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.

Evaluation of mechanical losses in a linear motor pressure wave generator

A moving magnet linear motor compressor or pressure wave generator (PWG) of 2 cc swept volume with dual opposed piston configuration has been developed to operate miniature pulse tube coolers. Prelimnary experiments yielded only a no-load cold end temperature of 180 K. Auxiliary tests and the interpretation of detailed modeling of a PWG suggest that much of the PV power has been lost in the form of blow-by at piston seals due to large and non-optimum clearance seal gap between piston and cylinder. The results of experimental parameters simulated using Sage provide the optimum seal gap value for maximizing the delivered PV power.

Catalytic activity of Peptidase N is required for adaptation of Escherichia coli to nutritional downshift and high temperature stress

Peptidase N (PepN), the sole M1 family member in Escherichia coli, displays broad substrate specificity and modulates stress responses: it lowers resistance to sodium salicylate (NaSal)-induced stress but is required during nutritional downshift and high temperature (NDHT) stress. The expression of PepN does not significantly change during different growth phases in LB or NaSal-induced stress; however, PepN amounts are lower during NDHT stress. To gain mechanistic insights on the roles of catalytic activity of PepN in modulating these two stress responses, alanine mutants of PepN replacing E264 (GAMEN motif) and E298 (HEXXH motif) were generated. There are no major structural changes between purified wild type (WT) and mutant proteins, which are catalytically inactive. Importantly, growth profiles of Delta pepN upon expression of WT or mutant proteins demonstrated the importance of catalytic activity during NDHT but not NaSal-induced stress. Further fluorescamine reactivity studies demonstrated that the catalytic activity of PepN is required to generate higher intracellular amounts of free N-terminal amino acids; consequently, the lower growth of Delta pepN during NDHT stress increases with high amounts of casamino acids. Together, this study sheds insights on the expression and functional roles of the catalytic activity of PepN during adaptation to NDHT stress. (C) 2012 Elsevier GmbH. All rights reserved.

Classification calibration dimension for general multiclass losses

We study consistency properties of surrogate loss functions for general multiclass classification problems, defined by a general loss matrix. We extend the notion of classification calibration, which has been studied for binary and multiclass 0-1 classification problems (and for certain other specific learning problems), to the general multiclass setting, and derive necessary and sufficient conditions for a surrogate loss to be classification calibrated with respect to a loss matrix in this setting. We then introduce the notion of \emph{classification calibration dimension} of a multiclass loss matrix, which measures the smallest `size' of a prediction space for which it is possible to design a convex surrogate that is classification calibrated with respect to the loss matrix. We derive both upper and lower bounds on this quantity, and use these results to analyze various loss matrices. In particular, as one application, we provide a different route from the recent result of Duchi et al.\ (2010) for analyzing the difficulty of designing `low-dimensional' convex surrogates that are consistent with respect to pairwise subset ranking losses. We anticipate the classification calibration dimension may prove to be a useful tool in the study and design of surrogate losses for general multiclass learning problems.

Extension of TSVM to multi-class and hierarchical text classification problems With general losses

Transductive SVM (TSVM) is a well known semi-supervised large margin learning method for binary text classification. In this paper we extend this method to multi-class and hierarchical classification problems. We point out that the determination of labels of unlabeled examples with fixed classifier weights is a linear programming problem. We devise an efficient technique for solving it. The method is applicable to general loss functions. We demonstrate the value of the new method using large margin loss on a number of multi-class and hierarchical classification datasets. For maxent loss we show empirically that our method is better than expectation regularization/constraint and posterior regularization methods, and competitive with the version of entropy regularization method which uses label constraints.

Nutritional benefits from domatia inhabitants in an ant-plant interaction: interlopers do pay the rent

1. How a symbiosis originates and is maintained are important evolutionary questions. Symbioses in myrmecophytes (plants providing nesting for ants) are believed to be maintained by protection and nutrients provided by specialist plant-ants in exchange for nesting spaces (called domatia) and nourishment offered by ant-plants. However, besides the benefits accrued from housing protective ants, the mechanisms contributing to the fitness advantages of bearing domatia have rarely been examined, especially because the domatia trait is usually constitutively expressed, and many myrmecophytes have obligate mutualisms with single ant species resulting in invariant conditions. 2. In the unspecialized ant-plant Humboldtia brunonis (Fabaceae) that offers extrafloral nectar to ants, only some plants produce domatia in the form of hollow internodes. These domatia have a self-opening slit making them more prone to interlopers and are occupied mostly by non-protective ants and other invertebrates, especially arboreal earthworms. The protection mutualism with ants is restricted in geographical extent, occurring only at a few sites in the southernmost part of this plant's range in the Western Ghats of India. 3. We examined nutrient flux from domatia residents to the plant using stable isotopes. We found that between 9% (earthworms) and 17% (protective or non-protective ants) of nitrogen of plant tissues nearest the domatium came from domatia inhabitants. Therefore, interlopers such as earthworms and non-protective ants contributed positively to the nitrogen budget of localized plant modules of this understorey tree. N-15-enriched feeding experiments with protective ants demonstrated that nutrients flowed from domatia inhabitants to nearby plant modules. Fruit set did not differ between paired hand-pollinated inflorescences on domatia and non-domatia bearing branches. This was possibly due to the nutrient flux from domatia to adjacent branches without domatia within localized modules. 4. This study has demonstrated the nutritive role of non-protective ants and non-ant invertebrates, hitherto referred to as interlopers, in an unspecialized myrmecophyte. Our study suggests that even before the establishment of a specialized ant-plant protection mutualism, nutritional benefits conferred by domatia inhabitants can explain the fitness benefits of bearing domatia, and thus the maintenance of a trait that facilitates the establishment of a specialized ant-plant symbiosis.

Surrogate Regret Bounds for Bipartite Ranking via Strongly Proper Losses

The problem of bipartite ranking, where instances are labeled positive or negative and the goal is to learn a scoring function that minimizes the probability of mis-ranking a pair of positive and negative instances (or equivalently, that maximizes the area under the ROC curve), has been widely studied in recent years. A dominant theoretical and algorithmic framework for the problem has been to reduce bipartite ranking to pairwise classification; in particular, it is well known that the bipartite ranking regret can be formulated as a pairwise classification regret, which in turn can be upper bounded using usual regret bounds for classification problems. Recently, Kotlowski et al. (2011) showed regret bounds for bipartite ranking in terms of the regret associated with balanced versions of the standard (non-pairwise) logistic and exponential losses. In this paper, we show that such (non-pairwise) surrogate regret bounds for bipartite ranking can be obtained in terms of a broad class of proper (composite) losses that we term as strongly proper. Our proof technique is much simpler than that of Kotlowski et al. (2011), and relies on properties of proper (composite) losses as elucidated recently by Reid and Williamson (2010, 2011) and others. Our result yields explicit surrogate bounds (with no hidden balancing terms) in terms of a variety of strongly proper losses, including for example logistic, exponential, squared and squared hinge losses as special cases. An important consequence is that standard algorithms minimizing a (non-pairwise) strongly proper loss, such as logistic regression and boosting algorithms (assuming a universal function class and appropriate regularization), are in fact consistent for bipartite ranking; moreover, our results allow us to quantify the bipartite ranking regret in terms of the corresponding surrogate regret. We also obtain tighter surrogate bounds under certain low-noise conditions via a recent result of Clemencon and Robbiano (2011).

Nutritional and anti-nutritional analyses of Azolla africana Desv. and Spirodela polyrrhiza L. Schleiden as feedstuffs for fish production

The chemical composition of Azolla africana and Spirodela polyrrhiza cultivated in earthen ponds were determined. Crude protein contents of the samples were 28.9~c0.6 and 25.6~c0.2% dry matter for A. africana and S. polyrrhiza respectively. Dry matter, crude fibre and lipid contents of A. africana were higher (P<0.05) than values obtained for S. polyrrhiza. Mineral analyses showed that S. polyrrhiza contained higher levels of Na, S, Ca, Mg and Fe than A. africana. Except for Ca content in S. polyrrhiza, heavy metals (Ni and Zn) accumulation in Azolla were very high. There were no wide differences in the individual amino acid indexes except for methionine. Some anti-nutritional factors were determined. Cyanide, tannin and phytin contents of fresh weed samples were higher than sun-dried samples. A. africana contained more cyanide and tannin than S. polyrrhiza both in fresh and sun-dried forms

Assessment of post-harvest losses in Nigeria fishery: the Kainji Lake model

The paper provides a description of a methodology used for quantitative assessment of post harvest losses in the Kainji Lake Fishery (Nigeria). The sample population was made up of 314 fisherfolk, 115 processors, 125 fish buyers and 111 fish sellers. For the determination of handling losses, 24,839 fishes weighing 2,389.31 kg belonging to 43 species were examined of which 10% by number and 9% by weight deteriorated at checking and 4% by number and 3% by weight at landing. Processing losses recorded 22% by number and 16% by weight deteriorated prior to and during smoking with the traditional 'Banda' kiln. During marketing, 16% of fish sold had deteriorated and 6% by weight of fish bought also deteriorated, mainly due to insect infestation during storage. Based on the 1995 yield estimate for Kainji Lake fishery, approximately 1000 tons of fish estimated at 80 million Naira were lost during handling alone. This figure would be much higher if the level of losses during processing and marketing are included. This assessment technique is recommended for use in obtaining quantifiable data on post harvest losses from other water bodies in Nigeria

A preliminary investigation into the post harvest losses of fish in Shiroro Lake area

The findings are presented of a study conducted to assess the post harvest losses in Shiroro Lake, Nigeria. The major objectives were to identify and quantify the types of losses, to provide recommendations that would enhance formulation of policy guidelines for utilization and exploitation of the declining fishery resources of the lake

Studies on the nutritional value of cassava (Manihot utillisima) peels as energy source in the diet of Oreochromis niloticus fry fed in indoor glass aquaria

This study, though, has as its core objective cost reduction in aquaculture nutrition was equally designed to investigate the value of the peels of cassava (Manihot utillisima) as energy source in the diet of Oreochromis niloticus fry. Three levels of cassava peels diet and a control (100% yellow maize in the carbohydrate mixture) was prepared and tested on O. niloticus fry for ten (10) weeks. The fry with mean weight of 0.32g were grouped fifteen (15) in each of the glass aquaria measuring 60x30x30cm with a maximum capacity of 52 litres of water. The fry were fed twice daily at 10% biomass. Weekly, the fry were weighed to determine the weight increment or otherwise and the quality of feed adjusted accordingly. Water quality parameters like temperature, pH and dissolved oxygen (D.0) were monitored and found to be at desirable level. DT 3 (97 % cassava peels and 3% yellow maize) in the carbohydrate mixture gave the best growth performance. The fry fed, this diet gained mean weight of 1.18g for the period of the experiment. However, the poorest performance in terms of growth was from fry fed the control diet (100%yellow maize in the carbohydrate mixture) fry fed this diet gained mean weight of 0.80 for the duration of the experiment. Analysis of the various growth indices like SGR, PER, FCR and NPU shows that DT3 was the overall best diet with an SGR value of2.40 and FCR of 43.83. However, DT 1 (70% cassava peels and 30% yellow maize) gave the poorest SGR of 1.61 and FCR of 67.58. The difference in weight gain among the fry fed the three levels of cassava peels diet and the control was not statically significant (P>0.05)

Nutritional habits and food consumption pattern of fishing communities around Lake Kainji, Nigeria

The findings are presented of a nutrition survey carried out in the framework of the Nigerian-German Kainji Lake Fisheries Promotion Project in March/April 1995. The major aim of the study was to obtain closer information on the food habits and food consumption patterns of the fishing population, especially the quantification of fish and small fish in particular, consumed at the household level. A standardized questionnaire was used to collect relevant information on demographic, education and occupational profiles; the information was collected in 39 villages in the southern sector of Kainji Lake. (PDF contains 75 pages)

The nutritional status of pre-school children, Kainji Lake communities, Nigeria: a baseline survey

This survey was carried out to provide the Kainji Lake Fisheries Promotion Project (KLFPP), whose overall goal is the improvement of the standard of living of fishing communities around Kainji Lake, Nigeria, and an increase in the availability of fish to consumers, with nutritional status baseline data for long-term monitoring and evaluation of the overall project goal. In a cross-sectional survey, baseline anthropometric data was collected from 768 children, aged 3-60 months in 389 fisherfolk households around the southern sector of Kainji Lake, Nigeria. In addition, data was collected on the nutritional status and fertility of the mothers, vaccination coverage of children and child survival indicators. For control purposes, 576 children and 292 mothers from non-fishing households around Kainji Lake were likewise covered by the survey. A standardised questionnaire was used to collect relevant information, while anthropometric measurements were made using appropriate equipment. Data compilation and analysis was carried out with DATAEASE registered and EPI-INFO registered software, using NCHS reference data for the analysis of anthropometric measurements. The prevalence of stunted children in fishing households was high at 40%, while the prevalence of wasted and underweight children was likewise high at 10% and 29% respectively. Children from non-fishing households had a marginally lower prevalence of stunting, wasting and underweight with 37%, 7% and 25 % respectively, although these differences were not statistically significant. Considering the fact that the survey was carried out during a period of relative food abundance, the prevalence of wasting and underweight children is likely to be much higher during periods of food shortage. The prevalence of stunting, wasting and underweight was relatively high for children aged 3 to 23 months, suggesting an increased risk of malnutrition during this period, most likely associated with inadequate weaning practices. The prevalence of malnourishment amongst women of child-bearing age was relatively high, irrespective of occupation of the household, with an average of 11% undernourished and 6% wasted. Vaccination coverage was very low while infant and child mortality were extremely high with about 1 in 5 children dying before their fifth birthday. Based on the ethical obligation to maximise the potential benefits of the survey, recommendations for activities to improve community nutrition and health were made for communication to relevant authorities. (PDF contains 52 pages)

Post harvest losses in the fisheries of Kainji Lake

An assessment is given of the post-harvest losses in the Lake Kainji fisheries of Nigeria. The study focussed on quantifiable information on post-harvest technology and post-harvest losses from fisherfolk, fish processors and fish traders operating within the Kainji Lake basin. The information was obtained from questionnaires sent to a total of 665 respondents, comprising 317 fishermen, 115 fish processors, 125 fish buyers, and 111 fish sellers in 45 fishing villages and collection centres within the lake basin. Considering the total catch from gillnets, longlines, traps and cast nets estimated at 14,000 in 1995 about 1,000 t of fish was either discarded or lost value due to spoilage during handling by fisherfolk. Assuming an average prices of 80 Naira/kg of fish, the loss to the economy amounted to 80 million Naira annually. Appropriate recommendations are made to significantly reduce post-harvest losses in the Kainji Lake fishery. (PDF contains 91 pages)

The nutritional status of pre-school children, Kainji Lake communities, Nigeria: a follow-up survey

This survey was carried out to provide the Kainji Lake Fisheries Promotion Project (KLFPP), whose overall goal is the improvement of the standard of living of fishing communities around Kainji Lake, Nigeria, managing the fisheries on a sustainable basis, with follow-up data for long-term monitoring and evaluation of the overall project goal. A similar survey, conducted in 1996, provided the baseline against which data from the current survey was evaluated. In a cross-sectional survey, anthropometric data was collected from 576 children aged 3-60 months in 282 fisherfolk households around the southern sector of Kainji Lake, Nigeria. In addition, data was collected on the nutritional status and fertility of the mothers, vaccination coverage of children and child survival indicators. For control purposes, 374 children and 181 mothers from non-fishing households around Kainji Lake were likewise covered by the survey. A standardised questionnaire was used to collect relevant data, while anthropometric measurements were made using appropriate equipment. Data compilation and analysis was carried out with a specially designed Microsoft Access application, using NCHS reference data for the analysis of anthropometric measurements. Statistical significance testing was done using EPI-INFO" software. The results of the follow-up survey indicate a slight increase in the percentage of stunted pre-school children in fishing households around Kainji Lake, from 40% in 1996 to 41% in 1999. This increase is however not statistically significant (p= 0.704). Over the same period, the percentage of stunted children in non-fishing households increased from 37% to 39% (p= 0.540), which is also not statistically significant. Likewise, there were no statistically significant differences between the 1996 and 1999 results for the prevalence of either wasted or underweight children in fishing households. The same applies to children from non-fishing households. In addition, vaccination coverage remains very low while infant and child mortality rates continue to be extremely high with about 1 in 5 children dying before their fifth birthday. There has been no perceptible and lasting improvement in the standard of living of fishing households over the course of the second project phase as indicated by the persistently high prevalence of stunting. The situation is the same for the control group, indicating that for the region as a whole, a number of factors beyond the immediate influence of the project continue to negatively impact on the standard of living. The results also show that the project activities have not had any negative long-term effect on the nutritional status of the beneficiaries. (PDF contains 44 pages)