Approximate inference in Boltzmann machines

Inference in Boltzmann machines is NP-hard in general. As a result approximations are often necessary. We discuss first order mean field and second order Onsager truncations of the Plefka expansion of the Gibbs free energy. The Bethe free energy is introduced and rewritten as a Gibbs free energy. From there a convergent belief optimization algorithm is derived to minimize the Bethe free energy. An analytic expression for the linear response estimate of the covariances is found which is exact on Boltzmann trees. Finally, a number of theorems is proven concerning the Plefka expansion, relating the first order mean field and the second order Onsager approximation to the Bethe approximation. Experiments compare mean field approximation, Onsager approximation, belief propagation and belief optimization.     

Data Communications for Internet Robots

For Internet robots, the most challenging and distinct difficulties are well recognized to be associated with Internet transmission delays, delay jitter and not-guaranteed bandwidth, which might degrade system performance dramatically or even lead to instability. Current approaches to data communication between the remote robot and the human operator employ directly one of the two currently available protocols, i.e. TCP and UPD. In this paper, a teleoperation-oriented data transmission mechanism is implemented. Compared to TCP, the presented scheme provides minimized transmission delays and delay jitter; in the steady state, its transmission rate is much smoother; when available network bandwidth changes, it adapts to the variation quickly without large overshoot. Compared to UDP, it is inter-protocol fairness convergent, intro-protocol convergent, and efficient convergent. The presented mechanism is deployed in a mobile robot teleoperation system developed. In the experiments, users successfully guided a Pioneer-2 mobile robot through a laboratory environment remotely over the Internet via a web browser.     

Family-Focused Preventive Interventions: Evaluating Parental Risk Moderation of Substance Use Trajectories.

Four years of longitudinal data from 373 families participating in a randomized intervention-control clinical trial were used to examine whether intervention effects on adolescent alcohol and tobacco use trajectories were moderated by family risk, as defined by parental social emotional maladjustment. Consistent with earlier outcome evaluations based on analyses of covariance, analyses confirmed that both the Preparing for the Drug Free Years program and the Iowa Strengthening Families Program favorably influenced alcohol use index trajectories across the time frame of the study; only the latter program, however, evidenced positive effects on a tobacco use index. Concerning the primary research question, analyses provided no support for family risk moderation of any intervention effect. Findings indicate the feasibility of developing universal preventive interventions that offer comparable benefits to all families. (PsycINFO Database Record (c) 2010 APA, all rights reserved)     

Über das Protein in Kartoffel-, Mais- und ReisstärkeHerrn Prof. Dr. Erwin Welte,Direktor des Instituts für Agrikulturchemie der Universität Göttingen,zum 60. Geburtstag gewidmet.

On the Protein in Potato, Maize and Rice Starch The protein content of potato, maize and rice starch was determined, and 21 amino acids could be identified. With the usual hydrochloric acid hydrolysis to set free the amino acids a portion (residual protein) remains in the dark residue (melanine), and can be isolated only after destruction of the latter, e.g. with sodium chlorite, and analyzed for amino acids. As shown by the nitrogen determination after Kjeldahl and amino acid analysis potato starch has the lowest, maize starch a mean and rice starch a high content of amino acids. The total amount of nitrogen in the amino acids compared with the Kjeldahl nitrogen is only 83% for potato protein and 80% for maize and rice protein. The residue from hydrolysis can vary substantially.     

Next-Generation Grade and Survival Expression Biomarkers of Human Gliomas Based on Algorithmically Reconstructed Molecular Pathways

In gliomas, expression of certain marker genes is strongly associated with survival and tumor type and often exceeds histological assessments. Using a human interactome model, we algorithmically reconstructed 7494 new-type molecular pathways that are centered each on an individual protein. Each single-gene expression and gene-centric pathway activation was tested as a survival and tumor grade biomarker in gliomas and their diagnostic subgroups (IDH mutant or wild type, IDH mutant with 1p/19q co-deletion, MGMT promoter methylated or unmethylated), including the three major molecular subtypes of glioblastoma (proneural, mesenchymal, classical). We used three datasets from The Cancer Genome Atlas and the Chinese Glioma Genome Atlas, which in total include 527 glioblastoma and 1097 low grade glioma profiles. We identified 2724 such gene and 2418 pathway survival biomarkers out of total 17,717 genes and 7494 pathways analyzed. We then assessed tumor grade and molecular subtype biomarkers and with the threshold of AUC > 0.7 identified 1322/982 gene biomarkers and 472/537 pathway biomarkers. This suggests roughly two times greater efficacy of the reconstructed pathway approach compared to gene biomarkers. Thus, we conclude that activation levels of algorithmically reconstructed gene-centric pathways are a potent class of new-generation diagnostic and prognostic biomarkers for gliomas.     

MDS-based segmentation model for the fusion of contour and texture cues in natural images

In this paper, we present an original image segmentation model based on a preliminary spatially adaptive non-linear data dimensionality reduction step integrating contour and texture cues. This new dimensionality reduction model aims at converting an input texture image into a noisy color image in order to greatly simplify its subsequent segmentation. In this latter de-texturing model, the (spatially adaptive) non-local constraints based on edge and contour cues allows us to efficiently regularize the reduced data (or the resulting de-textured color image) and to efficiently combine inhomogeneous region and edge based features in a data fusion/reduction model used as pre-processing step for a final segmentation task. In addition, a set of color/texture and edge-based adaptive spatial continuity constraints is imposed during the segmentation step. These improvements lead to an appealing and powerful two-step adaptive segmentation model, integrating contour and texture cues. Extensive experimental evaluation on the Berkeley image segmentation database demonstrates the efficiency of this hybrid segmentation model in terms of classification accuracy of pairwise pixels in the resulting segmentation map and in the precision–recall framework widespread used for evaluating contour detectors.     

Linear response algorithms for approximate inference in graphical models

Belief propagation (BP) on cyclic graphs is an efficient algorithm for computing approximate marginal probability distributions over single nodes and neighboring nodes in the graph. However, it does not prescribe a way to compute joint distributions over pairs of distant nodes in the graph. In this article, we propose two new algorithms for approximating these pairwise probabilities, based on the linear response theorem. The first is a propagation algorithm that is shown to converge if BP converges to a stable fixed point. The second algorithm is based on matrix inversion. Applying these ideas to gaussian random fields, we derive a propagation algorithm for computing the inverse of a matrix.     

模块化风机机械动力传动装置的先进解决方案

时下模块化风机齿轮箱的可靠性是一个普遍存在的问题。为此,Maag开发和试制了一种新型Maag DPPV-7-79齿轮箱,在平衡刚度和柔性的基础上能更好地实现载荷的分配均匀,具有较小的应力和最佳齿轮接触模式。