Variance inflation in high dimensional Support Vector Machines

Many important machine learning models, supervised and unsupervised, are based on simple Euclidean distance or orthogonal projection in a high dimensional feature space. When estimating such models from small training sets we face the problem that the span of the training data set input vectors is not the full input space. Hence, when applying the model to future data the model is effectively blind to the missed orthogonal subspace. This can lead to an inflated variance of hidden variables estimated in the training set and when the model is applied to test data we may find that the hidden variables follow a different probability law with less variance. While the problem and basic means to reconstruct and deflate are well understood in unsupervised learning, the case of supervised learning is less well understood. We here investigate the effect of variance inflation in supervised learning including the case of Support Vector Machines (SVMS) and we propose a non-parametric scheme to restore proper generalizability. We illustrate the algorithm and its ability to restore performance on a wide range of benchmark data sets.     

Monitoring of Probabilistic Timed Property Sequence Charts

Run‐time monitoring is an important technique to detect erroneous run‐time behaviors. Several techniques have been proposed to automatically generate monitors from specification languages to check temporal and real‐time properties. However, monitoring of probabilistic properties still requires manual generation. To overcome this problem, we define a formal property specification language called Probabilistic Timed Property Sequence Chart (PTPSC). PTPSC is a probabilistic and timed extension of the existing scenario‐based specification formalism Property Sequence Chart (PSC). We have defined a formal grammar‐based syntax and have implemented a syntax‐directed translator that can automatically generate a probabilistic monitor which combines timed B”uchi automata and a sequential statistical hypothesis test process. We validate the generated monitors with a set of experiments performed with our tool WS‐PSC Monitor. Copyright © 2011 John Wiley & Sons, Ltd.     

Can internet users protect themselves? Challenges and techniques of automated protection of HTTP communication

HTTPS enables secure access to web content and web-based services. Although supported by many content and service providers, HTTPS is oftentimes not enabled by default, as pointed out in an open letter sent to Google by security experts. In this article, we discuss if and how web users can protect themselves by using HTTPS instead of HTTP. We show that many websites allow for accessing content by HTTPS instead of HTTP. However, HTTPS access must be manually configured or requested by the user, or is impossible at all, e.g., for embedded objects. For this reason, we explore how to protect users transparently by automatically using HTTPS whenever possible. In order to enable this approach, one needs to determine whether using HTTPS yields the same content as using HTTP, even in the presence of dynamic websites incorporating advertisements and news. We show that this decision is possible for entire websites like amazon.com in short time by combining a fast content comparison algorithm, result caching, and observations on the structure of the website. Besides the concrete HTTP use case considered in this article, our results are of independent interest for any setting in which content can be accessed by various means. Finally, we present and discuss different approaches for implementing automated protection of HTTP connections.     

Hydrothermal vents in Lake Tanganyika harbor spore-forming thermophiles with extremely rapid growth

A thermophilic anaerobic bacterium was isolated from a sublacustrine hydrothermal vent site in Lake Tanganyika (East Africa) with recorded fluid temperatures of 66-103 °C and pH values of 7.7-8.9. The bacterium (strain TR10) was rod-shaped, about 1 by 5 μm in size, and readily formed distal endospores. Based on the 16 S rDNA sequence the novel strain was homologous to Thermobrachium celere and Caloramator indicus, which are closely related. The novel strain was strictly anaerobic, fermentative and had a doubling time as short as 10 min during growth on complex substrates, such as yeast extract and peptone. The optimum temperature for growth was 60 °C, while minimum and maximum temperatures were 40 and 75 °C. The pH response was alkalitolerant with optimum pH at 7.4 and 8.5 depending on the growth medium. The distinct feature of rapid proliferation and endospore formation may allow the novel organism to exploit the temporarily fluctuating growth conditions in the hydrothermal vent environments of Lake Tanganyika.     

A review and comparison of strategies for handling missing values in separate-and-conquer rule learning

In this paper, we review possible strategies for handling missing values in separate-and-conquer rule learning algorithms, and compare them experimentally on a large number of datasets. In particular through a careful study with data with controlled levels of missing values we get additional insights on the strategies’ different biases w.r.t. attributes with missing values. Somewhat surprisingly, a strategy that implements a strong bias against the use of attributes with missing values, exhibits the best average performance on 24 datasets from the UCI repository.     

Efficient model-checking of weighted CTL with upper-bound constraints

We present a symbolic extension of dependency graphs by Liu and Smolka in to model-check weighted Kripke structures against the computation tree logic with upper-bound weight constraints. Our extension introduces a new type of edges into dependency graphs and lifts the computation of fixed-points from boolean domain to nonnegative integers to cope with the weights. We present both global and local algorithms for the fixed-point computation on symbolic dependency graphs and argue for the advantages of our approach compared to the direct encoding of the model-checking problem into dependency graphs. We implement all algorithms in a publicly available tool and evaluate them on several experiments. The principal conclusion is that our local algorithm is the most efficient one with an order of magnitude improvement for model checking problems with a high number of “witnesses”.     

Analysis and evaluation of V*-kNN: an efficient algorithm for moving kNN queries

The moving k nearest neighbor (MkNN) query continuously finds the k nearest neighbors of a moving query point. MkNN queries can be efficiently processed through the use of safe regions. In general, a safe region is a region within which the query point can move without changing the query answer. This paper presents an incremental safe-region-based technique for answering MkNN queries, called the V*-Diagram, as well as analysis and evaluation of its associated algorithm, V*-kNN. Traditional safe-region approaches compute a safe region based on the data objects but independent of the query location. Our approach exploits the knowledge of the query location and the boundary of the search space in addition to the data objects. As a result, V*-kNN has much smaller I/O and computation costs than existing methods. We further provide cost models to estimate the number of data accesses for V*-kNN and a competitive technique, RIS-kNN. The V*-Diagram and V*-kNN are also applicable to the domain of spatial networks and we present algorithms to construct a spatial-network V*-Diagram. Our experimental results show that V*-kNN significantly outperforms the competitive technique. The results also verify the accuracy of the cost models.     

Spatial interpolation in wireless sensor networks: localized algorithms for variogram modeling and Kriging

Wireless sensor networks (WSNs) are rapidly emerging as the prominent technology for monitoring physical phenomena. However, large scale WSNs are known to suffer from coverage holes, i.e., large regions of deployment area where no sensing coverage can be provided. Such holes are the result of hardware failures, extensive costs for redeployment or the hostility of deployment areas. Coverage holes can adversely affect the accurate representation of natural phenomena that are monitored by a WSN. In this work, we propose to exploit the spatial correlation of physical phenomena to make monitoring systems more resilient to coverage holes. We show that a phenomenon can be interpolated inside a coverage hole with a high level of accuracy from the available nodal data given a model of its spatial correlation. However, due to energy limitations of sensor nodes it is imperative to perform this interpolation in an energy efficient manner that minimizes communication among nodes. In this paper, we present highly energy efficient methods for spatial interpolation in WSNs. First, we build a correlation model of the phenomenon being monitored in a distributed manner. Then, a purely localized and distributed spatial interpolation scheme based on Kriging interpolates the phenomenon inside coverage holes. We test the cost and accuracy of our scheme with extensive simulations and show that it is significantly more energy efficient than global interpolations and remarkably more accurate than simple averaging.     

MS analysis of rheumatoid arthritic synovial tissue identifies specific citrullination sites on fibrinogen

Purpose : Citrullination is a post‐translational modification of arginine residues to citrulline catalyzed by peptidyl arginine deiminases. Induced expression of citrullinated proteins are frequently detected in various inflammatory states including arthritis; however, direct detection of citrullination in arthritic samples has not been successfully performed in the past. Experimental design : Citrullination of human fibrinogen, a candidate autoantigen in arthritis, was studied. Accurate identification of citrullinated fibrinogen peptides from rheumatoid arthritis synovial tissue specimens was performed using accurate mass and retention time analysis. Results : A peptide with the sequence ESSSHHPGIAEFPSRGK corresponding to amino acids 559–575 of fibrinogen α‐chain was identified to be citrullinated with an occupancy rate between 1.4 and 2.5%. Citrullination of the peptide KREEAPSLRPAPPPISGGGYRARPAK corresponding to amino acids 52–77 of the fibrinogen β‐chain was identified with an occupancy rate of 1.2%. Conclusions and clinical relevance : We report a proof of principle study for the identification of citrullinated proteins and within them, identification of citrullination sites and quantification of their occupancies in synovial tissue from rheumatoid arthritis patients using high‐resolution MS. Detailed studies on which molecules are citrullinated in arthritis can provide information about their role in immune regulation and serve as novel biomarkers and potentially even as therapeutic targets.     

The maximum gain of increasing the number of preemptions in multiprocessor scheduling

We consider the optimal makespan C(P, m, i) of an arbitrary set P of independent jobs scheduled with i preemptions on a multiprocessor with m identical processors. We compare the ratio for such makespans for i and j preemptions, respectively, where i < j. This ratio depends on P, but we are interested in the P that maximizes this ratio, i.e. we calculate a formula for the worst case ratio G(m, i, j) defined as G(m,i,j)=max\fracC(P,m,i)C(P,m,j),{G(m,i,j)=\max \frac{C(P,m,i)}{C(P,m,j)},} where the maximum is taken over all sets P of independent jobs.