A new hierarchical ranking aggregation method

The problem of interpreting or aggregating multiple rankings is common to many real-world applications. Perhaps the simplest and most common approach is a weighted rank aggregation, wherein a (convex) weight is applied to each input ranking and then ordered. This paper describes a new tool for visualizing and displaying ranking information for the weighted rank aggregation method. Traditionally, the aim of rank aggregation is to summarize the information from the input rankings and provide one final ranking that hopefully represents a more accurate or truthful result than any one input ranking. While such an aggregated ranking is, and clearly has been, useful to many applications, it also obscures information. In this paper, we show the wealth of information that is available for the weighted rank aggregation problem due to its structure. We apply weight set decomposition to the set of convex multipliers, study the properties useful for understanding this decomposition, and visualize the indifference regions. This methodology reveals information–that is otherwise collapsed by the aggregated ranking–into a useful, interpretable, and intuitive decision support tool. Included are multiple illustrative examples, along with heuristic and exact algorithms for computing the weight set decomposition.

Normally, statistical methods are used to generate rankings for genes in terms of their ability to distinguish between normal and malignant tumors from a gene expression dataset. However, different statistical methods yield different ranks for same gene and there is no universally accepted method for ranking. Therefore rank aggregation is required to find the overall ranking of the set of genes. There are various rank aggregation methods in the existing literature to integrate the rankings produced by various statistical tests. Moreover, the problem of integration of some partial rankings, containing unequal numbers of genes, is more challenging. In this article, a multiobjective genetic algorithm based rank aggregation method is proposed to integrate some partial rankings in an unbiased way. The first objective is to minimize the total distance from the reference ranking to the input rankings. For distance calculation, the Scaled Footrule Distance is used. The second objective is to minimize the standard deviation among those distances in order to avoid bias toward a particular input ranking. The proposed method is applied on some real-life microarray gene expression datasets, and the performance of it is compared with that of several existing rank aggregation techniques with respect to accuracy and the AUC (Area under ROC curve) value. Again, for real-life datasets, accuracy is plotted for visual comparison.

This paper presents novel approaches for combining re-ranking and rank aggregation methods aiming at improving the effectiveness of Content-Based Image Retrieval (CBIR) systems. Given a query image as input, CBIR systems retrieve the most similar images in a collection by taking into account image visual properties. In this scenario, accurately ranking collection images is of great relevance. Aiming at improving the effectiveness of CBIR systems, re-ranking and rank aggregation algorithms have been proposed. However, different re-ranking and rank aggregation approaches, applied to different image descriptors, may produce different and complementary image rankings. In this paper, we present four novel approaches for combining these rankings aiming at obtaining more effective results. Several experiments were conducted involving shape, color, and texture descriptors. The proposed approaches are also evaluated on multimodal retrieval tasks, considering visual and textual descriptors. Experimental results demonstrate that our approaches can improve significantly the effectiveness of image retrieval systems.

The analytic hierarchy process (AHP) is a method to help people make a complex decision by analyzing and synthesizing multiple criteria for the decision objective in a hierarchy. We adopt the AHP as a new strategy for the semantic concept detection (SCD) in video so that multiple factors involved in SCD, including multiple modalities and relating concepts, can be hierarchically analyzed and synthesized. In this paper, we first explain, by an example, why and how the SCD problem can be analyzed by the AHP. Then, following the idea of the AHP, we develop a new rank aggregation (RA) method, called AHP-RA. Experimental results of RA for SCD in video show the effectiveness of this method.

Aggregating multiple rankings in a database is an important task well studied by the database community. High-stakes application domains include hiring, lending, and education where multiple decision makers rank candidates and their input is then combined into an overall consensus ranking. While state-of-art aggregation methods succeed in giving equal say to each decision maker, to date no methods ensure fair representation for groups of candidates being ranked, who risk being impacted by discriminatory bias. We present the first solution to this open problem of guaranteeing fairness for disadvantaged groups of candidates, while still producing a good consensus of the base rankings. We design a family of exact fair algorithms achieving optimality for fair rank aggregation. We also develop approximate methods achieving fairness with guaranteed minimal approximation error scaling to millions of candidates in the rankings. A comparative study evaluates our proposed methods, revealing trade-offs between aggregation accuracy and different degrees of unfair bias in a rich variety of rank aggregation scenarios. Our real-world case study demonstrates that our solutions mitigate unfair bias using real-world data.

Learning the true rank ordering among objects by aggregating a set of expert opinion rank order lists is an important and ubiquitous problem in many applications ranging from social choice theory to recommendation systems and search aggregation. We study the problem of unsupervised rank aggregation where no ground truth ordering information in available, neither about the true preference ordering among any set of objects nor about the quality of individual rank lists. Aggregating the often inconsistent and poor quality rank lists in such an unsupervised manner is a highly challenging problem, and standard consensus-based methods fall short in terms of both quality and scalability. In this manuscript we propose a novel framework to bypass these issues by using object attributes to augment the standard rank aggregation framework. We design algorithms that learn joint models on both rank lists and object features to obtain an aggregated rank ordering that is more accurate and robust, and also helps weed out rank lists of dubious validity. We validate our techniques on synthetic datasets where our algorithm is able to estimate the true rank ordering even when the rank lists are corrupted. Experiments on three real datasets, MQ2008, MQ2007 and OHSUMED, show that using object features can result in significant improvement in performance over existing rank aggregation methods that do not use object information. Furthermore, when at least some of the rank lists are of high quality, our methods are able to effectively exploit such information to output an aggregated rank ordering of high accuracy.

Random walk based rank aggregation to improving web search

For a given query and a set of images ranked lists retrieved from multiple search engines, the metasearch technique aims at combining these lists to build an unified ranking with improved relevance. Rank aggregation is an approach that has been widely used to support this task. This paper investigates the use of rank aggregation methods in the metasearch scenario for diverse image retrieval. Although metasearch systems are usually driven by the relevance of the final result, the impact on diversification has also been analyzed. The experimental findings suggest metasearch based on rank aggregation allows significant improvements, both in terms of relevance and diversity.

Rank aggregation is the process of aggregating multiple base rankers into a single but more comprehensive ranker, which plays an important role in many domains such as recommender system, meta-search, database, genomics, etc. Works related to the comparison of rank aggregation methods all don't have a suitable and general data generation mechanism to produce data with various characteristics and lack a more reasonable and effective algorithm evaluation performance index.Therefore, this paper presents a general data generation mechanism based on Mallows model to produce synthetic controllable datasets, uses generalized Kendall rank correlation coefficient and rank-biased overlap to evaluate and compare the performance of two kinds of methods under different settings.Besides, we also consider the comparison between indices and the impact of data characteristics on the algorithms.This paper may be helpful to researchers and decision-makers from multiple domains.

Metasearch engine is a system that provides unified access to multiple existing search engines. After the results returned from all used component search engines are collected, the metasearch system merges the results into a single ranked list which is expected to be better than the results of the best of the participating search systems. The success of a metasearch engine depends mainly on their rank aggregation method. The system is a better one, if the aggregated list of results displayed before the user satisfies the user with his information need. In this paper, we discuss the development of a metasearch engine that performs user feedback based metasearching using modified rough set based aggregation. Metasearching using the modified rough set based aggregation is performed in two phases namely the ranking rule learning phase and the rank aggregation phase. For each query in the training set, we mine the ranking rules and select the best rules-set by performing cross-validation test. Once the system is trained, we use the best rule set to get the overall ranking for the results returned from different search systems in response to other queries. We also present few snapshots of our system.

We study the problem of combining active learning suggestions to identify informative training examples by empirically comparing methods on benchmark datasets. Many active learning heuristics for classification problems have been proposed to help us pick which instance to annotate next. But what is the optimal heuristic for a particular source of data? Motivated by the success of methods that combine predictors, we combine active learners with bandit algorithms and rank aggregation methods. We demonstrate that a combination of active learners outperforms passive learning in large benchmark datasets and removes the need to pick a particular active learner a priori. We discuss challenges to finding good rewards for bandit approaches and show that rank aggregation performs well.

In the past few years, segmentation and classification techniques have become a cornerstone of many successful remote sensing algorithms aiming at delineating geographic target objects. One common strategy relies on using multiple complex features to guide the delineation process with the objective of gathering complementary information for improving classification results. However, a persistent problem in this approach is how to combine different and noncorrelated feature descriptors automatically. In this regard, one solution is to combine them through multiple classifier systems (MCSs) in which the diversity of simple/noncomplex classifiers is an essential issue in the definition of appropriate strategies for classifier fusion. In this paper, we propose a novel strategy for selecting classifiers (whereby a classifier is taken as a pair of learning method plus image descriptor) to be combined in MCS. In the proposed solution, diversity measures are used to assess the degree of agreement/disagreement between pairs of classifiers and ranked lists are created to sort them according to their diversity score. Thereafter, the classifiers are also sorted according to their performance through different evaluation measures (e.g., kappa and tau indices). In the end, a rank aggregation method is proposed to select the most suitable classifiers based on both the diversity and the effectiveness performance of classifiers. The proposed fusion framework has targeted at coffee crop classification and urban recognition but it is general enough to be used in a variety of other pattern recognition problems. Experimental results demonstrate that the novel strategy yields good results when compared to several baselines while using fewer classifiers and being much more efficient.

Rank aggregation has recently become a common approach for combining individual rankings into a consensus and for quantifying and improving performance in various applications, such as elections, web page rankings, and sports. During the past few years, rankings from many sources have become increasingly high-dimensional and partial. In this study, we develop a rank aggregation method by constructing a directed weighted competition graph. We introduce the concept of “ratio of out- and in-degrees (ROID)” to transform high-dimensional partial rankings into a single consensus. Moreover, we provide a novel effectiveness measure for the aggregate ranking according to its deviations from the ground truth ranking. The proposed method is compared with four typical methods with synthetic rankings. The results indicate that our method outperforms the other four by a significant margin and can be particularly efficient in aggregating high-dimensional rankings. The empirical results validate the effectiveness and feasibility of our method.

More than a decade has passed since the publication of the first report implicating T cells in the bone loss induced by ovariectomy (ovx).1 Since then there has been extraordinary progress in the understanding of the regulatory network that links the hemopoietic and the mesenchymal compartments of the bone marrow (BM), the interactions between the immune system and bone, and the role of lymphocytes as mediators of the effects of calciotrophic hormones in bone. Collectively this body of knowledge has led to the firm establishment of "osteoimmunology" as a novel discipline and a promising area of investigation. The objective of this perspective is to revisit the exciting hypothesis that T cells play a pivotal role in the mechanism of ovx-induced bone loss. The bone-sparing activity of estrogens is due to a repression of bone remodeling coupled with a balancing effect on bone formation and resorption.2 The dominant acute effect of estrogen is the blockade of new osteoclast formation, cells which arise by cytokine-driven proliferation and differentiation of monocyte precursors that circulate within the hematopoietic cell.3 This process is facilitated by BM stromal cells (SCs), which provide physical support for nascent osteoclasts and produce soluble and membrane-associated factors essential for the proliferation and differentiation of osteoclast precursors. This inhibitory effect on osteoclastogenesis is associated with a repressive effect on osteoblastogenesis.4 Another critical effect of estrogen is that of increasing osteoclast apoptosis5-7 while blocking the apoptosis of osteoblasts and osteocytes.2, 8 It has been proposed that the effects of estrogens on the generation of osteoblasts and the lifespan of osteoblasts and osteoclasts result from extranuclear actions of the endoplasmic reticulum (ER) and activation of cytoplasmic kinases.2, 7 The decline of ovarian function at menopause results in decreased production of estrogen and a parallel increase in pituitary follicle-stimulating hormone (FSH) levels. The combined effects of estrogen deprivation and elevated FSH production cause a marked stimulation of bone resorption9 and a period of rapid bone loss that is central to the pathogenesis of postmenopausal osteoporosis. In mice the acute effects of menopause are modeled by ovx, a procedure that stimulates bone resorption by increasing osteoclast formation3 and lifespan.5-7 This initial phase of bone loss is followed by a slower but more prolonged loss of mainly cortical bone due to incomplete refilling of the resorption cavities due to insufficient osteoblast (OB) activity and lifespan.10 An expansion of the osteoclastic pool is therefore the key mechanism responsible for the bone loss that occurs early after ovx. The net bone loss caused by ovx is limited, in part, by an increase in bone formation resulting from stimulated osteoblastogenesis.11 This compensation is fueled by an expansion of the pool of BM–SCs, increased commitment of such pluripotent precursors toward the osteoblastic lineage,11 and enhanced proliferation of early OB precursors.4 Subsequent escalations in OB apoptosis12, 13 extensions of osteoclast lifespan5, 6 and increased secretion of cytokines that suppress bone formation, such as interleukin 7 (IL-7) and tumor necrosis factor (TNF), are the likely reasons for why bone formation does not increase as much as resorption after ovx. The stimulatory effect of ovx on SCs is equally relevant for osteoclastogenesis as one of the consequences of estrogen deprivation is the formation of osteoblastic cells with an increased osteoclastogenic activity14; ie, the capacity to support osteoclast formation. Evidence now suggests that T cells play a pivotal role in the mechanism of ovx-induced bone loss. Core observations include reports showing that ovx fails to induce trabecular and cortical bone loss in: T-cell–deficient nude mice,1, 15-17, wild-type (WT) mice deleted of T cells by injection of anti–T cell antibodies,18 mice treated with Abatacept19 (an agent that blocks T cell costimulation and induces T cell anergy and apoptosis)20, 21 and mice lacking the T cell costimulatory molecule CD40 ligand (CD40L).18 The fact that nude mice are protected against ovx-induced bone loss has been confirmed independently.22 By contrast, Lee and colleagues23 showed that nude mice are protected against the loss of cortical, but not trabecular bone induced by ovx. In another study, T-cell–deficient and B-cell–deficient mice were found to lose bone after ovx.24 The discrepancy between our reports1, 15-19, 25 and those of others23, 24 is likely explained by differences in the experimental design, the lack of B cells in some models, and compensatory mechanisms such as an increase in natural killer (NK) cells producing the osteoclastogenic factor IL-17.26 Both CD4+ and CD8+ cells have been found to play a role in ovx-induced bone loss. CD4+ cells include the TH1, Th2, and Th17 subsets. Th17 are regarded as the most osteoclastogenic subset of T CD4+ cells because they produce high levels of IL-17, receptor activator of NF-κB ligand (RANKL), and TNF, and low levels of interferon gamma (IFNγ).27, 28 The differentiation of Th17 is inhibited by estrogen via a direct effect mediated by estrogen receptor alpha (ERα).29 The role of Th17 in ovx-induced bone loss remains to be determined because one study reported that IL-17R–null mice are more susceptible to ovx-induced bone loss than controls,30 while another group found these mice to be protected from ovx-induced bone loss.26 More abundant information is available about regulatory T cells (Tregs), a population capable of suppressing the effector function of TH1, Th2, and Th17 T cells. Tregs are defined by the expression of the transcription factor FoxP3. Tregs inhibit monocyte differentiation into osteoclasts in vitro and in vivo, and blunt bone resorption31, 32 through the secretion of IL-4, IL-10, and transforming growth factor beta 1 (TGFβ1).33 Attesting to the relevance of Tregs, studies have shown that estrogen increases the relative number of Tregs.27, 34 Moreover, transgenic mice overexpressing Tregs develop progressive high bone mass due to inhibition of bone resorption, and are protected against ovx-induced bone loss.35 Moreover, adoptive transfer of Tregs into T cell–deficient mice increases bone mass, indicating that Tregs directly affect bone homeostasis without the need to engage other T cell lineages.35 Two mechanisms have been described to explain how T cells contribute to ovx-induced bone loss (Fig. 1). The first involves an increase in T cell activation, leading to enhanced production of TNF by BM T cells. The second is a regulatory crosstalk between T cells and SCs, resulting in enhanced production of osteoclastogenic cytokines by SCs. Schematic representation of the role of T cells in the mechanism by which ovx promotes osteoclastogenesis, osteoblastogenesis, and hemopoiesis. Estrogen deficiency promotes T cell activation by increasing the interaction of antigen (Ag)-loaded MHC molecules with bone marrow macrophages (BMM) and dendritic cells (DC) with the T cell receptor (TCR). The Ags are likely to be non-self peptides derived from the intestinal macrobiota. T cell activation also requires at least two costimulatory signals provided by the binding of BMM and DC-expressed CD40 and CD80 to the T cell surface molecules CD40L and CD28, respectively. A critical upstream event is the increased production of reactive oxygen species (ROS) that activate DCs by increasing their expression of CD80. The expansion of T cells in the BM is partially driven by an ovx-induced increase in the thymic output of naïve T cells. Activated T cells secrete TNF that stimulates osteoclast formation primarily by potentiating the response to RANKL. In addition, T cell–expressed CD40L and DLK1/FA-1 increase the osteoclastogenic activity of SC by blunting their secretion of OPG and augmenting their production of RANKL, M-CSF, and other proinflammatory factors. The survival of naïve T cells and some memory T cells requires the low-affinity engagement of the T cell receptor (TCR) by a diverse repertoire of self-antigens (Ags) bound to major histocompatibility complex (MHC) molecules expressed on Ag-presenting cells.36, 37 By contrast, binding of foreign Ags such as bacterial Ags with MHC molecules is followed by high-affinity interactions with TCRs that drive T cell activation. Both low-affinity interactions with self-Ags and high-affinity foreign Ag/MHC/TCR interactions are referred to as "Ag presentation." Ovx induces T cell expression of activation markers and promotes T cell proliferation, expansion, and acquisition of effector functions. These are all features of T cells exposed to foreign Ags.36 The gastrointestinal tract is colonized for life with 100 trillion indigenous bacteria, creating a diverse ecosystem known as the microbiota, whose contributions to human health are profound.38 The macrobiota is likely to represent the source of foreign Ag that drives the expansion of T cells induced by ovx. Although direct verification of a role for macrobiota in T cell expansion is still lacking, compelling supportive data are available. For example, adoptive transfer of T cells into T cell–deficient mice is followed by rapid engraftment and expansion of donor T cells into the host. This process is driven by foreign Ags.36 Attesting to a role of the macrobiota, the expansion of transferred T cells into T cell–deficient mice is greatly reduced in host mice raised in a germ-free environment.36 Moreover, mice maintained in germ-free conditions display increased bone mass due to the lack of immune cell activation.39 T cells are key inducers of bone-wasting because ovx increases T cell TNF production to a level sufficient to augment RANKL-induced osteoclastogenesis.1 This effect is due to an increased number of TNF-producing T cells15 and enhanced production of TNF per cell.18, 40 Ovx also increases the population of premature senescent CD4 + CD28–T cells,40 a lineage that produces high levels of TNF. The presence of increased levels of T cell–produced TNF in the BM of ovx animals is well documented.15, 19, 40, 41 Studies have also shown that menopause increases T cell activation and T cell production of TNF and RANKL in humans.42, 43 The role of TNF in ovx-induced bone loss has been demonstrated in multiple models. For example, ovx fails to induce bone loss in TNF-null mice and in animals lacking the p55 TNF receptor.15 Likewise, transgenic mice insensitive to TNF due to the overexpression of a soluble TNF receptor,44 and mice treated with the TNF inhibitor TNF binding protein45 are protected from ovx-induced bone loss. The specific relevance of T cell TNF production in vivo was demonstrated by the finding that although reconstitution of nude recipient mice with T cells from WT mice restores the capacity of ovx to induce bone loss, reconstitution with T cells from TNF-deficient mice does not.15 The mechanism by which estrogen deficiency expands the pool of TNF-producing T cells is summarized in Figure 1 and involves reactivation of thymic function and induction of T cell activation in the BM. T cell activation is driven by enhanced Ag presentation by macrophages and dendritic cells (DCs).46, 47 The thymus undergoes progressive structural and functional decline with age, coinciding with increased circulating sex-steroid levels at puberty.48 By middle age most parenchymal tissue is replaced by fat, and in both mice and humans fewer T cells are produced and exported to secondary lymphoid organs. However, the thymus continues to generate new T cells even into old age.49, 50 In fact, active lymphocytic thymic tissue has been documented in adults up to 107 years of age.51 Under severe T cell depletion secondary to human immunodeficiency virus (HIV) infection, chemotherapy, or bone marrow transplant, an increase in thymic output (known as thymic rebound) becomes critical for long-term restoration of T cell homeostasis. For example, middle-aged women treated with autologous bone marrow transplants develop thymic hypertrophy and a resurgence of thymic T cell output, which contributes to the restoration of a wide T cell repertoire,52 although the intensity of thymic rebound declines with age. The mechanism driving thymic rebound is not completely understood, but one factor involved is IL-7.53 Both androgens and estrogen suppress thymic function.54, 55 Accordingly, castration reverses thymic atrophy and increases export of recent thymic emigrants to the periphery,56 whereas sex steroids inhibit thymic regeneration by promoting thymocyte apoptosis and arresting thymocyte/prelymphocyte differentiation.57 Restoration of thymic function after castration occurs in young58 as well as in very old rodents.59, 60 In accordance with the notion that estrogen deficiency induces a rebound in thymic function, ovx increases the thymic export of naïve T cells.61 Indeed, stimulated thymic T cell output accounts for ∼50% of the increase in the number of T cells in the periphery. Moreover, thymectomy decreases the bone loss induced by ovx by ∼50%, thus demonstrating that the thymus plays a previously unrecognized role in the pathogenesis of ovx-induced bone loss in mice.61 The remaining bone loss is a consequence of the peripheral expansion of naïve and memory T cells. This finding suggests the tantalizing hypothesis that estrogen deficiency–induced thymic rebound may be responsible for the exaggerated bone loss in young women undergoing surgical menopause62 or for the rapid bone loss characteristic of women in their first 5 to 7 years after natural menopause.63 Indeed, an age-related decrease in estrogen deficiency–induced thymic rebound could mitigate the stimulatory effects of sex steroid deprivation and explain why the rate of bone loss in postmenopausal women diminishes as aging progresses.63 The most upstream effects of ovx in the BM are to stimulate the production of reactive oxygen species (ROS) and to impair the generation of antioxidants.10, 19, 64, 65 In response to ovx, ROS are produced by most BM cells, including T cells.41 ROS play an important role in postmenopausal bone loss by generating a more oxidized bone microenvironment.66, 67 Multiple enzymatic pathways regulate the intracellular redox state through modulation of ROS levels.68 Ovx blunts the BM levels of glutathione (GSH), a critical ROS scavenger, and reduces expression of APE1/Ref-1 and Prx-1 proteins, which collectively limit the production of intracellular ROS.69 ROS have important direct effects on osteoblasts and osteoclasts; these effects have been addressed elsewhere.7, 70, 71 However, additional pivotal effects of ROS include expanding the pool of mature DCs that express the costimulatory molecule CD80, and increasing DC-mediated Ag presentation.19 Antioxidants potently inhibit DC differentiation and their ability to activate T cells,72, 73 in part by suppressing expression of MHC class II and costimulatory molecules in response to Ag.74 N-acetyl-cysteine (NAC), which acts as an intracellular scavenger of ROS by restoring intracellular concentrations of GSM, can block DC maturation75 and DC-mediated T cell activation.76 In vivo support for a role of ROS is provided by experiments demonstrating that administration of antioxidants prevents ovx-induced bone loss,19, 64, 70 while depletion of glutathione by buthionine sulfoximine (BSO), which inhibits glutathione synthesis, enhances bone loss.64 Bone loss caused by BSO has significant similarities to bone loss induced by estrogen deficiency, as both processes are TNF-dependent.77 A second, direct upstream effect of estrogen deficiency is to blunt BM levels of TGFβ,78 a powerful repressor of T cell activation. TGFβ acts as an immunosuppressant by inhibiting T cell activation and T cell production of inflammatory cytokines, including IFNγ. Demonstrating the relevance of the repressive effects of TGFβ on T cell function, mice with T cell–specific blockade of TGFβ signaling were found to be completely resistant to the bone-sparing effects of estrogen.16 Gain of function experiments confirmed that elevation of the systemic levels of TGFβ prevents ovx-induced bone loss and bone turnover.16 The key downstream mechanism by which ovx increases Ag presentation by macrophages is a stimulatory effect on the expression of the gene encoding Class II Transactivator (CIITA). The product of CIITA is a non-DNA binding factor induced by IFNγ that functions as a transcriptional coactivator at the MHC II promoter.79 Increased CIITA expression in macrophages results from ovx-mediated increases in IFNγ production by CD4+ T cells and the responsiveness of CIITA to IFNγ.46 This cytokine was initially described as an anti-osteoclastogenic cytokine because it is a potent inhibitor of osteoclastogenesis in vitro.80 The notion that IFNγ is an inhibitor of bone resorption was reinforced by the finding that silencing of IFNγR−/− signaling leads to a more rapid onset of collagen-induced arthritis and bone resorption81 as compared to WT controls, and by the report that IFNγ decreases serum calcium and osteoclastic bone resorption in nude mice.82, 83 Several mechanisms have been proposed to explain the anti-osteoclastogenic activity of IFNγ, including inhibition of RANKL signaling through the degradation of TNF receptor–associated factor 6 (TRAF6),80 stimulation of apoptosis mediated by Fas/FasL signals,84 and inhibition of RANK and c-Fms gene expression.85 However, the finding that IFNγ is an effective treatment for osteopetrosis both in humans86 and rodents87 demonstrates that the net effect of IFNγ in vivo is to stimulate osteoclastic bone resorption. In keeping with a net pro-resorptive effect of IFNγ in vivo are reports demonstrating that IFNγ−/− and IFNγR−/− mice are protected against ovx-induced bone loss.17, 46 Mice lacking IFNγ production are also protected against infection-induced alveolar bone loss,88 whereas in erosive tuberculoid leprosy and psoriatic arthritis IFNγ production correlates positively with tissue destruction.89 In addition, randomized controlled trials have shown that IFNγ does not prevent bone loss in patients with rheumatoid arthritis (RA),90 nor the bone-wasting effect of cyclosporine A.91 Finally, disruption of IFNγ signaling in vivo results in a strong and sustained inhibition of markers of osteoclastic activity.92 These opposing in vitro and in vivo effects of IFNγ are explained by the fact that IFNγ influences osteoclast formation via both direct and indirect effects.17 IFNγ directly blocks osteoclast formation through targeting of maturing osteoclast.93 However, IFNγ is also a potent inducer of antigen presentation and thus of T cell activation. Therefore, when IFNγ levels are increased in vivo, activated T cells secrete pro-osteoclastogenic factors and this activity offsets the anti-osteoclastogenic effects of IFNγ.17 It should also be mentioned that it is now recognized that IFNγ affects bone turnover by promoting the commitment of BM–SCs into the osteoblastic lineage and their differentiation into mature osteoblasts.94 Accordingly, treatment with IFNγ reverses ovx-induced bone loss by promoting bone formation.92 Another mechanism by which estrogen regulates T cell TNF production is by repressing the production of IL-7, a potent lymphopoietic cytokine and an inducer of bone destruction in vivo.95 Attesting to the relevance of this factor, IL-7 levels are significantly elevated following ovx,61, 96 and in vivo IL-7 blockade is effective in preventing ovx-induced bone destruction.96 The elevated BM levels of IL-7 contribute to the expansion of the T cell population in peripheral lymphoid organs through several mechanisms. First, IL-7 directly stimulates T cell proliferation.61 Second, IL-7 increases antigen presentation by upregulating the production of IFNγ. Third, IL-7 and TGFβ regulate The in TGFβ signaling characteristic of estrogen deficiency may to stimulate IL-7 thus driving the of osteoclastogenic cytokine production and bone In estrogen deficiency, IL-7 bone loss by suppressing bone formation and thus bone formation from resorption. A that the crosstalk between T cells and SCs is relevant for ovx-induced bone loss was provided by the finding that activated T cells induce SC apoptosis via the ligand a which blunts the compensatory increase in bone formation that bone loss in ovx More abundant information is available about the T crosstalk driven by the also known as is a key surface ligand expressed on T CD40L to and several CD40 is expressed on cells, hemopoietic and cells of the osteoblastic CD40L has been to because T cells, through the production of the anti-osteoclastogenic factor by B mice a reduced bone due to exaggerated bone bone has also been found in by a in which CD40L production is due to a in the CD40L However, mice lacking T cell–expressed CD40L are protected against hormone bone the that CD40L may in while promoting bone resorption conditions of bone Studies with mice and with WT mice treated with the CD40L have that silencing of CD40L completely prevents ovx-induced bone A mechanism was shown to be First, silencing of CD40L blocks the activation of T cells and the resulting production of TNF. Second, CD40L is for ovx to increase the proliferation and the differentiation of SCs and their capacity to support osteoclast formation through enhanced production of factor and RANKL, and secretion of a critical additional mechanism by which T cells bone homeostasis in ovx mice is through a crosstalk between T cells and SCs that results in enhanced osteoclastogenesis to a enhanced In ovx increases the number of activated T cells that the expression of and RANKL by SCs, and ovx the SC production of The net result is a significant increase in the rate of The interaction of CD40L with CD40 on SCs in the of estrogen deficiency to the effects of costimulation on B cell OPG the of osteoclast formation in of bone loss. An additional mechanism of T crosstalk the novel of bone mass antigen has been for a known as This factor can be to generate a soluble known as In conditions DLK1/FA-1 are produced by SCs, B cells, and T cells. These factors stimulate osteoclastogenesis and block by the production of TNF, IL-7, and other inflammatory cytokines by SCs. Ovx increases the production of DLK1/FA-1 by activated CD4+ and CD8+ T cells, resulting in a stimulation of the production of osteoclastogenic cytokines by SCs. Attesting to the relevance of mice are significantly protected against ovx-induced bone The hypothesis that the immune system is pivotal for the bone loss about by menopause remains to be demonstrated in it remains to the of why the immune system in and is involved in the mechanism of ovx-induced bone loss. may by postmenopausal bone loss as an of an event critical for the need to stimulate bone resorption in the This process is essential to the increased for calcium about by The for this event is the acute in estrogen levels A second to the period is in associated with is by proliferation of the and formation of These are by and is now that regulates and to by the production of RANKL and a system that is for bone homeostasis is also a key of The in levels in the period results in reduced tissue levels of RANK and RANKL, with the effect of the for The increase in the and TNF levels in the BM induced by estrogen levels the of calcium from the to the and then to the a state of to the is induced by an increase in the number of A to the of is the loss of such immune and the restoration of a immune This is through a decrease in the resulting from the acute in steroid levels. However, a between immune to the and bone is provided by the that OPG is expressed by human it is to that of ovarian function induces bone loss through an immune response because natural has these key within the immune system to (Fig. Schematic representation of in the period and of the involved cells and The that estrogen and are followed by a stimulation of bone resorption, to and of immune to the Activated T cells drive bone resorption by TNF. A decrease in the number of Tregs induces the of the immune The loss of OPG may contribute to the and the increased bone resorption. A decrease in levels of RANKL and RANK induce the of the for with increased of calcium from the the in the contribute to an event critical for The data the hypothesis that of ovarian function induces bone loss through an immune response because natural has key within the immune system to The data support the hypothesis that the bone loss induced by estrogen deficiency is due to a complex of hormones and cytokines that to the process of bone Ovx T cell TNF production by increasing T cell activity in the and the peripheral lymphoid organs. T cell precursors the BM and to the T cell and expansion in the of from the these new T cells to peripheral lymphoid including the BM Ovx induces T cell activation in the BM in part by directly promoting antigen and in part via stimulation of IL-7 and IFNγ production and of TGFβ The net result of these actions is an increase in the number of TNF-producing T cells. The elevated levels of TNF increase RANKL-induced osteoclast formation. Estrogen deficiency also T cell activation and osteoclastogenesis by leading to an in The combined effect of IFNγ and ROS enhances Ag T cell activation. T cells stimulate RANKL and production by SCs, through CD40L and The that has of

In credit risk evaluation the accuracy of a classifier is very significant for classifying the high-risk loan applicants correctly. Feature selection is one way of improving the accuracy of a classifier. It provides the classifier with important and relevant features for model development. This study uses the ensemble of multiple feature ranking techniques for feature selection of credit data. It uses five individual rank based feature selection methods. It proposes a novel rank aggregation algorithm for combining the ranks of the individual feature selection methods of the ensemble. This algorithm uses the rank order along with the rank score of the features in the ranked list of each feature selection method for rank aggregation. The ensemble of multiple feature selection techniques uses the novel rank aggregation algorithm and selects the relevant features using the 80%, 60%, 40% and 20% thresholds from the top of the aggregated ranked list for building the C4.5, MLP, C4.5 based Bagging and MLP based Bagging models. It was observed that the performance of models using the ensemble of multiple feature selection techniques is better than the performance of 5 individual rank based feature selection methods. The average performance of all the models was observed as best for the ensemble of feature selection techniques at 60% threshold. Also, the bagging based models outperformed the individual models most significantly for the 60% threshold. This increase in performance is more significant from the fact that the number of features were reduced by 40% for building the highest performing models. This reduces the data dimensions and hence the overall data size phenomenally for model building. The use of the ensemble of feature selection techniques using the novel aggregation algorithm provided more accurate models which are simpler, faster and easy to interpret.