Self-Organizing Information Fusion and Hierarchical Knowledge Discovery: A New Framework Using Artmap Neural Networks

Classifying novel terrain or objects from sparse, complex data may require the resolution of conflicting information from sensors woring at different times, locations, and scales, and from sources with different goals and situations. Information fusion methods can help resolve inconsistencies, as when eveidence variously suggests that and object's class is car, truck, or airplane. The methods described her address a complementary problem, supposing that information from sensors and experts is reliable though inconsistent, as when evidence suggests that an object's class is car, vehicle, and man-made. Underlying relationships among classes are assumed to be unknown to the autonomated system or the human user. The ARTMAP information fusion system uses distributed code representations that exploit the neural network's capacity for one-to-many learning in order to produce self-organizing expert systems that discover hierachical knowlege structures. The fusion system infers multi-level relationships among groups of output classes, without any supervised labeling of these relationships. The procedure is illustrated with two image examples, but is not limited to image domain.

Fusion Artmap: A Neural Network Architecture for Multi-Channel Data Fusion and Classification

Fusion ARTMAP is a self-organizing neural network architecture for multi-channel, or multi-sensor, data fusion. Single-channel Fusion ARTMAP is functionally equivalent to Fuzzy ART during unsupervised learning and to Fuzzy ARTMAP during supervised learning. The network has a symmetric organization such that each channel can be dynamically configured to serve as either a data input or a teaching input to the system. An ART module forms a compressed recognition code within each channel. These codes, in turn, become inputs to a single ART system that organizes the global recognition code. When a predictive error occurs, a process called paraellel match tracking simultaneously raises vigilances in multiple ART modules until reset is triggered in one of them. Parallel match tracking hereby resets only that portion of the recognition code with the poorest match, or minimum predictive confidence. This internally controlled selective reset process is a type of credit assignment that creates a parsimoniously connected learned network. Fusion ARTMAP's multi-channel coding is illustrated by simulations of the Quadruped Mammal database.

Fusion ARTMAP: An Adaptive Fuzzy Network for Multi-Channel Classification

Fusion ARTMAP is a self-organizing neural network architecture for multi-channel, or multi-sensor, data fusion. Fusion ARTMAP generalizes the fuzzy ARTMAP architecture in order to adaptively classify multi-channel data. The network has a symmetric organization such that each channel can be dynamically configured to serve as either a data input or a teaching input to the system. An ART module forms a compressed recognition code within each channel. These codes, in turn, beco1ne inputs to a single ART system that organizes the global recognition code. When a predictive error occurs, a process called parallel match tracking simultaneously raises vigilances in multiple ART modules until reset is triggered in one of thmn. Parallel match tracking hereby resets only that portion of the recognition code with the poorest match, or minimum predictive confidence. This internally controlled selective reset process is a type of credit assignment that creates a parsimoniously connected learned network.

Identification of ribosomal proteins specific to higher eukaryotic organisms

This report describes the identification of a novel protein named PS1D (Genbank accession number ), which is composed of an S1-like RNA-binding domain, a (cysteine)x3-(histidine) CCCH-zinc finger, and a very basic carboxyl domain. PS1D is expressed as two isoforms, probably resulting from the alternative splicing of mRNA. The long PS1D isoform differs from the short one by the presence of 48 additional amino acids at its amino-terminal extremity. Analysis of PS1D subcellular distribution by cell fractionation reveals that this protein belongs to the core of the eukaryotic 60S ribosomal subunit. Interestingly, PS1D protein is a highly conserved protein among mammalians as murine, human, and simian PS1D homologues share more than 95% identity. In contrast, no homologous protein is found in lower eukaryotes such as yeast and Caenorhabditis elegans. These observations indicate that PS1D is the first eukaryotic ribosomal protein that is specific to higher eukaryotes.

Nuclear ingredients for cross section calculation of exotic nuclei

The increasing need for cross sections far from the valley of stability, especially for applications such as nuclear astrophysics, poses a challenge for nuclear reaction models. So far, predictions of cross sections have relied on more or less phenomenological approaches, depending on parameters adjusted to available experimental data or deduced from systematic relations. While such predictions are expected to be reliable for nuclei not too far from the experimentally known regions, it is clearly preferable to use more fundamental approaches, based on sound physical bases, when dealing with very exotic nuclei. Thanks to the high computer power available today, all major ingredients required to model a nuclear reaction can now be (and have been) microscopically (or semi-microscopically) determined starting from the information provided by an effective nucleon-nucleon interaction. All these microscopic ingredients have been included in the latest version of the TALYS nuclear reaction code (http://www.talys.eu/).

Latest development of the combinatorial model of nuclear level densities

The combinatorial model of nuclear level densities has now reached a level of accuracy comparable to that of the best global analytical expressions without suffering from the limits imposed by the statistical hypothesis on which the latter expressions rely. In particular, it provides, naturally, non-Gaussian spin distribution as well as non-equipartition of parities which are known to have an impact on cross section predictions at low energies [1, 2, 3]. Our previous global models developed in Refs. [1, 2] suffered from deficiencies, in particular in the way the collective effects - both vibrational and rotational - were treated. We have recently improved this treatment using simultaneously the single-particle levels and collective properties predicted by a newly derived Gogny interaction [4], therefore enabling a microscopic description of energy-dependent shell, pairing and deformation effects. In addition for deformed nuclei, the transition to sphericity is coherently taken into account on the basis of a temperature-dependent Hartree-Fock calculation which provides at each temperature the structure properties needed to build the level densities. This new method is described and shown to give promising results with respect to available experimental data.

Optimized approach to decision fusion of heterogeneous data for breast cancer diagnosis.

As more diagnostic testing options become available to physicians, it becomes more difficult to combine various types of medical information together in order to optimize the overall diagnosis. To improve diagnostic performance, here we introduce an approach to optimize a decision-fusion technique to combine heterogeneous information, such as from different modalities, feature categories, or institutions. For classifier comparison we used two performance metrics: The receiving operator characteristic (ROC) area under the curve [area under the ROC curve (AUC)] and the normalized partial area under the curve (pAUC). This study used four classifiers: Linear discriminant analysis (LDA), artificial neural network (ANN), and two variants of our decision-fusion technique, AUC-optimized (DF-A) and pAUC-optimized (DF-P) decision fusion. We applied each of these classifiers with 100-fold cross-validation to two heterogeneous breast cancer data sets: One of mass lesion features and a much more challenging one of microcalcification lesion features. For the calcification data set, DF-A outperformed the other classifiers in terms of AUC (p < 0.02) and achieved AUC=0.85 +/- 0.01. The DF-P surpassed the other classifiers in terms of pAUC (p < 0.01) and reached pAUC=0.38 +/- 0.02. For the mass data set, DF-A outperformed both the ANN and the LDA (p < 0.04) and achieved AUC=0.94 +/- 0.01. Although for this data set there were no statistically significant differences among the classifiers' pAUC values (pAUC=0.57 +/- 0.07 to 0.67 +/- 0.05, p > 0.10), the DF-P did significantly improve specificity versus the LDA at both 98% and 100% sensitivity (p < 0.04). In conclusion, decision fusion directly optimized clinically significant performance measures, such as AUC and pAUC, and sometimes outperformed two well-known machine-learning techniques when applied to two different breast cancer data sets.

Co-regulation of nuclear respiratory factor-1 by NFkappaB and CREB links LPS-induced inflammation to mitochondrial biogenesis.

The nuclear respiratory factor-1 (NRF1) gene is activated by lipopolysaccharide (LPS), which might reflect TLR4-mediated mitigation of cellular inflammatory damage via initiation of mitochondrial biogenesis. To test this hypothesis, we examined NRF1 promoter regulation by NFκB, and identified interspecies-conserved κB-responsive promoter and intronic elements in the NRF1 locus. In mice, activation of Nrf1 and its downstream target, Tfam, by Escherichia coli was contingent on NFκB, and in LPS-treated hepatocytes, NFκB served as an NRF1 enhancer element in conjunction with NFκB promoter binding. Unexpectedly, optimal NRF1 promoter activity after LPS also required binding by the energy-state-dependent transcription factor CREB. EMSA and ChIP assays confirmed p65 and CREB binding to the NRF1 promoter and p65 binding to intron 1. Functionality for both transcription factors was validated by gene-knockdown studies. LPS regulation of NRF1 led to mtDNA-encoded gene expression and expansion of mtDNA copy number. In cells expressing plasmid constructs containing the NRF-1 promoter and GFP, LPS-dependent reporter activity was abolished by cis-acting κB-element mutations, and nuclear accumulation of NFκB and CREB demonstrated dependence on mitochondrial H(2)O(2). These findings indicate that TLR4-dependent NFκB and CREB activation co-regulate the NRF1 promoter with NFκB intronic enhancement and redox-regulated nuclear translocation, leading to downstream target-gene expression, and identify NRF-1 as an early-phase component of the host antibacterial defenses.

Cytokinesis proteins Tum and Pav have a nuclear role in Wnt regulation.

Wg/Wnt signals specify cell fates in both invertebrate and vertebrate embryos and maintain stem-cell populations in many adult tissues. Deregulation of the Wnt pathway can transform cells to a proliferative fate, leading to cancer. We have discovered that two Drosophila proteins that are crucial for cytokinesis have a second, largely independent, role in restricting activity of the Wnt pathway. The fly homolog of RacGAP1, Tumbleweed (Tum)/RacGAP50C, and its binding partner, the kinesin-like protein Pavarotti (Pav), negatively regulate Wnt activity in fly embryos and in cultured mammalian cells. Unlike many known regulators of the Wnt pathway, these molecules do not affect stabilization of Arm/beta-catenin (betacat), the principal effector molecule in Wnt signal transduction. Rather, they appear to act downstream of betacat stabilization to control target-gene transcription. Both Tum and Pav accumulate in the nuclei of interphase cells, a location that is spatially distinct from their cleavage-furrow localization during cytokinesis. We show that this nuclear localization is essential for their role in Wnt regulation. Thus, we have identified two modulators of the Wnt pathway that have shared functions in cell division, which hints at a possible link between cytokinesis and Wnt activity during tumorigenesis.

Delimiting species without nuclear monophyly in Madagascar's mouse lemurs.

BACKGROUND: Speciation begins when populations become genetically separated through a substantial reduction in gene flow, and it is at this point that a genetically cohesive set of populations attain the sole property of species: the independent evolution of a population-level lineage. The comprehensive delimitation of species within biodiversity hotspots, regardless of their level of divergence, is important for understanding the factors that drive the diversification of biota and for identifying them as targets for conservation. However, delimiting recently diverged species is challenging due to insufficient time for the differential evolution of characters--including morphological differences, reproductive isolation, and gene tree monophyly--that are typically used as evidence for separately evolving lineages. METHODOLOGY: In this study, we assembled multiple lines of evidence from the analysis of mtDNA and nDNA sequence data for the delimitation of a high diversity of cryptically diverged population-level mouse lemur lineages across the island of Madagascar. Our study uses a multi-faceted approach that applies phylogenetic, population genetic, and genealogical analysis for recognizing lineage diversity and presents the most thoroughly sampled species delimitation of mouse lemur ever performed. CONCLUSIONS: The resolution of a large number of geographically defined clades in the mtDNA gene tree provides strong initial evidence for recognizing a high diversity of population-level lineages in mouse lemurs. We find additional support for lineage recognition in the striking concordance between mtDNA clades and patterns of nuclear population structure. Lineages identified using these two sources of evidence also exhibit patterns of population divergence according to genealogical exclusivity estimates. Mouse lemur lineage diversity is reflected in both a geographically fine-scaled pattern of population divergence within established and geographically widespread taxa, as well as newly resolved patterns of micro-endemism revealed through expanded field sampling into previously poorly and well-sampled regions.

Utility of telomerase-pot1 fusion protein in vascular tissue engineering.

While advances in regenerative medicine and vascular tissue engineering have been substantial in recent years, important stumbling blocks remain. In particular, the limited life span of differentiated cells that are harvested from elderly human donors is an important limitation in many areas of regenerative medicine. Recently, a mutant of the human telomerase reverse transcriptase enzyme (TERT) was described, which is highly processive and elongates telomeres more rapidly than conventional telomerase. This mutant, called pot1-TERT, is a chimeric fusion between the DNA binding protein pot1 and TERT. Because pot1-TERT is highly processive, it is possible that transient delivery of this transgene to cells that are utilized in regenerative medicine applications may elongate telomeres and extend cellular life span while avoiding risks that are associated with retroviral or lentiviral vectors. In the present study, adenoviral delivery of pot1-TERT resulted in transient reconstitution of telomerase activity in human smooth muscle cells, as demonstrated by telomeric repeat amplification protocol (TRAP). In addition, human engineered vessels that were cultured using pot1-TERT-expressing cells had greater collagen content and somewhat better performance in vivo than control grafts. Hence, transient delivery of pot1-TERT to elderly human cells may be useful for increasing cellular life span and improving the functional characteristics of resultant tissue-engineered constructs.

An immunoglobulin C kappa-reactive single chain antibody fusion protein induces tolerance through receptor editing in a normal polyclonal immune system.

Understanding immune tolerance mechanisms is a major goal of immunology research, but mechanistic studies have generally required the use of mouse models carrying untargeted or targeted antigen receptor transgenes, which distort lymphocyte development and therefore preclude analysis of a truly normal immune system. Here we demonstrate an advance in in vivo analysis of immune tolerance that overcomes these shortcomings. We show that custom superantigens generated by single chain antibody technology permit the study of tolerance in a normal, polyclonal immune system. In the present study we generated a membrane-tethered anti-Igkappa-reactive single chain antibody chimeric gene and expressed it as a transgene in mice. B cell tolerance was directly characterized in the transgenic mice and in radiation bone marrow chimeras in which ligand-bearing mice served as recipients of nontransgenic cells. We find that the ubiquitously expressed, Igkappa-reactive ligand induces efficient B cell tolerance primarily or exclusively by receptor editing. We also demonstrate the unique advantages of our model in the genetic and cellular analysis of immune tolerance.

Mitochondrial fusion is regulated by Reaper to modulate Drosophila programmed cell death.

In most multicellular organisms, the decision to undergo programmed cell death in response to cellular damage or developmental cues is typically transmitted through mitochondria. It has been suggested that an exception is the apoptotic pathway of Drosophila melanogaster, in which the role of mitochondria remains unclear. Although IAP antagonists in Drosophila such as Reaper, Hid and Grim may induce cell death without mitochondrial membrane permeabilization, it is surprising that all three localize to mitochondria. Moreover, induction of Reaper and Hid appears to result in mitochondrial fragmentation during Drosophila cell death. Most importantly, disruption of mitochondrial fission can inhibit Reaper and Hid-induced cell death, suggesting that alterations in mitochondrial dynamics can modulate cell death in fly cells. We report here that Drosophila Reaper can induce mitochondrial fragmentation by binding to and inhibiting the pro-fusion protein MFN2 and its Drosophila counterpart dMFN/Marf. Our in vitro and in vivo analyses reveal that dMFN overexpression can inhibit cell death induced by Reaper or γ-irradiation. In addition, knockdown of dMFN causes a striking loss of adult wing tissue and significant apoptosis in the developing wing discs. Our findings are consistent with a growing body of work describing a role for mitochondrial fission and fusion machinery in the decision of cells to die.

Electrostatic Energetics of Bacillus subtilis Ribonuclease P Protein Determined by Nuclear Magnetic Resonance-Based Histidine pKa Measurements.

The pKa values of ionizable groups in proteins report the free energy of site-specific proton binding and provide a direct means of studying pH-dependent stability. We measured histidine pKa values (H3, H22, and H105) in the unfolded (U), intermediate (I), and sulfate-bound folded (F) states of RNase P protein, using an efficient and accurate nuclear magnetic resonance-monitored titration approach that utilizes internal reference compounds and a parametric fitting method. The three histidines in the sulfate-bound folded protein have pKa values depressed by 0.21 ± 0.01, 0.49 ± 0.01, and 1.00 ± 0.01 units, respectively, relative to that of the model compound N-acetyl-l-histidine methylamide. In the unliganded and unfolded protein, the pKa values are depressed relative to that of the model compound by 0.73 ± 0.02, 0.45 ± 0.02, and 0.68 ± 0.02 units, respectively. Above pH 5.5, H22 displays a separate resonance, which we have assigned to I, whose apparent pKa value is depressed by 1.03 ± 0.25 units, which is ∼0.5 units more than in either U or F. The depressed pKa values we observe are consistent with repulsive interactions between protonated histidine side chains and the net positive charge of the protein. However, the pKa differences between F and U are small for all three histidines, and they have little ionic strength dependence in F. Taken together, these observations suggest that unfavorable electrostatics alone do not account for the fact that RNase P protein is intrinsically unfolded in the absence of ligand. Multiple factors encoded in the P protein sequence account for its IUP property, which may play an important role in its function.

The evolutionary history of ferns inferred from 25 low-copy nuclear genes.

UNLABELLED: • PREMISE OF THE STUDY: Understanding fern (monilophyte) phylogeny and its evolutionary timescale is critical for broad investigations of the evolution of land plants, and for providing the point of comparison necessary for studying the evolution of the fern sister group, seed plants. Molecular phylogenetic investigations have revolutionized our understanding of fern phylogeny, however, to date, these studies have relied almost exclusively on plastid data.• METHODS: Here we take a curated phylogenomics approach to infer the first broad fern phylogeny from multiple nuclear loci, by combining broad taxon sampling (73 ferns and 12 outgroup species) with focused character sampling (25 loci comprising 35877 bp), along with rigorous alignment, orthology inference and model selection.• KEY RESULTS: Our phylogeny corroborates some earlier inferences and provides novel insights; in particular, we find strong support for Equisetales as sister to the rest of ferns, Marattiales as sister to leptosporangiate ferns, and Dennstaedtiaceae as sister to the eupolypods. Our divergence-time analyses reveal that divergences among the extant fern orders all occurred prior to ∼200 MYA. Finally, our species-tree inferences are congruent with analyses of concatenated data, but generally with lower support. Those cases where species-tree support values are higher than expected involve relationships that have been supported by smaller plastid datasets, suggesting that deep coalescence may be reducing support from the concatenated nuclear data.• CONCLUSIONS: Our study demonstrates the utility of a curated phylogenomics approach to inferring fern phylogeny, and highlights the need to consider underlying data characteristics, along with data quantity, in phylogenetic studies.