Joint multiple dictionary learning for tensor sparse coding

Traditional dictionary learning algorithms are used for finding a sparse representation on high dimensional data by transforming samples into a one-dimensional (1D) vector. This 1D model loses the inherent spatial structure property of data. An alternative solution is to employ Tensor Decomposition for dictionary learning on their original structural form —a tensor— by learning multiple dictionaries along each mode and the corresponding sparse representation in respect to the Kronecker product of these dictionaries. To learn tensor dictionaries along each mode, all the existing methods update each dictionary iteratively in an alternating manner. Because atoms from each mode dictionary jointly make contributions to the sparsity of tensor, existing works ignore atoms correlations between different mode dictionaries by treating each mode dictionary independently. In this paper, we propose a joint multiple dictionary learning method for tensor sparse coding, which explores atom correlations for sparse representation and updates multiple atoms from each mode dictionary simultaneously. In this algorithm, the Frequent-Pattern Tree (FP-tree) mining algorithm is employed to exploit frequent atom patterns in the sparse representation. Inspired by the idea of K-SVD, we develop a new dictionary update method that jointly updates elements in each pattern. Experimental results demonstrate our method outperforms other tensor based dictionary learning algorithms.

Discriminative pattern mining in software fault detection

We present a method to enhance fault localization for software systems based on a frequent pattern mining algorithm. Our method is based on a large set of test cases for a given set of programs in which faults can be detected. The test executions are recorded as function call trees. Based on test oracles the tests can be classified into successful and failing tests. A frequent pattern mining algorithm is used to identify frequent subtrees in successful and failing test executions. This information is used to rank functions according to their likelihood of containing a fault. The ranking suggests an order in which to examine the functions during fault analysis. We validate our approach experimentally using a subset of Siemens benchmark programs.

Efficient mining of discriminative molecular fragments

Frequent pattern discovery in structured data is receiving an increasing attention in many application areas of sciences. However, the computational complexity and the large amount of data to be explored often make the sequential algorithms unsuitable. In this context high performance distributed computing becomes a very interesting and promising approach. In this paper we present a parallel formulation of the frequent subgraph mining problem to discover interesting patterns in molecular compounds. The application is characterized by a highly irregular tree-structured computation. No estimation is available for task workloads, which show a power-law distribution in a wide range. The proposed approach allows dynamic resource aggregation and provides fault and latency tolerance. These features make the distributed application suitable for multi-domain heterogeneous environments, such as computational Grids. The distributed application has been evaluated on the well known National Cancer Institute’s HIV-screening dataset.

Genetic evidence of frequent long-distance recruitment in a vertebrate-dispersed tree

The importance of dispersal for the maintenance of biodiversity, while long-recognized, has remained unresolved. We used molecular markers to measure effective dispersal in a natural population of the vertebrate-dispersed Neotropical tree, Simarouba amara (Simaroubaceae) by comparing the distances between maternal parents and their offspring and comparing gene movement via seed and pollen in the 50 ha plot of the Barro Colorado Island forest, Central Panama. In all cases (parent-pair, mother-offspring, father-offspring, sib-sib) distances between related pairs were significantly greater than distances to nearest possible neighbours within each category. Long-distance seedling establishment was frequent: 74% of assigned seedlings established > 100 m from the maternal parent [mean = 392 +/- 234.6 m (SD), range = 9.3-1000.5 m] and pollen-mediated gene flow was comparable to that of seed [mean = 345.0 +/- 157.7 m (SD), range 57.6-739.7 m]. For S. amara we found approximately a 10-fold difference between distances estimated by inverse modelling and mean seedling recruitment distances (39 m vs. 392 m). Our findings have important implications for future studies in forest demography and regeneration, with most seedlings establishing at distances far exceeding those demonstrated by negative density-dependent effects.

A model analysis of climate and CO2 controls on tree growth in a semi-arid woodland

We used a light-use efficiency model of photosynthesis coupled with a dynamic carbon allocation and tree-growth model to simulate annual growth of the gymnosperm Callitris columellaris in the semi-arid Great Western Woodlands, Western Australia, over the past 100 years. Parameter values were derived from independent observations except for sapwood specific respiration rate, fine-root turnover time, fine-root specific respiration rate and the ratio of fine-root mass to foliage area, which were estimated by Bayesian optimization. The model reproduced the general pattern of interannual variability in radial growth (tree-ring width), including the response to the shift in precipitation regimes that occurred in the 1960s. Simulated and observed responses to climate were consistent. Both showed a significant positive response of tree-ring width to total photosynthetically active radiation received and to the ratio of modeled actual to equilibrium evapotranspiration, and a significant negative response to vapour pressure deficit. However, the simulations showed an enhancement of radial growth in response to increasing atmospheric CO2 concentration (ppm) ([CO2]) during recent decades that is not present in the observations. The discrepancy disappeared when the model was recalibrated on successive 30-year windows. Then the ratio of fine-root mass to foliage area increases by 14% (from 0.127 to 0.144 kg C m-2) as [CO2] increased while the other three estimated parameters remained constant. The absence of a signal of increasing [CO2] has been noted in many tree-ring records, despite the enhancement of photosynthetic rates and water-use efficiency resulting from increasing [CO2]. Our simulations suggest that this behaviour could be explained as a consequence of a shift towards below-ground carbon allocation.

Struggles over family land? Tree crops, land and labour in Ghana's Brong-Ahafo region

Agricultural land use in much of Brong-Ahafo region, Ghana has been shifting from the production of food crops towards increased cashew nut cultivation in recent years. This article explores everyday, less visible, gendered and generational struggles over family farms in West Africa, based on qualitative, participatory research in a rural community that is becoming increasingly integrated into the global capitalist system. As a tree crop, cashew was regarded as an individual man's property to be passed on to his wife and children rather than to extended family members, which differed from the communal land tenure arrangements governing food crop cultivation. The tendency for land, cash crops and income to be controlled by men, despite women's and young people's significant labour contributions to family farms, and for women to rely on food crop production for their main source of income and for household food security, means that women and girls are more likely to lose out when cashew plantations are expanded to the detriment of land for food crops. Intergenerational tensions emerged when young people felt that their parents and elders were neglecting their views and concerns. The research provides important insights into gendered and generational power relations regarding land access, property rights and intra-household decision-making processes. Greater dialogue between genders and generations may help to tackle unequal power relations and lead to shared decision-making processes that build the resilience of rural communities.

Analysis of dietary pattern impact on weight status for personalised nutrition through on-line advice: the Food4Me Spanish cohort

Obesity prevalence is increasing. The management of this condition requires a detailed analysis of the global risk factors in order to develop personalised advice. This study is aimed to identify current dietary patterns and habits in Spanish population interested in personalised nutrition and investigate associations with weight status. Self-reported dietary and anthropometrical data from the Spanish participants in the Food4Me study, were used in a multidimensional exploratory analysis to define specific dietary profiles. Two opposing factors were obtained according to food groups’ intake: Factor 1 characterised by a more frequent consumption of traditionally considered unhealthy foods; and Factor 2, where the consumption of “Mediterranean diet” foods was prevalent. Factor 1 showed a direct relationship with BMI (β = 0.226; r2 = 0.259; p < 0.001), while the association with Factor 2 was inverse (β = −0.037; r2 = 0.230; p = 0.348). A total of four categories were defined (Prudent, Healthy, Western, and Compensatory) through classification of the sample in higher or lower adherence to each factor and combining the possibilities. Western and Compensatory dietary patterns, which were characterized by high-density foods consumption, showed positive associations with overweight prevalence. Further analysis showed that prevention of overweight must focus on limiting the intake of known deleterious foods rather than exclusively enhance healthy products.