Ranking Task Research Articles

Beyond majority: Label ranking ensembles based on voting rules

Label ranking is a machine learning task that deals with mapping an instance to a ranking of labels, representing the labels’ ordered relevance to the instance. Three recent studies have suggested the use of ensembles to improve the performance of simple label ranking models. However, none of them has explicitly inspected the question of how should the results obtained by simple models be aggregated into a single combined output. While classification tasks and regression tasks typically employ trivial aggregation techniques (i.e., majority voting and averaging, respectively), the case of label ranking tasks is not straightforward. The contribution of this paper is twofold. First, we propose to apply voting rules, typically used in the field of social choice, as the aggregation technique for label ranking ensembles. Our evaluation reveals that there is no single rule that consistently outperforms all other voting rules, and that under different settings different voting rules perform the best. Second, we propose a novel aggregation method for label ranking ensembles that learns the best voting rule to be used in a given setting. An extensive evaluation of the proposed method on semi-synthetic as well as real-world datasets shows that it obtains prediction performance that is significantly higher than that of the aggregation techniques currently used by state-of-the-art label ranking ensembles.

Social acceptance of distributed energy systems in Swiss, German, and Austrian energy transitions

Distributed energy systems (DESs) on a local scale constitute a promising niche to leverage renewable energy provision. These DESs (e.g., micro-cogeneration, multi-energy hubs) integrate renewable sources, combined heat and power production, various methods of energy conversion and storage, and active demand-side management. Since the broader uptake of such systems is still at an early stage, research about the adoption potentials of DESs in existing neighborhood contexts is scarce. Given their potential to contribute to energy transitions, it is relevant to gain a better understanding of the conditions under which these systems are diffused. Based on a quantitative study, we investigated the perceived responsibilities and the intended technology uptake of DESs among different societal groups in Switzerland, Germany, and Austria. We analyzed a sample of 2104 survey participants who were exposed to ranking tasks and a framing experiment and were asked questions about both opportunities and challenges in implementing DESs. Our results show an openness to engage with DESs; on average, the opportunities in applying DESs are rated higher than the challenges. However, in all three countries, the participants place the responsibility for the energy system transitions on the national government and large energy supply utilities. Municipalities and households are not yet considered relevant innovators. Moreover, the support of DESs is found to be independent from framing the DES implementation in relation to different scales of rationale (global climate change, national energy independence, or local autarky). Our findings illustrate not only the agents’ attitudes toward DESs but also the structural path dependencies involving the implementation of distributed energy technologies. Our study implies the need to address the willingness to invest in new distributed energy infrastructures at the household level in the future. We discuss the necessity of identifying barriers to and drivers of technology applications ex-ante.

Quantum protocols for secure multi-party summation

Quantum secure multi-party summation is a kind of privacy-preserving summation whereby multiple mutually distrustful parties can securely compute the summation of their secret data, which can be utilized to execute many tasks of quantum secure multi-party computation, such as quantum anonymous surveying. In this paper, we present two quantum secure multi-party summation protocols, both of which allow multiple mutually distrustful parties to securely compute the summation of their secret data, where the dataset of the summation is supposed to be a set of nonnegative integers. Our protocols have two main common features. One is the assumption of a semi-honest third party who helps multiple parties to perform the summation computations and announces the results, at the same time he is allowed to misbehave on his own, but cannot conspire with any party. Another is the use of the entanglement swapping of d-level cat states and Bell states to maintain privacy, and pass information between each party and the semi-honest third party. We analyze the success probability of our two protocols, respectively. In addition, we generalize the use of our second protocol to execute the tasks of quantum anonymous ranking and quantum anonymous voting. What’s more, we show that our protocols can resist various attacks from both outside attackers and inside participants.

Supervised and extended restart in random walks for ranking and link prediction in networks.

Given a real-world graph, how can we measure relevance scores for ranking and link prediction? Random walk with restart (RWR) provides an excellent measure for this and has been applied to various applications such as friend recommendation, community detection, anomaly detection, etc. However, RWR suffers from two problems: 1) using the same restart probability for all the nodes limits the expressiveness of random walk, and 2) the restart probability needs to be manually chosen for each application without theoretical justification. We have two main contributions in this paper. First, we propose Random Walk with Extended Restart (RWER), a random walk based measure which improves the expressiveness of random walks by using a distinct restart probability for each node. The improved expressiveness leads to superior accuracy for ranking and link prediction. Second, we propose SuRe (Supervised Restart for RWER), an algorithm for learning the restart probabilities of RWER from a given graph. SuRe eliminates the need to heuristically and manually select the restart parameter for RWER. Extensive experiments show that our proposed method provides the best performance for ranking and link prediction tasks.

Task effects determine whether recognition memory is mediated discretely or continuously.

How recognition memory is mediated has been of interest to researchers for decades. But the apparent consensus implicating continuous mediation has been challenged. McAdoo, Key, and Gronlund (Journal of Experimental Psychology: Learning, Memory, and Cognition,2018. Advanced online publication) demonstrated that recognition memory can be mediated by either discrete or continuous evidence, depending on target-filler similarity. The present paper expands on this research by showing that different recognition tasks also can be mediated by different evidence. Specifically, recognition memory was mediated by continuous evidence in a ranking task, but by discrete evidence in a confidence-rating task. We posit that participants utilize a control process that induces a reliance on discrete or continuous evidence as a function of efficiency (Malmberg, 2008) to suit the demands of the task.

Ranking and Selection

We introduce a new sampling scheme for selecting the best alternative out of a given set of systems that are evaluated with respect to their expected performances. We assume that the systems are simulated on a computer and that a joint observation of all systems has a multivariate normal distribution with unknown mean and unknown covariance matrix. In particular, the observations of the systems may be stochastically dependent as is the case if common random numbers are used for simulation. In each iteration of the algorithm, we allocate a fixed budget of simulation runs to the alternatives. We use a Bayesian set-up with a noninformative prior distribution and derive a new closed-form approximation for the posterior distributions that allows provision of a lower bound for the posterior probability of a correct selection (PCS). Iterations are continued until this lower bound is greater than 1−α for a given α. We also introduce a new allocation strategy that allocates the available budget according to posterior error probabilities. Our procedure needs no additional prior parameters and can cope with different types of ranking and selection tasks. Our numerical experiments show that our strategy is superior to other procedures from the literature, namely, KN ++ and P luck . In all of our test scenarios, these procedures needed more observation and/or had an empirical PCS below the required 1−α. Our procedure always had its empirical PCS above 1−α, underlining the practicability of our approximation of the posterior distribution.

Blockwise Based Detection of Local Defects

Print quality is an important criterion for a printer’s performance. The detection, classification, and assessment of printing defects can reflect the printer’s working status and help to locate mechanical problems inside. To handle all these questions, an efficient algorithm is needed to replace the traditionally visual checking method. In this paper, we focus on pages with local defects including gray spots and solid spots. We propose a coarse-to-fine method to detect local defects in a block-wise manner, and aggregate the blockwise attributes to generate the feature vector of the whole test page for a further ranking task. In the detection part, we first select candidate regions by thresholding a single feature. Then more detailed features of candidate blocks are calculated and sent to a decision tree that is previously trained on our training dataset. The final result is given by the decision tree model to control the false alarm rate while maintaining the required miss rate. Our algorithm is proved to be effective in detecting and classifying local defects compared with previous methods.

Hardness Perception Based on Dynamic Stiffness in Tapping.

A human can judge the hardness of an object based on the damped natural vibration caused by tapping the surface of the object using a fingertip. In this study, we investigated the influence of the dynamic characteristics of vibrations on the hardness perceived by tapping. Subjectively reported hardness values were related to the dynamic stiffness of several objects. The dynamic stiffness, which characterizes the impulsive response of an object, was acquired across the 40–1,000 Hz frequency range for cuboids of 14 types of materials by administering a hammering test. We performed two psychophysical experiments—a ranking task and a magnitude-estimation tasks—wherein participants rated the perceived hardness of each block by tapping it with a finger. We found that the perceptual effect of dynamic stiffness depends on the frequency. Its effect displayed a peak around 300 Hz and decreased or disappeared at higher frequencies, at which human perceptual capabilities are limited. The acquired results help design hardness experienced by products.

DeepRank: improving unsupervised node ranking via link discovery

This paper proposes an unsupervised node-ranking model that considers not only the attributes of nodes in a graph but also the incompleteness of the graph structure. We formulate the unsupervised ranking task into an optimization task and propose a deep neural network (DNN) structure to solve it. The rich representation capability of the DNN structure together with a novel design of the objectives allow the proposed model to significantly outperform the state-of-the-art ranking solutions.

Enhancing Question Retrieval in Community Question Answering Using Word Embeddings

Community Question Answering (CQA) services have evolved into a popular way of online information seeking, where users can interact and exchange knowledge in the form of questions and answers. In this paper, we study the problem of finding historical questions that are semantically equivalent to the queried ones, assuming that the answers to the similar questions should also answer the new ones. The major challenge of question retrieval is the word mismatch problem between questions, as users can formulate the same question using different wording. Most existing methods measure the similarity between questions based on the bag-of-words (BOWs) representation capturing no semantics between words. Therefore, this study proposes to use word embeddings, which can capture semantic and syntactic information from contexts, to vectorize the questions. The questions are clustered using Kmeans to speed up the search and ranking tasks. The similarity between the questions is measured using cosine similarity based on their weighted continuous valued vectors. We run our experiments on real world data set from Yahoo! Answers in English and Arabic to show the efficiency and generality of our proposed method.

Listwise View Ranking for Image Cropping

Rank-based Learning with deep neural network has been widely used for image cropping. However, the performance of ranking-based methods is often poor and this is mainly due to two reasons: 1) image cropping is a listwise ranking task rather than pairwise comparison; 2) the rescaling caused by pooling layer and the deformation in view generation damage the performance of composition learning. In this paper, we develop a novel model to overcome these problems. To address the first problem, we formulate the image cropping as a listwise ranking problem to find the best view composition. For the second problem, a refined view sampling (called RoIRefine) is proposed to extract refined feature maps for candidate view generation. Given a series of candidate views, the proposed model learns the Top-1 probability distribution of views and picks up the best one. By integrating refined sampling and listwise ranking, the proposed network called LVRN achieves the state-of-the-art performance both in accuracy and speed.

Predictions and constructing explanations: an investigation into introductory chemistry students’ understanding of structure–property relationships

The relationship between chemical structure and physical and chemical properties is essential to chemistry. Studies have shown that students have difficulty using structural representations to predict properties, which is not surprising because of the sequence of inferences that are required for sense-making. However, obtaining a nuanced model of students’ understanding depends on how information is elicited. This study investigated how the phrasing of the question prompt may elicit students’ understanding of structure–property relationships. Students were given a two-part assessment: (1) four multiple-choice questions assessing students’ self-reported abilities to predict structure–property relationships, and (2) three questions requiring students to predict, argue, and explain a boiling point trend. Two groups of students were selected to determine the sensitivity of the instrument (one with less explicit instruction of structure–property relationships and one with more explicit instruction). We found that Part I of the assessment was able to differentiate between these two groups of students. The group with more explicit instruction was further analyzed to determine how their prediction on a boiling point task connected to their arguments and explanations of the phenomenon. Even though 64% of students answered the boiling point ranking task correctly, the students typically provided less complete arguments as to why that structure had a higher boiling point. However, after scaffolding (i.e., providing relevant information for the phenomenon) and asking for an explanation, students’ responses began to include a much more mechanistic understanding, suggesting that having students provide explanations instead of constructing an argument would display their reasoning at a deeper level.

Tripolar Stimulation Improves Polyphonic Pitch Detection in Cochlear Implant Users.

Cochlear implant (CI) users struggle with pitch perception, particularly for polyphonic stimuli. Tripolar (TP) stimulation has been proposed as a way to mitigate the broad spread of neural excitation observed in traditional monopolar (MP) stimulation, thereby potentially improving perception of polyphony. Prospective cohort study. Tertiary academic center. Eleven postlingually deafened adults with Advanced Bionics HiRes 90K CIs. We performed pitch ranking and polyphonic pitch detection testing under MP and TP configurations. To assess pitch ranking, users were asked to identify the higher pitch between two notes. In polyphonic pitch detection, users were asked to distinguish between single-pitch tones and two-pitch tones. Two-pitch stimuli consisted of one pitch of three base frequencies (392, 523, 740 Hz) and a second pitch between 1 and 12 semitones above the base frequency. Pitch performance was analyzed as a function of current delivery mode (tripolar vs. monopolar), with smaller semitone interval pitch resolution indicating better performance. In pitch ranking tasks, TP configuration did not confer an advantage over MP stimulation. In polyphonic perception, however, tripolar stimulation improved performance in lower frequencies and resulted in statistically significant (p < 0.05) improvement at the highest base frequency, 740 Hz. These data suggest that TP configuration may confer an advantage in the perception of polyphonic pitch, which may not be observed in monophonic pitch ranking tasks. Since music is typically polyphonic, such data offer approaches toward improving perception of real-world musical stimuli.

One Best Way? Leader Behavior and Different Aspects of Team Performance

Abstract. In two experiments, we tested differential effects of transformational, empowering, and task-oriented leader behavior on various aspects of team performance (quantity, quality, independence, subjective group performance) using three different tasks. In Experiment 1, 60 three-person teams completed a rank-order task and in Experiment 2, 54 three-person teams completed a construction and an information-search task. In both experiments, a videotaped team leader displayed transformational, empowering, or task-oriented leadership. Results suggested that leadership effectiveness may be a function of group outcome. Only for the tasks of Experiment 2 did transformational leadership result in superior outcomes, specifically regarding productivity and originality. Empowering leadership, in turn, yielded superior outcomes in terms of leader-independent thinking, while task-oriented leadership resulted in the highest subjective group performance ratings – except for the ranking task in which satisfaction ratings were highest for transformational leadership. Implications for research and practice are discussed.

A comparison between preference judgments of curvature and sharpness in architectural façades

ABSTRACTCan curvature drive preference, perceived familiarity, complexity, stability and approachability for architectural façades? In this study, we generated four versions of the same reference building, varying only the amount of curvature introduced in the façade. Participants’ judgments were measured using three experimental methodologies. Multidimensional scaling on forced choices showed that the curved façade was the most preferred. Multidimensional unfolding on ranking task showed that the majority of participants expressed higher preferences for the curved façade compared to the sharp-angled and rectilinear ones. Ratings on different psychological variables provided supporting evidence for curvature significantly influencing liking and approachability judgments. Results from image analyses –using a dynamical model of the visual cortex and a model that characterizes discomfort in terms of adherence to the statistics of natural images – matched behavioural data. We discuss the implications of the findings on our understanding of human preferences, which are intrinsically dynamic and influenced by context and experience.

The Relative Contribution of Jawbone and Cheekbone Prominence, Eyebrow Thickness, Eye Size, and Face Length to Evaluations of Facial Masculinity and Attractiveness: A Conjoint Data-Driven Approach.

Recent work demonstrates the methodological rigor of a type of data-driven analysis (i.e., conjoint analysis; CA), which accounts for the relative contribution of different facial morphological cues to interpersonal perceptions of romantic partner quality. This study extends this literature by using a conjoint face ranking task to predict the relative contribution of five sexually dimorphic facial shape features (jawbone and cheekbone prominence, eyebrow thickness, eye size, face length) to participants’ (N = 922) perceptions of facial attractiveness and sex-typicality (i.e., masculinity/femininity). For overall partner attractiveness, eyebrow thickness and jawbone prominence were relatively more salient than cheekbone prominence and eye size. Interestingly, masculinized (i.e., thicker) eyebrows were marginally more attractive for female than male faces, particularly within a long-term mating context. Masculinized jawbone prominence was more attractive for male than female faces, and feminized jawbone prominence was more attractive for female than male faces. For perceptions of masculinity, eyebrow thickness, jawbone prominence, and facial height were relatively more salient than cheekbone prominence and eye size, although facial height was more important for female than male faces, and jawbone prominence was marginally more important for male than female faces. These findings highlight the prominence of eyebrows, the jawline, and facial height during perception of facial attractiveness and masculinity – though it should be noted that many of these differences were small to moderate in effect size. Findings are interpreted in the context of prior research, and future directions for studying why these facial traits exhibit superior signaling capacity are discussed.

“You’re a great driver/Your driving scares me”: Reactions of older drivers to family comments

Background and objectivesAge related changes often compromise the ability to drive safely. Some elders report that they listen to others’ advice, but there have been no studies that have examined how much influence families have on their driving decisions. This study fills this gap by exploring older drivers’ perceptions of how family members affect their driving decisions. Research design and methodsA quasi-experimental ranking task was used to examine the expected influence that 48 hypothetical statements, drawn from related literature, have on a convenience sample of 37 adults, age 70–96 years. Analysis was conducted by gender and by habit strength for driving. ResultsFour domains of influence were identified from the literature, representing family aims to influence the elder to: (1) continue driving, (2) modify driving, (3) reduce driving, and (4) stop driving. Statements aimed at supporting the driver to continued driving (62%) and to reduce driving (62%) were found to have the most influence, with slightly less support for statements aimed to influence them to stop driving (58%) or modify their driving (52%). Women rated statements supporting them to continue driving as having the highest influence on them, whereas men reported that they were more influenced by statements encouraging them to reduce their driving. In addition, participants with high habit strength for driving were less likely to rate family statements as having a high influence on them. Discussion and implicationsThis study contributes new understanding of the influence of families on the driving decisions of older adults. These findings may be used to support families, in addition to informing the development of effective programs aimed at older driver safety.

Learning to rank with relational graph and pointwise constraint for cross-modal retrieval

Cross-modal retrieval (i.e., image–query–text or text–query–image) is a hot research topic for multimedia information retrieval, but the heterogeneity gap between different modalities generates a critical challenge for multimodal data. Some researchers regard the cross-modal retrieval as a leaning to rank task, and they usually consider to measure similarity between two different modalities in the embedding shared subspace. However, previous methods almost pay more attention to construct a discriminative objective function to optimize common space, ignoring to exploit correlation between the single modality. In this paper, we consider the cross-modal retrieval task, from the perspective of optimizing ranking model, as a listwise ranking problem, and propose a novel method called learning to rank with relational graph and pointwise constraint ( $$ {\text{LR}}^{2} {\text{GP}} $$ ). In $$ {\text{LR}}^{2} {\text{GP}} $$ , we first propose a discriminative ranking model, which makes use of the relation between the single modality to improve ranking performance so as to learn an optimal embedding common subspace. Then, a pointwise constraint is introduced in the low-dimension embedding subspace to make up for the real loss in the training phase since listwise method introduced merely considers directly optimize latent permutation from the perspective of the overall. Finally, a dynamic interpolation algorithm, which gradually transits from pointwise and pairwise to listwise learning, is selected to deal with the problem of fusion of loss function reasonable. Experiments on the benchmark datasets about Wikipedia and Pascal demonstrate the effectiveness for proposed method.

Applying Collaborative Ranking Tasks to Improve Students’ Concept Mastery and Generic Science Skills

The lack of students concept mastery and generic science skills underlying this research. The purpose of this research was to improve the students’ concept mastery and generic science skill by applying the collaborative ranking-task model in Earth and Space Science learning. This type of research was a classroom action research, conducted in two cycles. The data were collected by observation technique, test, and questionnaire. The study was conducted on physics students amounted to 25 people (10 male and 15 female). The success of the study was determined by the grade point average which should be at least 70. The obtained data were analyzed descriptively. The results showed that there was an increase in the students’ concept mastery starting from cycle 1 to cycle 2 ( x1= 50, s1 = 11.4 in the ‘deficient’ category; and x2= 64, s2 = 8.3 in the ‘sufficient’ category). The generic science skills were numerically increased from cycle 1 to cycle 2 (x1= 58.4, s1 = 13.3; x2= 62.8, s2 =10.2; in the ‘moderate’category). The developed indicators of generic science skills included indirect observation, logic inference, and concept building. The mean score of student response was x = 58.6; s = 6.7 which was in the positive category.

Analysis on the potential of an EA–surrogate modelling tandem for deep learning parametrization: an example for cancer classification from medical images

The paper introduces a novel modality to efficiently tune the convolutional layers of a deep neural network (CNN) and an approach to also rank the importance of the involved hyperparameters. Evolutionary algorithms (EA) offer a flexible solution to this twofold issue, while the expensive simulations of the deep learner with the generated configurations are resolved by surrogate modelling. Three models have been used and evaluated as surrogates: random forests (RF), support vector machines (SVM) and Kriging. Sample convolutional configurations are generated by Latin hypercube sampling and have attached computed accuracy outcomes from real CNN runs. For the hyperparameter estimation task, the fitness of an individual from the EA associated with a surrogate model is subsequently derived from the CNN accuracy estimation on those variable values. With respect to the ranking and variable selection task, RF includes implicit variable selection, the SVM can be straightforwardly supported by a second EA, and Kriging offers a ranking based on the corresponding θ values. The estimated accuracy of the found hyperparameter values is compared with the true validation accuracy, and they are next used for the prediction on the test cases. The ranking of the variables for each of the three surrogate models is compared, and their influence is also revealed by response surface methodology. The experimental testing of the proposed EA–surrogate approaches is conducted on a real-world scenario of histopathological image interpretation in colorectal cancer diagnosis.