Supervised latent dirichlet allocation models for efficient activity representation

Local spatio-temporal features with a Bag-of-visual words model is a popular approach used in human action recognition. Bag-of-features methods suffer from several challenges such as extracting appropriate appearance and motion features from videos, converting extracted features appropriate for classification and designing a suitable classification framework. In this paper we address the problem of efficiently representing the extracted features for classification to improve the overall performance. We introduce two generative supervised topic models, maximum entropy discrimination LDA (MedLDA) and class- specific simplex LDA (css-LDA), to encode the raw features suitable for discriminative SVM based classification. Unsupervised LDA models disconnect topic discovery from the classification task, hence yield poor results compared to the baseline Bag-of-words framework. On the other hand supervised LDA techniques learn the topic structure by considering the class labels and improve the recognition accuracy significantly. MedLDA maximizes likelihood and within class margins using max-margin techniques and yields a sparse highly discriminative topic structure; while in css-LDA separate class specific topics are learned instead of common set of topics across the entire dataset. In our representation first topics are learned and then each video is represented as a topic proportion vector, i.e. it can be comparable to a histogram of topics. Finally SVM classification is done on the learned topic proportion vector. We demonstrate the efficiency of the above two representation techniques through the experiments carried out in two popular datasets. Experimental results demonstrate significantly improved performance compared to the baseline Bag-of-features framework which uses kmeans to construct histogram of words from the feature vectors.

Optimal allocation of conservation effort among subpopulations of a threatened species: How important is patch quality?

Money is often a limiting factor in conservation, and attempting to conserve endangered species can be costly. Consequently, a framework for optimizing fiscally constrained conservation decisions for a single species is needed. In this paper we find the optimal budget allocation among isolated subpopulations of a threatened species to minimize local extinction probability. We solve the problem using stochastic dynamic programming, derive a useful and simple alternative guideline for allocating funds, and test its performance using forward simulation. The model considers subpopulations that persist in habitat patches of differing quality, which in our model is reflected in different relationships between money invested and extinction risk. We discover that, in most cases, subpopulations that are less efficient to manage should receive more money than those that are more efficient to manage, due to higher investment needed to reduce extinction risk. Our simple investment guideline performs almost as well as the exact optimal strategy. We illustrate our approach with a case study of the management of the Sumatran tiger, Panthera tigris sumatrae, in Kerinci Seblat National Park (KSNP), Indonesia. We find that different budgets should be allocated to the separate tiger subpopulations in KSNP. The subpopulation that is not at risk of extinction does not require any management investment. Based on the combination of risks of extinction and habitat quality, the optimal allocation for these particular tiger subpopulations is an unusual case: subpopulations that occur in higher-quality habitat (more efficient to manage) should receive more funds than the remaining subpopulation that is in lower-quality habitat. Because the yearly budget allocated to the KSNP for tiger conservation is small, to guarantee the persistence of all the subpopulations that are currently under threat we need to prioritize those that are easier to save. When allocating resources among subpopulations of a threatened species, the combined effects of differences in habitat quality, cost of action, and current subpopulation probability of extinction need to be integrated. We provide a useful guideline for allocating resources among isolated subpopulations of any threatened species. © 2010 by the Ecological Society of America.

Resource allocation in two species systems: Is it worth acknowledging species interactions?

It is becoming increasingly popular to consider species interactions when managing ecological foodwebs. Such an approach is useful in determining how management can affect multiple species, with either beneficial or detrimental consequences. Identifying such actions is particularly valuable in the context of conservation decision making as funding is severely limited. This paper outlines a new approach that simplifies the resource allocation problem in a two species system for a range of species interactions: independent, mutualism, predator-prey, and competitive exclusion. We assume that both species are endangered and we do not account for decisions over time. We find that optimal funding allocation is to the conservation of the species with the highest marginal gain in expected probability of survival and that, across all except mutualist interaction types, optimal conservation funding allocation differs between species. Loss in efficiency from ignoring species interactions was most severe in predator-prey systems. The funding problem we address, where an ecosystem includes multiple threatened species, will only become more commonplace as increasing numbers of species worldwide become threatened. © 2011 Elsevier B.V.

A new approach to measure tactical voting: evidence from the British elections

Although tactical voting attracts a great deal of attention, it is very hard to measure as it requires knowledge of both individuals’ voting choices as well as their unobserved preferences. In this article, we present a simple empirical strategy to nonparametrically identify tactical voting patterns directly from balloting results. This approach allows us to study the magnitude and direction of strategic voting as well as to verify which information voters and parties take into account to determine marginal constituencies. We show that tactical voting played a significant role in the 2010 election, mainly for Liberal–Democratic voters supporting Labour. Moreover, our results suggest that voters seem to form their expectations based on a national swing in vote shares rather than newspaper guides published in the main media outlets or previous election outcomes. We also present some evidence that suggests that campaign spending is not driving tactical voting.

On the benefits of equicorrelation for portfolio allocation

The importance of modelling correlation has long been recognised in the field of portfolio management, with largedimensional multivariate problems increasingly becoming the focus of research. This paper provides a straightforward and commonsense approach toward investigating a number of models used to generate forecasts of the correlation matrix for large-dimensional problems.We find evidence in favour of assuming equicorrelation across various portfolio sizes, particularly during times of crisis. During periods of market calm, however, the suitability of the constant conditional correlation model cannot be discounted, especially for large portfolios. A portfolio allocation problem is used to compare forecasting methods. The global minimum variance portfolio and Model Confidence Set are used to compare methods, while portfolio weight stability and relative economic value are also considered.

On optimum stratification with proportional allocation for a class of pareto distributions

Cum ./LSTA_A_8828879_O_XML_IMAGES/LSTA_A_8828879_O_ILM0001.gif rule [Singh (1975)] has been suggested in the literature for finding approximately optimum strata boundaries for proportional allocation, when the stratification is done on the study variable. This paper shows that for the class of density functions arising from the Wang and Aggarwal (1984) representation of the Lorenz Curve (or DBV curves in case of inventory theory), the cum ./LSTA_A_8828879_O_XML_IMAGES/LSTA_A_8828879_O_ILM0002.gif rule in place of giving approximately optimum strata boundaries, yields exactly optimum boundaries. It is also shown that the conjecture of Mahalanobis (1952) “. . .an optimum or nearly optimum solutions will be obtained when the expected contribution of each stratum to the total aggregate value of Y is made equal for all strata” yields exactly optimum strata boundaries for the case considered in the paper.

The Learning Component of Dynamic Allocation Indices

For a multiarmed bandit problem with exponential discounting the optimal allocation rule is defined by a dynamic allocation index defined for each arm on its space. The index for an arm is equal to the expected immediate reward from the arm, with an upward adjustment reflecting any uncertainty about the prospects of obtaining rewards from the arm, and the possibilities of resolving those uncertainties by selecting that arm. Thus the learning component of the index is defined to be the difference between the index and the expected immediate reward. For two arms with the same expected immediate reward the learning component should be larger for the arm for which the reward rate is more uncertain. This is shown to be true for arms based on independent samples from a fixed distribution with an unknown parameter in the cases of Bernoulli and normal distributions, and similar results are obtained in other cases.

Sequential allocation in clinical trials

Suppose two treatments with binary responses are available for patients with some disease and that each patient will receive one of the two treatments. In this paper we consider the interests of patients both within and outside a trial using a Bayesian bandit approach and conclude that equal allocation is not appropriate for either group of patients. It is suggested that Gittins indices should be used (using an approach called dynamic discounting by choosing the discount rate based on the number of future patients in the trial) if the disease is rare, and the least failures rule if the disease is common. Some analytical and simulation results are provided.

Gittins indices and constrained allocation in clinical trials

We explore the use of Gittins indices to search for near optimality in sequential clinical trials. Some adaptive allocation rules are proposed to achieve the following two objectives as far as possible: (i) to reduce the expected successes lost, (ii) to minimize the error probability at the end. Simulation results indicate the merits of the rules based on Gittins indices for small trial sizes. The rules are generalized to the case when neither of the response densities is known. Asymptotic optimality is derived for the constrained rules. A simple allocation rule is recommended for one-stage models. The simulation results indicate that it works better than both equal allocation and Bather's randomized allocation. We conclude with a discussion of possible further developments.

Using global formulary apportionment for international profit allocation: The case of Indonesia's mining industry

My thesis examined an alternative approach, referred to as the unitary taxation approach to the allocation of profit, which arises from the notion that as a multinational group exists as a single economic entity, it should be taxed as one taxable unit. The plausibility of a unitary taxation regime achieving international acceptance and agreement is highly contestable due to its implementation issues, and economic and political feasibility. Using a case-study approach focusing on Freeport-McMoRan and Rio Tinto's mining operations in Indonesia, this thesis compares both tax regimes against the criteria for a good tax system - equity, efficiency, neutrality and simplicity. This thesis evaluates key issues that arise when implementing a unitary taxation approach with formulary apportionment based on the context of mining multinational firms in Indonesia.

Tactical Pasture Management: Enhancing Profits from Poplar Box Country

The project assembled basic information to allow effective management and manipulation of native pastures in the southern Maranoa region of Queensland. This involved a range of plant studies, including a grazing trial, to quantify the costs of poor pasture composition. While the results focus on perennial grasses, we recognise the important dietary role played by broad-leaved herbs. The plant manipulation studies focussed on ways to change the proportions of plants in a grazed pasture, eg. by recruitment or accelerated morbidity of existing plants. As most perennial grasses have a wide range of potential flowering times outside of mid-winter, rainfall exerts the major influence on flowering and seedset; exceptions are black speargrass, rough speargrass and golden beardgrass that flower only for a restricted period each year. This simplifies potential control options through reducing seedset. Data from field growth studies of four pasture grasses have been used to refine the State's pasture production model GRASP. We also provide detailed data on the forage value of many native species at different growth stages. Wiregrass dominance in pastures on a sandy red earth reduced wool value by only 5-10% at Roma in 1994/95 when winters were very dry and grass seed problems were minimal.

Effect of timing of a deficit-irrigation allocation on corn evapotranspiration, yield, water use efficiency and dry mass

Water regulations have decreased irrigation water supplies in Nebraska and some other areas of the USA Great Plains. When available water is not enough to meet crop water requirements during the entire growing cycle, it becomes critical to know the proper irrigation timing that would maximize yields and profits. This study evaluated the effect of timing of a deficit-irrigation allocation (150 mm) on crop evapotranspiration (ETc), yield, water use efficiency (WUE = yield/ETc), irrigation water use efficiency (IWUE = yield/irrigation), and dry mass (DM) of corn (Zea mays L.) irrigated with subsurface drip irrigation in the semiarid climate of North Platte, NE. During 2005 and 2006, a total of sixteen irrigation treatments (eight each year) were evaluated, which received different percentages of the water allocation during July, August, and September. During both years, all treatments resulted in no crop stress during the vegetative period and stress during the reproductive stages, which affected ETc, DM, yield, WUE and IWUE. Among treatments, ETc varied by 7.2 and 18.8%; yield by 17 and 33%; WUE by 12 and 22%, and IWUE by 18 and 33% in 2005 and 2006, respectively. Yield and WUE both increased linearly with ETc and with ETc/ETp (ETp = seasonal ETc with no water stress), and WUE increased linearly with yield. The yield response factor (ky) averaged 1.50 over the two seasons. Irrigation timing affected the DM of the plant, grain, and cob, but not that of the stover. It also affected the percent of DM partitioned to the grain (harvest index), which increased linearly with ETc and averaged 56.2% over the two seasons, but did not affect the percent allocated to the cob or stover. Irrigation applied in July had the highest positive coefficient of determination (R2) with yield. This high positive correlation decreased considerably for irrigation applied in August, and became negative for irrigation applied in September. The best positive correlation between the soil water deficit factor (Ks) and yield occurred during weeks 12-14 from crop emergence, during the "milk" and "dough" growth stages. Yield was poorly correlated to stress during weeks 15 and 16, and the correlation became negative after week 17. Dividing the 150 mm allocation about evenly among July, August and September was a good strategy resulting in the highest yields in 2005, but not in 2006. Applying a larger proportion of the allocation in July was a good strategy during both years, and the opposite resulted when applying a large proportion of the allocation in September. The different results obtained between years indicate that flexible irrigation scheduling techniques should be adopted, rather than relying on fixed timing strategies.

A multi-field bio-economic model of irrigated grain-cotton farming systems.

We present a participatory modelling framework that integrates information from interviews and discussions with farmers and consultants, with dynamic bio-economic models to answer complex questions on the allocation of limited resources at the farm business level. Interviews and discussions with farmers were used to: describe the farm business; identify relevant research questions; identify potential solutions; and discuss and learn from the whole-farm simulations. The simulations are done using a whole-farm, multi-field configuration of APSIM (APSFarm). APSFarm results were validated against farmers' experience. Once the model was accepted by the participating farmers as a fair representation of their farm business, the model was used to explore changes in the tactical or strategic management of the farm and results were then discussed to identify feasible options for improvement. Here we describe the modelling framework and present an example of the application of integrative whole farm system tools to answer relevant questions from an irrigated farm business case study near Dalby (151.27E - 27.17S), Queensland, Australia. Results indicated that even though cotton crops generates more farm income per hectare a more diversified rotation with less cotton would be relatively more profitable, with no increase in risk, as a more cotton dominated traditional rotation. Results are discussed in terms of the benefits and constraints from developing and applying more integrative approaches to represent farm businesses and their management in participatory research projects with the aim of designing more profitable and sustainable irrigated farming systems.

Comparative growth and biomass allocation of two varieties of cat's claw creeper, Dolichandra unguis-cati (Bignoniaceae) in Australia

Introduced as an ornamental vine, cat's claw creeper Dolichandra unguis-cati (syn. Macfadyena unguis-cati) has invaded coastal and subcoastal areas of subtropical eastern Australia. Two varieties have been indentified, one of which ('short-pod') is found throughout south-eastern Australia, while the other ('long-pod') appears to be restricted to several sites in south-eastern Queensland. We compared the growth and biomass allocation patterns of the two varieties in the field over a 22-month period to determine if a higher growth rate and/or more efficient allocation of biomass may contribute to this disparity in distribution. The long-pod variety produced greater aboveground and total biomass than the short-pod variety in both riparian and non-riparian zones. Belowground the two varieties produced a similar number of tubers and overall biomass, though the long-pod variety allocated a smaller portion of its carbon belowground. High growth rates and greater biomass allocation aboveground are characteristic of invasive species, allowing them to outcompete and crowd out existing vegetation. There was no significant site by variety interaction, an indication of consistency in variety performance across riparian and non-riparian sites. Results from our study suggest that differences in growth and biomass allocations are unlikely to have contributed to the disparity in distribution of the two varieties. Despite currently occupying a relatively small range, the long-pod variety may be a more adept invader than the short-pod variety, and could become more prevalent in the future. © 2012 CSIRO.