Validation based sparse Gaussian processes for ordinal regression

This paper proposes a sparse modeling approach to solve ordinal regression problems using Gaussian processes (GP). Designing a sparse GP model is important from training time and inference time viewpoints. We first propose a variant of the Gaussian process ordinal regression (GPOR) approach, leave-one-out GPOR (LOO-GPOR). It performs model selection using the leave-one-out cross-validation (LOO-CV) technique. We then provide an approach to design a sparse model for GPOR. The sparse GPOR model reduces computational time and storage requirements. Further, it provides faster inference. We compare the proposed approaches with the state-of-the-art GPOR approach on some benchmark data sets. Experimental results show that the proposed approaches are competitive.

Elastic Resources Framework in IaaS, preserving performance SLAs

Elasticity in cloud systems provides the flexibility to acquire and relinquish computing resources on demand. However, in current virtualized systems resource allocation is mostly static. Resources are allocated during VM instantiation and any change in workload leading to significant increase or decrease in resources is handled by VM migration. Hence, cloud users tend to characterize their workloads at a coarse grained level which potentially leads to under-utilized VM resources or under performing application. A more flexible and adaptive resource allocation mechanism would benefit variable workloads, such as those characterized by web servers. In this paper, we present an elastic resources framework for IaaS cloud layer that addresses this need. The framework provisions for application workload forecasting engine, that predicts at run-time the expected demand, which is input to the resource manager to modulate resource allocation based on the predicted demand. Based on the prediction errors, resources can be over-allocated or under-allocated as compared to the actual demand made by the application. Over-allocation leads to unused resources and under allocation could cause under performance. To strike a good trade-off between over-allocation and under-performance we derive an excess cost model. In this model excess resources allocated are captured as over-allocation cost and under-allocation is captured as a penalty cost for violating application service level agreement (SLA). Confidence interval for predicted workload is used to minimize this excess cost with minimal effect on SLA violations. An example case-study for an academic institute web server workload is presented. Using the confidence interval to minimize excess cost, we achieve significant reduction in resource allocation requirement while restricting application SLA violations to below 2-3%.

Give-and-take based peer-to-peer content distribution networks

Content Distribution Networks (CDNs) are widely used to distribute data to large number of users. Traditionally, content is being replicated among a number of surrogate servers, leading to high operational costs. In this context, Peer-to-Peer (P2P) CDNs have emerged as a viable alternative. An issue of concern in P2P networks is that of free riders, i.e., selfish peers who download files and leave without uploading anything in return. Free riding must be discouraged. In this paper, we propose a criterion, the Give-and-Take (G&T) criterion, that disallows free riders. Incorporating the G&T criterion in our model, we study a problem that arises naturally when a new peer enters the system: viz., the problem of downloading a `universe' of segments, scattered among other peers, at low cost. We analyse this hard problem, and characterize the optimal download cost under the G&T criterion. We propose an optimal algorithm, and provide a sub-optimal algorithm that is nearly optimal, but runs much more quickly; this provides an attractive balance between running time and performance. Finally, we compare the performance of our algorithms with that of a few existing P2P downloading strategies in use. We also study the computation time for prescribing the strategy for initial segment and peer selection for the newly arrived peer for various existing and proposed algorithms, and quantify cost-computation time trade-offs.

Regionalization of extreme rainfall in India

Regionalization of extreme rainfall is useful for various applications in hydro-meteorology. There is dearth of regionalization studies on extreme rainfall in India. In this perspective, a set of 25 regions that are homogeneous in 1-, 2-, 3-, 4- and 5-day extreme rainfall is delineated based on seasonality measure of extreme rainfall and location indicators (latitude, longitude and altitude) by using global fuzzy c-means (GFCM) cluster analysis. The regions are validated for homogeneity in L-moment framework. One of the applications of the regions is in arriving at quantile estimates of extreme rainfall at sparsely gauged/ungauged locations using options such as regional frequency analysis (RFA). The RFA involves use of rainfall-related information from gauged sites in a region as the basis to estimate quantiles of extreme rainfall for target locations that resemble the region in terms of rainfall characteristics. A procedure for RFA based on GFCM-delineated regions is presented and its effectiveness is evaluated by leave-one-out cross validation. Error in quantile estimates for ungauged sites is compared with that resulting from the use of region-of-influence (ROI) approach that forms site-specific regions exclusively for quantile estimation. Results indicate that error in quantile estimates based on GFCM regions and ROI are fairly close, and neither of them is consistent in yielding the least error over all the sites. The cluster analysis approach was effective in reducing the number of regions to be delineated for RFA.

Simultaneous perturbation methods for adaptive labor staffing in service systems

We consider the problem of optimizing the workforce of a service system. Adapting the staffing levels in such systems is non-trivial due to large variations in workload and the large number of system parameters do not allow for a brute force search. Further, because these parameters change on a weekly basis, the optimization should not take longer than a few hours. Our aim is to find the optimum staffing levels from a discrete high-dimensional parameter set, that minimizes the long run average of the single-stage cost function, while adhering to the constraints relating to queue stability and service-level agreement (SLA) compliance. The single-stage cost function balances the conflicting objectives of utilizing workers better and attaining the target SLAs. We formulate this problem as a constrained parameterized Markov cost process parameterized by the (discrete) staffing levels. We propose novel simultaneous perturbation stochastic approximation (SPSA)-based algorithms for solving the above problem. The algorithms include both first-order as well as second-order methods and incorporate SPSA-based gradient/Hessian estimates for primal descent, while performing dual ascent for the Lagrange multipliers. Both algorithms are online and update the staffing levels in an incremental fashion. Further, they involve a certain generalized smooth projection operator, which is essential to project the continuous-valued worker parameter tuned by our algorithms onto the discrete set. The smoothness is necessary to ensure that the underlying transition dynamics of the constrained Markov cost process is itself smooth (as a function of the continuous-valued parameter): a critical requirement to prove the convergence of both algorithms. We validate our algorithms via performance simulations based on data from five real-life service systems. For the sake of comparison, we also implement a scatter search based algorithm using state-of-the-art optimization tool-kit OptQuest. From the experiments, we observe that both our algorithms converge empirically and consistently outperform OptQuest in most of the settings considered. This finding coupled with the computational advantage of our algorithms make them amenable for adaptive labor staffing in real-life service systems.

A recursive multi-scaling approach to regional flood frequency analysis

Scaling approaches are widely used by hydrologists for Regional Frequency Analysis (RFA) of floods at ungauged/sparsely gauged site(s) in river basins. This paper proposes a Recursive Multi-scaling (RMS) approach to RFA that overcomes limitations of conventional simple- and multi-scaling approaches. The approach involves identification of a separate set of attributes corresponding to each of the sites (being considered in the study area/region) in a recursive manner according to their importance, and utilizing those attributes to construct effective regional regression relationships to estimate statistical raw moments (SMs) of peak flows. The SMs are then utilized to arrive at parameters of flood frequency distribution and quantile estimate(s) corresponding to target return period(s). Effectiveness of the RMS approach in arriving at flood quantile estimates for ungauged sites is demonstrated through leave-one-out cross-validation experiment on watersheds in Indiana State, USA. Results indicate that the approach outperforms index-flood based Region-of-Influence approach, simple- and multi-scaling approaches and a multiple linear regression method. (C) 2015 Elsevier B.V. All rights reserved.

How to be a fig wasp parasite on the fig-fig wasp mutualism

The nursery pollination mutualism between figs and pollinating fig wasps is based on adaptations that allow wasps to enter the enclosed inflorescences of figs, to facilitate seed set, and to have offspring that develop within the nursery and that leave to enter other inflorescences for pollination. This closed mutualistic system is not immune to parasitic fig wasps. Although the life histories and basic biology of the mutualists have been investigated, the biology of the fig wasp parasites has been severely neglected. This review brings together current knowledge of the many different ways in which parasites can enter the system, and also points to the serious lacunae in our understanding of the intricate interactions between gallers, kleptoparasites, seed eaters and parasitoids within this mutualism.