Structured Learning for Non-Smooth Ranking losses

Learning to rank from relevance judgment is an active research area. Itemwise score regression, pairwise preference satisfaction, and listwise structured learning are the major techniques in use. Listwise structured learning has been applied recently to optimize important non-decomposable ranking criteria like AUC (area under ROC curve) and MAP(mean average precision). We propose new, almost-lineartime algorithms to optimize for two other criteria widely used to evaluate search systems: MRR (mean reciprocal rank) and NDCG (normalized discounted cumulative gain)in the max-margin structured learning framework. We also demonstrate that, for different ranking criteria, one may need to use different feature maps. Search applications should not be optimized in favor of a single criterion, because they need to cater to a variety of queries. E.g., MRR is best for navigational queries, while NDCG is best for informational queries. A key contribution of this paper is to fold multiple ranking loss functions into a multi-criteria max-margin optimization.The result is a single, robust ranking model that is close to the best accuracy of learners trained on individual criteria. In fact, experiments over the popular LETOR and TREC data sets show that, contrary to conventional wisdom, a test criterion is often not best served by training with the same individual criterion.

Finding optimal alternatives based on efficient comparative preference inference

Choosing the right or the best option is often a demanding and challenging task for the user (e.g., a customer in an online retailer) when there are many available alternatives. In fact, the user rarely knows which offering will provide the highest value. To reduce the complexity of the choice process, automated recommender systems generate personalized recommendations. These recommendations take into account the preferences collected from the user in an explicit (e.g., letting users express their opinion about items) or implicit (e.g., studying some behavioral features) way. Such systems are widespread; research indicates that they increase the customers' satisfaction and lead to higher sales. Preference handling is one of the core issues in the design of every recommender system. This kind of system often aims at guiding users in a personalized way to interesting or useful options in a large space of possible options. Therefore, it is important for them to catch and model the user's preferences as accurately as possible. In this thesis, we develop a comparative preference-based user model to represent the user's preferences in conversational recommender systems. This type of user model allows the recommender system to capture several preference nuances from the user's feedback. We show that, when applied to conversational recommender systems, the comparative preference-based model is able to guide the user towards the best option while the system is interacting with her. We empirically test and validate the suitability and the practical computational aspects of the comparative preference-based user model and the related preference relations by comparing them to a sum of weights-based user model and the related preference relations. Product configuration, scheduling a meeting and the construction of autonomous agents are among several artificial intelligence tasks that involve a process of constrained optimization, that is, optimization of behavior or options subject to given constraints with regards to a set of preferences. When solving a constrained optimization problem, pruning techniques, such as the branch and bound technique, point at directing the search towards the best assignments, thus allowing the bounding functions to prune more branches in the search tree. Several constrained optimization problems may exhibit dominance relations. These dominance relations can be particularly useful in constrained optimization problems as they can instigate new ways (rules) of pruning non optimal solutions. Such pruning methods can achieve dramatic reductions in the search space while looking for optimal solutions. A number of constrained optimization problems can model the user's preferences using the comparative preferences. In this thesis, we develop a set of pruning rules used in the branch and bound technique to efficiently solve this kind of optimization problem. More specifically, we show how to generate newly defined pruning rules from a dominance algorithm that refers to a set of comparative preferences. These rules include pruning approaches (and combinations of them) which can drastically prune the search space. They mainly reduce the number of (expensive) pairwise comparisons performed during the search while guiding constrained optimization algorithms to find optimal solutions. Our experimental results show that the pruning rules that we have developed and their different combinations have varying impact on the performance of the branch and bound technique.

Bactrocera dorsalis Preference for and performance on two mango varieties at three stages of ripeness

Most tropical fruit flies only lay into mature fruit, but a small number can also oviposit into unripe fruit. Little is known about the link between adult oviposition preference and offspring performance in such situations. In this study we examine the influence of different ripening stages of two mango Mangifera indica L. (Anacardiaceae) varieties on the preference and performance of the Oriental fruit fly, Bactrocera dorsalis (Hendel) (Diptera: Tephritidae), a fly known to be able to develop in unripe fruit. Work was carried out as a series of laboratory-based choice and no-choice oviposition experiments and larval growth trials. In oviposition choice trials, female B. dorsalis demonstrated a preference for ripe fruit of mango variety Namdorkmai over variety Oakrong, but generally the dependent variable most influencing oviposition results was fruit ripening stage. Ripe and fully-ripe mangoes were most preferred for oviposition by B. dorsalis. In contrast, unripe mango was infrequently used by ovipositing females, particularly in choice trials. Consistent with the results of oviposition preference, ripe and fully-ripe mangoes were also best for offspring survival, with a higher percentage of larval survival to pupation and shorter development times in comparison to unripe mango. Changes in Total Soluble Solids, TSS, and skin toughness correlate with changing host use across the ripening stages. Regardless of the mango variety or ripeness stage, B. dorsalis had difficulty penetrating the pericarp of our experimental fruit. Larval survival was also often poor. We discuss the possibility that there may be differences in the ability of laboratory and wild flies to penetrate fruit for oviposition, or that in the field flies more regularly utilize natural fruit wounds as oviposition sites.

Health, endogenous time preference, and inequality : a political economy perspective

We study an overlapping-generations model in which agents' mortality risks, and consequently impatience, are endogenously determined by private and public investment in health care. Revenues allocated for public health care arc determined by a voting process. We find that the degree of substitutability between public and private health expenditures matters for macroeconomic outcomes of the model. Higher substitutability implies a “crowding-out" effect, which in turn impacts adversely on morality risks and impatience leading to lower public expenditures on health care in the political equilibrium. Consequently, higher substitutability is associated with greater polarization in wealth, and long-run distributions that are bimodal.

Patient preference for falls prevention in hospitals revealed through willingness-to-pay, contingent valuation survey

Rationale, aims and objectives: Patient preference for interventions aimed at preventing in-hospital falls has not previously been investigated. This study aims to contrast the amount patients are willing to pay to prevent falls through six intervention approaches. ----- ----- Methods: This was a cross-sectional willingness-to-pay (WTP), contingent valuation survey conducted among hospital inpatients (n = 125) during their first week on a geriatric rehabilitation unit in Queensland, Australia. Contingent valuation scenarios were constructed for six falls prevention interventions: a falls consultation, an exercise programme, a face-to-face education programme, a booklet and video education programme, hip protectors and a targeted, multifactorial intervention programme. The benefit to participants in terms of reduction in risk of falls was held constant (30% risk reduction) within each scenario. ----- ----- Results: Participants valued the targeted, multifactorial intervention programme the highest [mean WTP (95% CI): $(AUD)268 ($240, $296)], followed by the falls consultation [$215 ($196, $234)], exercise [$174 ($156, $191)], face-to-face education [$164 ($146, $182)], hip protector [$74 ($62, $87)] and booklet and video education interventions [$68 ($57, $80)]. A ‘cost of provision’ bias was identified, which adversely affected the valuation of the booklet and video education intervention. ----- ----- Conclusion: There may be considerable indirect and intangible costs associated with interventions to prevent falls in hospitals that can substantially affect patient preferences. These costs could substantially influence the ability of these interventions to generate a net benefit in a cost–benefit analysis.

Preference among four Bactrocera species (Diptera: Tephritidae) by Diachasmimorpha kraussii (Fullaway) (Hymenoptera: Braconidae)

Diachasmimorpha kraussii is an endoparasitoid of larval dacine fruit flies. To date the only host preference study done on D. kraussii has used fruit flies from outside its native range (Australia, Papua New Guinea, Solomon Islands). In contrast, this paper investigates host preference for four fly species (Bactrocera cacuminata, B. cucumis, B. jarvisi and B. tryoni) which occur sympatrically with the wasp in the Australian component of the native range. Diachasmimorpha kraussii oviposition preference, host suitability (parasitism rate, number of progeny, sex ratio), and offspring performance measures (body length, hind tibial length, developmental time) were investigated with respect to the four fly species in the laboratory in both no-choice and choice situations. The parasitoid accepted all four fruit fly species for oviposition in both no-choice and choice tests; however, adult wasps only emerged from B. jarvisi and B. tryoni. Through dissection, it was demonstrated that parasitoid eggs were encapsulated in both B. cacuminata and B. cucumis. Between the two suitable hosts, measurements of oviposition preference, host suitability and offspring performance measurements either did not vary significantly, or varied in an inconsistent manner. Based on our results, and a related study by other authors, we conclude that D. krausii, at the point of oviposition, cannot discriminate between physiologically suitable and unsuitable hosts.

Team role preference and online role play

Role play approaches have been used in online environments in an effort to create a mix of contested ideas and to promote participant engagement. While it is recognised that there is an aspect of ‘fun’ associated with role play there is a need to understand role assignment more rigorously than simply levels of reported participation and enjoyment. It is the contention of this paper that individuals are unlikely to be able to authentically play a role and, that in fact, there may be little purpose to contrived roles. Additionally, the literature has widely reported that personality factors, such as introversion and extroversion continue to be of significance in the way that individuals contribute in online contexts. The findings in the study reported in this paper confirm that introversion and extroversion do, indeed, play a role in the way individuals contribute in online environments. Thus, this paper argues that an active consideration needs to be given to individuals preferred (or natural) way of working even where use is made of online role play.

How do level of experience, purpose for riding, and preference for facilities affect location of riding?

Characteristics of the road infrastructure affect both the popularity of bicycling and its safety, but comparisons of the safety performance of infrastructure may be confounded by differences in the profiles of cyclists who use them. Data from a survey of 2,532 adult bicycle riders in Queensland, Australia, demonstrated that many riders rode reluctantly in particular locations and that preference for riding location was influenced by degree of experience and riding purpose. Most riders rode most often and furthest per week on urban roads, but approximately one-third of all riders (and more new riders) rode there reluctantly. Almost two-thirds of riders rode on bicycle paths, most by choice, not reluctantly. New riders rode proportionally more on bicycle paths, but continuing riders rode further in absolute terms. Utilitarian riders were more likely to ride on bicycle paths than social and fitness riders and almost all of this riding was by choice. Fitness riders were more reluctant in their use of bicycle paths, but still most of their use was by choice. One-third of the respondents reported riding on the sidewalk (legal in Queensland), with approximately two-thirds doing so reluctantly. The frequency and distance ridden on the sidewalk was less than for urban roads and bicycle paths. Sidewalks and bicycle paths were important facilities for both inexperienced and experienced riders and for utilitarian riding, especially when urban roads were considered a poor choice for cycling.