimensional domains. The results underline the capability of 2D models in predicting the fluxes inside the DSF channel and in the domain. Furthermore, comparisons among the velocity profiles estimated by adopting different turbulence formulations highlight only slight variations, especially in proximity to the perturbated areas of the cavity.

SET-UP DUE TO RANDOM WAVES: INFLUENCE OF THE DIRECTIONAL SPECTRUM

The correct estimation of set-up is very important to evaluate coastal hazard and to design coastal structures. In this paper, we derived a mathematical expression for wave set-up in the context of random waves. The solution to this expression assumes straight, parallel depth contours and constant average flow parameters in the longshore direction. We then investigated the effect of different types of sea state taking account of different frequency spectrum and spreading function assumed in the expression on estimates of wave set-up. We found the set-up was highly influenced by the frequency spectrum used. Finally, we applied this expression to estimate set-up values at locations in Italy and in the United States using buoy data provided by ISPRA (Istituto Superiore per la Protezione e la Ricerca Ambientale) and NDBC (National Data Buoy Centre).

High Precision Half‑Life Measurement of the Extinct Radio‑Lanthanide Dysprosium-154

Sixty years after the discovery of 154Dy, the half-life of this pure alpha-emitter was re-measured. 154Dy was radiochemically separated from proton-irradiated tantalum samples. Sector field- and multicollector-inductively coupled plasma mass spectrometry were used to determine the amount of 154Dy retrieved. The disintegration rate of the radio-lanthanide was measured by means of α-spectrometry. The half-life value was determined as (1.33 ± 0.07)∙106 y, with an uncertainty reduced by a factor of ~10 compared to the currently adopted value of (3.0 ± 1.5)∙106 y. This precise half-life value is crucial for the correct estimation of p-process nucleosynthetic reactions in the lanthanide region, as well as for the safe disposal of irradiated target material from spallation facilities. As a first application of the half-life value found in this work, the excitation functions for the production of 154Dy in proton-irradiated Ta, Pb, and W targets were re-evaluated, which found to be in agreement with theoretical calculations.

A Survey Data Approach for Determining the Probability Values of Vehicle-to-Grid Service Provision

One of the key aspects of vehicle-to-grid technology (V2G) is the analysis of uncertainty in electric vehicle user behavior. Correct estimation of the amount of available energy from electric vehicles that are expected to provide ancillary services to the electricity system operator or to secure the end user’s demand is essential to design these services in an appropriate way. Therefore, it is necessary to analyze the probabilities of V2G service performance for different scenarios. This paper presents the author’s approach to determining the values of V2G service provision probabilities using survey data. It was found that estimating these values using simulation and forecasting tools makes sense when the coefficients resulting from survey responses are used as initial data. Thus, the paper also presents the results of the surveys that were conducted. As the results from the simulations show, the values of the probabilities of V2G services are not high, which should induce future operators of V2G services to offer a beneficial product for the customer.

Clustering in massive neutrino cosmologies via Eulerian Perturbation Theory

We introduce an Eulerian Perturbation Theory to study the clustering of tracers for cosmologies in the presence of massive neutrinos. Our approach is based on mapping recently-obtained Lagrangian Perturbation Theory results to the Eulerian framework. We add Effective Field Theory counterterms, IR-resummations and a biasing scheme to compute the one-loop redshift-space power spectrum. To assess our predictions, we compare the power spectrum multipoles against synthetic halo catalogues from the QUIJOTE simulations, finding excellent agreement on scales k ≲ 0.25 h Mpc-1. One can obtain the same fitting accuracy using higher wave-numbers, but then the theory fails to give a correct estimation of the linear bias parameter. We further discuss the implications for the tree-level bispectrum. Finally, calculating loop corrections is computationally costly, hence we derive an accurate approximation wherein we retain only the main features of the kernels, as produced by changes to the growth rate. As a result, we show how FFTLog methods can be used to further accelerate the loop computations with these reduced kernels.

Natural isotope correction improves analysis of protein modification dynamics

Stable isotope labelling in combination with high-resolution mass spectrometry approaches are increasingly used to analyze both metabolite and protein modification dynamics. To enable correct estimation of the resulting dynamics, it is critical to correct the measured values for naturally occurring stable isotopes, a process commonly called isotopologue correction or deconvolution. While the importance of isotopologue correction is well recognized in metabolomics, it has received far less attention in proteomics approaches. Although several tools exist that enable isotopologue correction of mass spectrometry data, the majority is tailored for the analysis of low molecular weight metabolites. We here present PICor which has been developed for isotopologue correction of complex isotope labelling experiments in proteomics or metabolomics and demonstrate the importance of appropriate correction for accurate determination of protein modifications dynamics, using histone acetylation as an example.

Kuyper’s Spirituality in Its Calvin-Context

This article explores Abraham Kuyper’s spirituality by comparing it to that of John Calvin. Calvin’s Institutes exhibits three dimensions of his spirituality in the context of the mystical union with Christ, namely, the affective character of this union, its effects and its significance for a correct estimation of the world. By comparison, Kuyper put a greater emphasis on the importance of the affections in mystical union because he gives more weight to the regenerated life. This focus also coheres with Kuyper’s more optimistic approach to the Christian life, which contrasts with Calvin’s emphasis on the need for daily justification. In Kuyper’s approach the indwelling of the Spirit represents the union between heaven and earth, while Calvin stresses that God’s future kingdom is beyond the here and now.