802 resultados para discounted cumulative gain


Relevância:

100.00% 100.00%

Publicador:

Resumo:

In the TREC Web Diversity track, novelty-biased cumulative gain (α-NDCG) is one of the official measures to assess retrieval performance of IR systems. The measure is characterised by a parameter, α, the effect of which has not been thoroughly investigated. We find that common settings of α, i.e. α=0.5, may prevent the measure from behaving as desired when evaluating result diversification. This is because it excessively penalises systems that cover many intents while it rewards those that redundantly cover only few intents. This issue is crucial since it highly influences systems at top ranks. We revisit our previously proposed threshold, suggesting α be set on a query-basis. The intuitiveness of the measure is then studied by examining actual rankings from TREC 09-10 Web track submissions. By varying α according to our query-based threshold, the discriminative power of α-NDCG is not harmed and in fact, our approach improves α-NDCG's robustness. Experimental results show that the threshold for α can turn the measure to be more intuitive than using its common settings.

Relevância:

100.00% 100.00%

Publicador:

Resumo:

Novelty-biased cumulative gain (α-NDCG) has become the de facto measure within the information retrieval (IR) community for evaluating retrieval systems in the context of sub-topic retrieval. Setting the incorrect value of parameter α in α-NDCG prevents the measure from behaving as desired in particular circumstances. In fact, when α is set according to common practice (i.e. α = 0.5), the measure favours systems that promote redundant relevant sub-topics rather than provide novel relevant ones. Recognising this characteristic of the measure is important because it affects the comparison and the ranking of retrieval systems. We propose an approach to overcome this problem by defining a safe threshold for the value of α on a query basis. Moreover, we study its impact on system rankings through a comprehensive simulation.

Relevância:

100.00% 100.00%

Publicador:

Resumo:

Discounted Cumulative Gain (DCG) is a well-known ranking evaluation measure for models built with multiple relevance graded data. By handling tagging data used in recommendation systems as an ordinal relevance set of {negative,null,positive}, we propose to build a DCG based recommendation model. We present an efficient and novel learning-to-rank method by optimizing DCG for a recommendation model using the tagging data interpretation scheme. Evaluating the proposed method on real-world datasets, we demonstrate that the method is scalable and outperforms the benchmarking methods by generating a quality top-N item recommendation list.

Relevância:

100.00% 100.00%

Publicador:

Resumo:

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.

Relevância:

100.00% 100.00%

Publicador:

Resumo:

The content-based image retrieval is important for various purposes like disease diagnoses from computerized tomography, for example. The relevance, social and economic of image retrieval systems has created the necessity of its improvement. Within this context, the content-based image retrieval systems are composed of two stages, the feature extraction and similarity measurement. The stage of similarity is still a challenge due to the wide variety of similarity measurement functions, which can be combined with the different techniques present in the recovery process and return results that aren’t always the most satisfactory. The most common functions used to measure the similarity are the Euclidean and Cosine, but some researchers have noted some limitations in these functions conventional proximity, in the step of search by similarity. For that reason, the Bregman divergences (Kullback Leibler and I-Generalized) have attracted the attention of researchers, due to its flexibility in the similarity analysis. Thus, the aim of this research was to conduct a comparative study over the use of Bregman divergences in relation the Euclidean and Cosine functions, in the step similarity of content-based image retrieval, checking the advantages and disadvantages of each function. For this, it was created a content-based image retrieval system in two stages: offline and online, using approaches BSM, FISM, BoVW and BoVW-SPM. With this system was created three groups of experiments using databases: Caltech101, Oxford and UK-bench. The performance of content-based image retrieval system using the different functions of similarity was tested through of evaluation measures: Mean Average Precision, normalized Discounted Cumulative Gain, precision at k, precision x recall. Finally, this study shows that the use of Bregman divergences (Kullback Leibler and Generalized) obtains better results than the Euclidean and Cosine measures with significant gains for content-based image retrieval.

Relevância:

100.00% 100.00%

Publicador:

Resumo:

The growing availability and popularity of opinion rich resources on the online web resources, such as review sites and personal blogs, has made it convenient to find out about the opinions and experiences of layman people. But, simultaneously, this huge eruption of data has made it difficult to reach to a conclusion. In this thesis, I develop a novel recommendation system, Recomendr that can help users digest all the reviews about an entity and compare candidate entities based on ad-hoc dimensions specified by keywords. It expects keyword specified ad-hoc dimensions/features as input from the user and based on those features; it compares the selected range of entities using reviews provided on the related User Generated Contents (UGC) e.g. online reviews. It then rates the textual stream of data using a scoring function and returns the decision based on an aggregate opinion to the user. Evaluation of Recomendr using a data set in the laptop domain shows that it can effectively recommend the best laptop as per user-specified dimensions such as price. Recomendr is a general system that can potentially work for any entities on which online reviews or opinionated text is available.

Relevância:

80.00% 80.00%

Publicador:

Resumo:

In this paper, we use reinforcement learning (RL) as a tool to study price dynamics in an electronic retail market consisting of two competing sellers, and price sensitive and lead time sensitive customers. Sellers, offering identical products, compete on price to satisfy stochastically arriving demands (customers), and follow standard inventory control and replenishment policies to manage their inventories. In such a generalized setting, RL techniques have not previously been applied. We consider two representative cases: 1) no information case, were none of the sellers has any information about customer queue levels, inventory levels, or prices at the competitors; and 2) partial information case, where every seller has information about the customer queue levels and inventory levels of the competitors. Sellers employ automated pricing agents, or pricebots, which use RL-based pricing algorithms to reset the prices at random intervals based on factors such as number of back orders, inventory levels, and replenishment lead times, with the objective of maximizing discounted cumulative profit. In the no information case, we show that a seller who uses Q-learning outperforms a seller who uses derivative following (DF). In the partial information case, we model the problem as a Markovian game and use actor-critic based RL to learn dynamic prices. We believe our approach to solving these problems is a new and promising way of setting dynamic prices in multiseller environments with stochastic demands, price sensitive customers, and inventory replenishments.

Relevância:

80.00% 80.00%

Publicador:

Resumo:

Includes bibliography

Relevância:

80.00% 80.00%

Publicador:

Resumo:

Using two standard cycle methodologies (Classical and Deviation Cycle) and a comprehensive sample of 83 countries worldwide, including all developing regions, we show that the Latin American and Caribbean cycle exhibits two distinctive features. First, and most importantly, its expansion performance is shorter and for the most par less imtense than that of the rest of the regions considered, and in particular than that of East Asia and the Pacific, East Asia and the Pacific's expansions last five years longer than those of LAC, and its output gain is 50% greater than that of LAC. Second, LAC tends to exhibit contractions that are not significantly different in terms of duration and amplitude than t those of other regions. Both these features imply that the complete Latin American and Caribbean cycle has, overall, the shortest duration and smallest amplitude in relation to other regions. The specificities of the Latin American and Caribbean cycle are not confined to the short run. These are also reflected in variables such as productivity and investment, which are linked to long-run growth. East Asia and the Pacific's cumulative gain in labor productivity during the expansionary phase is twice that of LAC. Moreover, the evidence also shows that the effects of the contraction in public investment surpass those of the expansion leading to a declining trend over the entire cycle. In this sense we suggest that policy analysis needs to increase its focus on the expansionary phase of the cycle. Improving our knowledge of the differences in the expansionary dynamics of countries and regions, can further our understanding of the differences in their rates of growth and levels of development. We also suggest that while, the management of the cycle affects the short-run fluctuations of economic activity and hence volatility, it is not trend neutral. Hence, the effects of aggregate demand management policies may be more persistent over time and less transitory than currently thought.

Relevância:

80.00% 80.00%

Publicador:

Resumo:

This report analyses the agriculture, health and tourism sectors in Saint Lucia to assess the potential economic impacts of climate change on the sectors. The fundamental aim of this report is to assist with the development of strategies to deal with the potential impact of climate change in Saint Lucia. It also has the potential to provide essential input for identifying and preparing policies and strategies to help advance the Caribbean subregion closer to solving problems associated with climate change and attaining individual and regional sustainable development goals. Some of the key anticipated impacts of climate change for the Caribbean include elevated air and sea-surface temperatures, sea-level rise, possible changes in extreme events and a reduction in freshwater resources. The economic impact of climate change on the three sectors was estimated for the A2 and B2 IPCC scenarios until 2050. An evaluation of various adaptation strategies for each sector was also undertaken using standard evaluation techniques. The key subsectors in agriculture are expected to have mixed impacts under the A2 and B2 scenarios. Banana, fisheries and root crop outputs are expected to fall with climate change, but tree crop and vegetable production are expected to rise. In aggregate, in every decade up to 2050, these sub-sectors combined are expected to experience a gain under climate change with the highest gains under A2. By 2050, the cumulative gain under A2 is calculated as approximately US$389.35 million and approximately US$310.58 million under B2, which represents 17.93% and 14.30% of the 2008 GDP respectively. This result was unexpected and may well be attributed to the unavailability of annual data that would have informed a more robust assessment. Additionally, costs to the agriculture sector due to tropical cyclones were estimated to be $6.9 million and $6.2 million under the A2 and B2 scenarios, respectively. There are a number of possible adaptation strategies that can be employed by the agriculture sector. The most attractive adaptation options, based on the benefit-cost ratio are: (1) Designing and implementation of holistic water management plans (2) Establishment of systems of food storage and (3) Establishment of early warning systems. Government policy should focus on the development of these adaption options where they are not currently being pursued and strengthen those that have already been initiated, such as the mainstreaming of climate change issues in agricultural policy. The analysis of the health sector placed focus on gastroenteritis, schistosomiasis, ciguatera poisoning, meningococal meningitis, cardiovascular diseases, respiratory diseases and malnutrition. The results obtained for the A2 and B2 scenarios demonstrate the potential for climate change to add a substantial burden to the health system in the future, a factor that will further compound the country’s vulnerability to other anticipated impacts of climate change. Specifically, it was determined that the overall Value of Statistical Lives impacts were higher under the A2 scenario than the B2 scenario. A number of adaptation cost assumptions were employed to determine the damage cost estimates using benefit-cost analysis. The benefit-cost analysis suggests that expenditure on monitoring and information provision would be a highly efficient step in managing climate change and subsequent increases in disease incidence. Various locations in the world have developed forecasting systems for dengue fever and other vector-borne diseases that could be mirrored and implemented. Combining such macro-level policies with inexpensive micro-level behavioural changes may have the potential for pre-empting the re-establishment of dengue fever and other vector-borne epidemic cycles in Saint Lucia. Although temperature has the probability of generating significant excess mortality for cardiovascular and respiratory diseases, the power of temperature to increase mortality largely depends on the education of the population about the harmful effects of increasing temperatures and on the existing incidence of these two diseases. For these diseases it is also suggested that a mix of macro-level efforts and micro-level behavioural changes can be employed to relieve at least part of the threat that climate change poses to human health. The same principle applies for water and food-borne diseases, with the improvement of sanitation infrastructure complementing the strengthening of individual hygiene habits. The results regarding the tourism sector imply that the tourism climatic index was likely to experience a significant downward shift in Saint Lucia under the A2 as well as the B2 scenario, indicative of deterioration in the suitability of the island for tourism. It is estimated that this shift in tourism features could cost Saint Lucia about 5 times the 2009 GDP over a 40-year horizon. In addition to changes in climatic suitability for tourism, climate change is also likely to have important supply-side effects on species, ecosystems and landscapes. Two broad areas are: (1) coral reefs, due to their intimate link to tourism, and, (2) land loss, as most hotels tend to lie along the coastline. The damage related to coral reefs was estimated at US$3.4 billion (3.6 times GDP in 2009) under the A2 scenario and US$1.7 billion (1.6 times GDP in 2009) under the B2 scenario. The damage due to land loss arising from sea level rise was estimated at US$3.5 billion (3.7 times GDP) under the A2 scenario and US$3.2 billion (3.4 times GDP) under the B2 scenario. Given the potential for significant damage to the industry a large number of potential adaptation measures were considered. Out of these a short-list of 9 potential options were selected by applying 10 evaluation criteria. Using benefit-cost analyses 3 options with positive ratios were put forward: (1) increased recommended design speeds for new tourism-related structures; (2) enhanced reef monitoring systems to provide early warning alerts of bleaching events, and, (3) deployment of artificial reefs or other fish-aggregating devices. While these options had positive benefit-cost ratios, other options were also recommended based on their non-tangible benefits. These include the employment of an irrigation network that allows for the recycling of waste water, development of national evacuation and rescue plans, providing retraining for displaced tourism workers and the revision of policies related to financing national tourism offices to accommodate the new climate realities.

Relevância:

30.00% 30.00%

Publicador:

Resumo:

Few studies have evaluated the reliability of lifetime sun exposure estimated from inquiring about the number of hours people spent outdoors in a given period on a typical weekday or weekend day (the time-based approach). Some investigations have suggested that women have a particularly difficult task in estimating time outdoors in adulthood due to their family and occupational roles. We hypothesized that people might gain additional memory cues and estimate lifetime hours spent outdoors more reliably if asked about time spent outdoors according to specific activities (an activity-based approach). Using self-administered, mailed questionnaires, test-retest responses to time-based and to activity-based approaches were evaluated in 124 volunteer radiologic technologist participants from the United States: 64 females and 60 males 48 to 80 years of age. Intraclass correlation coefficients (ICC) were used to evaluate the test-retest reliability of average number of hours spent outdoors in the summer estimated for each approach. We tested the differences between the two ICCs, corresponding to each approach, using a t test with the variance of the difference estimated by the jackknife method. During childhood and adolescence, the two approaches gave similar ICCs for average numbers of hours spent outdoors in the summer. By contrast, compared with the time-based approach, the activity-based approach showed significantly higher ICCs during adult ages (0.69 versus 0.43, P = 0.003) and over the lifetime (0.69 versus 0.52, P = 0.05); the higher ICCs for the activity-based questionnaire were primarily derived from the results for females. Research is needed to further improve the activity-based questionnaire approach for long-term sun exposure assessment. (Cancer Epidemiol Biomarkers Prev 2009;18(2):464–71)

Relevância:

30.00% 30.00%

Publicador:

Resumo:

Three drafts of Bos indicus cross steers (initially 178-216 kg) grazed Leucaena-grass pasture [Leucaena leucocephala subspecies glabrata cv. Cunningham with green panic (Panicum maximum cv. trichoglume)] from late winter through to autumn during three consecutive years in the Burnett region of south-east Queensland. Measured daily weight gain (DWGActual) of the steers was generally 0.7-1.1 kg/day during the summer months. Estimated intakes of metabolisable energy and dry matter (DM) were calculated from feeding standards as the intakes required by the steers to grow at the DWGActual. Diet attributes were predicted from near infrared reflectance spectroscopy spectra of faeces (F.NIRS) using established calibration equations appropriate for northern Australian forages. Inclusion of some additional reference samples from cattle consuming Leucaena diets into F.NIRS calibrations based on grass and herbaceous legume-grass pastures improved prediction of the proportion of Leucaena in the diet. Mahalanobis distance values supported the hypothesis that the F.NIRS predictions of diet crude protein concentration and DM digestibility (DMD) were acceptable. F.NIRS indicated that the percentage of Leucaena in the diet varied widely (10-99%). Diet crude protein concentration and DMD were usually high, averaging 12.4 and 62%, respectively, and were related asymptotically to the percentage of Leucaena in the diet (R2 = 0.48 and 0.33, respectively). F.NIRS calibrations for DWG were not satisfactory to predict this variable from an individual faecal sample since the s.e. of prediction were 0.33-0.40 kg/day. Cumulative steer liveweight (LW) predicted from F.NIRS DWG calibrations, which had been previously developed with tropical grass and grass-herbaceous legume pastures, greatly overestimated the measured steer LW; therefore, these calibrations were not useful. Cumulative steer LW predicted from a modified F.NIRS DWG calibration, which included data from the present study, was strongly correlated (R2 = 0.95) with steer LW but overestimated LW by 19-31 kg after 8 months. Additional reference data are needed to develop robust F.NIRS calibrations to encompass the diversity of Leucaena pastures of northern Australia. In conclusion, the experiment demonstrated that F.NIRS could improve understanding of diet quality and nutrient intake of cattle grazing Leucaena-grass pasture, and the relationships between nutrient supply and cattle growth.

Relevância:

30.00% 30.00%

Publicador:

Resumo:

The interplay between obesity, physical activity, weight gain and genetic variants in mTOR pathway have not been studied in renal cell carcinoma (RCC). We examined the associations between obesity, weight gain, physical activity and RCC risk. We also analyzed whether genetic variants in the mTOR pathway could modify the association. Incident renal cell carcinoma cases and healthy controls were recruited from the University of Texas MD Anderson Cancer Center in Houston, Texas. Cases and controls were frequency-matched by age (±5 years), ethnicity, sex, and county of residence. Epidemiologic data were collected via in-person interview. A total of 577 cases and 593 healthy controls (all white) were included. One hundred ninety-two (192) SNPs from 22 genes were available and their genotyping data were extracted from previous genome-wide association studies. Logistic regression and regression spline were performed to obtain odds ratios. Obesity at age 20, 40, and 3 years prior to diagnosis/recruitment, and moderate and large weight gain from age 20 to 40 were each significantly associated with increased RCC risk. Low physical activity was associated with a 4.08-fold (95% CI: 2.92-5.70) increased risk. Five single nucleotide polymorphisms (SNPs) were significantly associated with RCC risk and their cumulative effect increased the risk by up to 72% (95% CI: 1.20-2.46). Strata specific effects for weight change and genotyping cumulative groups were observed. However, no interaction was suggested by our study. In conclusion, energy balance related risk factors and genetic variants in the mTOR pathway may jointly influence susceptibility to RCC. ^

Relevância:

20.00% 20.00%

Publicador:

Resumo:

This chapter describes the use of collaborative learning as an approach to enhance English language learning by students from non-English speaking backgrounds. Communicative Language Teaching (CLT) principles were applied to two case studies, one comprising of undergraduate English as Foreign Language Learners in Turkey and the other involved English as Second Language learners in Australia. Social constructivism inspired communicative language teaching using collaborative learning activities such as team work, interactive peer-based learning, and iterative stages of learning matrix were incorporated to enhance students' learning outcomes. Data collected after the CLT intervention was made up of field notes, reflective logs and focus group interviews which revealed complementarities, as well as subtle differences between the two cases. The findings were summarized as learning dispositions; speaking fluency and confidence; learning diagnostics and completion deficiencies; task engagement, flow theory and higher order thinking skills; in addition to self efficacy and development of student identity. CLT has the potential to provide a more inclusive and dynamic education for diverse learners through vital outcomes and benefits which resonate with the real world.