Sequential anomaly detection in the presence of noise and limited feedback

This paper describes a methodology for detecting anomalies from sequentially observed and potentially noisy data. The proposed approach consists of two main elements: 1) filtering, or assigning a belief or likelihood to each successive measurement based upon our ability to predict it from previous noisy observations and 2) hedging, or flagging potential anomalies by comparing the current belief against a time-varying and data-adaptive threshold. The threshold is adjusted based on the available feedback from an end user. Our algorithms, which combine universal prediction with recent work on online convex programming, do not require computing posterior distributions given all current observations and involve simple primal-dual parameter updates. At the heart of the proposed approach lie exponential-family models which can be used in a wide variety of contexts and applications, and which yield methods that achieve sublinear per-round regret against both static and slowly varying product distributions with marginals drawn from the same exponential family. Moreover, the regret against static distributions coincides with the minimax value of the corresponding online strongly convex game. We also prove bounds on the number of mistakes made during the hedging step relative to the best offline choice of the threshold with access to all estimated beliefs and feedback signals. We validate the theory on synthetic data drawn from a time-varying distribution over binary vectors of high dimensionality, as well as on the Enron email dataset. © 1963-2012 IEEE.

Finite mixture distributions, sequential likelihood and the EM algorithm

A popular way to account for unobserved heterogeneity is to assume that the data are drawn from a finite mixture distribution. A barrier to using finite mixture models is that parameters that could previously be estimated in stages must now be estimated jointly: using mixture distributions destroys any additive separability of the log-likelihood function. We show, however, that an extension of the EM algorithm reintroduces additive separability, thus allowing one to estimate parameters sequentially during each maximization step. In establishing this result, we develop a broad class of estimators for mixture models. Returning to the likelihood problem, we show that, relative to full information maximum likelihood, our sequential estimator can generate large computational savings with little loss of efficiency.

A model of sequential heart and composite tissue allotransplant in rats.

BACKGROUND: Some of the 600,000 patients with solid organ allotransplants need reconstruction with a composite tissue allotransplant, such as the hand, abdominal wall, or face. The aim of this study was to develop a rat model for assessing the effects of a secondary composite tissue allotransplant on a primary heart allotransplant. METHODS: Hearts of Wistar Kyoto rats were harvested and transplanted heterotopically to the neck of recipient Fisher 344 rats. The anastomoses were performed between the donor brachiocephalic artery and the recipient left common carotid artery, and between the donor pulmonary artery and the recipient external jugular vein. Recipients received cyclosporine A for 10 days only. Heart rate was assessed noninvasively. The sequential composite tissue allotransplant consisted of a 3 x 3-cm abdominal musculocutaneous flap harvested from Lewis rats and transplanted to the abdomen of the heart allotransplant recipients. The abdominal flap vessels were connected to the femoral vessels. No further immunosuppression was administered following the composite tissue allotransplant. Ten days after composite tissue allotransplantation, rejection of the heart and abdominal flap was assessed histologically. RESULTS: The rat survival rate of the two-stage transplant surgery was 80 percent. The transplanted heart rate decreased from 150 +/- 22 beats per minute immediately after transplant to 83 +/- 12 beats per minute on day 20 (10 days after stopping immunosuppression). CONCLUSIONS: This sequential allotransplant model is technically demanding. It will facilitate investigation of the effects of a secondary composite tissue allotransplant following primary solid organ transplantation and could be useful in developing future immunotherapeutic strategies.

Scheduling batches with sequential job processing for two-machine flow and open shops

In this paper, we study a problem of scheduling and batching on two machines in a flow-shop and open-shop environment. Each machine processes operations in batches, and the processing time of a batch is the sum of the processing times of the operations in that batch. A setup time, which depends only on the machine, is required before a batch is processed on a machine, and all jobs in a batch remain at the machine until the entire batch is processed. The aim is to make batching and sequencing decisions, which specify a partition of the jobs into batches on each machine, and a processing order of the batches on each machine, respectively, so that the makespan is minimized. The flow-shop problem is shown to be strongly NP-hard. We demonstrate that there is an optimal solution with the same batches on the two machines; we refer to these as consistent batches. A heuristic is developed that selects the best schedule among several with one, two, or three consistent batches, and is shown to have a worst-case performance ratio of 4/3. For the open-shop, we show that the problem is NP-hard in the ordinary sense. By proving the existence of an optimal solution with one, two or three consistent batches, a close relationship is established with the problem of scheduling two or three identical parallel machines to minimize the makespan. This allows a pseudo-polynomial algorithm to be derived, and various heuristic methods to be suggested.

Contingent screening for down syndrome is an efficient alternative to non-disclosure sequential screening

Objective To present a first and second trimester Down syndrome screening strategy, whereby second-trimester marker determination is contingent on the first-trimester results. Unlike non-disclosure sequential screening (the Integrated test), which requires all women to have markers in both trimesters, this allows a large proportion of the women to complete screening in the first trimester. Methods Two first-trimester risk cut-offs defined three types of results: positive and referred for early diagnosis; negative with screening complete; and intermediate, needing second-trimester markers. Multivariate Gaussian modelling with Monte Carlo simulation was used to estimate the false-positive rate for a fixed 85% detection rate. The false-positive rate was evaluated for various early detection rates and early test completion rates. Model parameters were taken from the SURUSS trial. Results Completion of screening in the first trimester for 75% of women resulted in a 30% early detection rate and a 55% second trimester detected rate (net 85%) with a false-positive rate only 0.1% above that achievable by the Integrated test. The screen-positive rate was 0.1% in the first trimester and 4.7% for those continuing to be tested in the second trimester. If the early detection rate were to be increased to 45% or the early completion rate were to be increased to 80%, there would be a further 0.1% increase in the false-positive rate. Conclusion Contingent screening can achieve results comparable with the Integrated test but with earlier completion of screening for most women. Both strategies need to be evaluated in large-scale prospective studies particularly in relation to psychological impact and practicability.

Performance of a Sequential Reactive Barrier for Bioremediation of Coal Tar Contaminated Groundwater

Following a thorough site investigation, a biological Sequential Reactive Barrier (SEREBAR), designed to remove Polycyclic Aromatic Hydrocarbons (PAHs) and BTEX compounds, was installed at a Former Manufactured Gas Plant (FMGP) site. The novel design of the barrier comprises, in series, an interceptor and six reactive chambers. The first four chambers (2 nonaerated-2 aerated) were filled with sand to encourage microbial colonization. Sorbant Granular Activated Carbon (GAC) was present in the final two chambers in order to remove any recalcitrant compounds. The SEREBAR has been in continuous operation for 2 years at different operational flow rates (ranging from 320 L/d to 4000 L/d, with corresponding residence times in each chamber of 19 days and 1.5 days, respectively). Under low flow rate conditions (320-520 L/d) the majority of contaminant removal (>93%) occurred biotically within the interceptor and the aerated chambers. Under high flow rates (1000-4000 L/d) and following the installation of a new interceptor to prevent passive aeration, the majority of contaminant removal (>80%) again occurred biotically within the aerated chambers. The sorption zone (GAC) proved to be an effective polishing step, removing any remaining contaminants to acceptable concentrations before discharge down-gradient of the SEREBAR (overall removals >95%).

Bcl-2 expression in sequential biopsies of potentially malignant oral mucosal lesions assessed by immunocytochemistry

OBJECTIVE: To examine, for the first time Bcl-2 expression in sequential (autogenous) oral mucosal biopsies taken from the same sites in a gender, risk-factor matched, Caucasoid sample, over a 21-year period,

Modification of smooth muscle Ca2+-sparks by tetracaine: evidence for sequential RyR activation.

Spontaneous Ca(2+)-sparks were imaged using confocal line scans of fluo-4 loaded myocytes in retinal arterioles. Tetracaine produced concentration-dependent decreases in spark frequency, and modified the spatiotemporal characteristics of residual sparks. Tetracaine (10 microM) reduced the rate of rise but prolonged the average rise time so that average spark amplitude was unaltered. The mean half-time of spark decay was also unaffected, suggesting that spark termination, although delayed, remained well synchronized. Sparks spread transversely across the myocytes in these vessels, and the speed of spread within individual sparks was slowed by approximately 60% in 10 microM tetracaine, as expected if the spark was propagated across the cell but the average P(o) for RyRs was reduced. Staining of isolated vessels with BODIPY-ryanodine and di-4-ANEPPS showed that RyRs were located both peripherally, adjacent to the plasma membrane, and in transverse extensions of the SR from one side of the cell to the other. Immuno-labelling of retinal flat mounts demonstrated the presence RyR(2) in arteriole smooth muscle but not RyR(1). We conclude that Ca(2+)-sparks in smooth muscle can result from sequential activation of RyRs distributed over an area of several microm(2), rather than from tightly clustered channels as in striated muscle.