On matrices of neural nets

The structure of a complete or incomplete neural net is represented here by several matrices. The activity equation of the net follows in a general form. A chain or cycle is defined as a neural structure whose connection matrix is unitary. We can compute the number of simple chains by a recurrent formula. Academia Sinica     

Beyond the binary case in random nets

Experiments on random binary, ternary, etc. (P=2, 3,…, 10) switching nets are reported. Behavioral cycle lengths are examined as functions of output variety,P, input connectance,K, and net size,N. Overall, output variety appears an influential, well-behaved net property. Strong, but well-behaved interactions appear among net variables. In high connectance nets, median cycle length grows approx. asP ^N/2. Other factors constant, one-connected nets show the shortest cycles, and connectance effects appear to converge asymptotically aroundN. Data for cycle length as a function of net size suggest a concavity not compatible with the Kauffman “square root law” (Kauffman, 1969). Evidence of a positive relationship between cycle length and run-in length is found in two-input nets; weaker evidence is obtained that in higher connectance nets this relationship becomes negative in sign. The “modular complexity” ofP>2 nets is examined briefly.     

Control of switching nets

Switching nets have long been used as models of biological phenomena. In this paper, control properties of switching nets are discussed. Bounds on control parameters for general nets, linear nets and nets with linear inputs are obtained. In particular, bounds are obtained on the maximum distance between two states and on the length of input sequences needed to synchronize two copies of the same net. Bounds on the number of input lines that must be added to an autonomous net to make it controllable are obtained in terms of the number of cycles and transient chains of the net.     

Using distance matrices to choose between competing theories and an application to the origin of modern humans

This paper examines competing theories for cases in which both the data and the hypotheses can be represented as distance matrices. A test due to Dow & Cheverud has been used for such comparisons in anthropology, but when data are spatially, temporally, or phylogenetically autocorrelated, this test may be far too liberal. We examine a classification procedure based on ratios of probabilities obtained from Mantel tests of the competing hypotheses and find that design matrices describing only lag-one connections and those eliminating common connections of competing hypotheses are the most informative. We apply this method to simulated gene-frequency data in a 7×7 chessboard representing a stepping-stone model and discriminate between alternative theories with a 7% misclassification rate. We also apply these techniques to the current controversy concerning the origin of anatomically modern humans by testing design matrices representing regional continuity and single African origins. The outcome for lag-one matrices and those showing only unique lag-one differences indicate that the single African origin of anatomically modern humans fits the distance matrix based on 165 characters of 83 fossil crania better than the competing theory. However, we also tested a design matrix describing single origin out of southwest Asia. This design matrix was clearly most similar to the data in all tested cases. These results make the regional-continuity theory a less likely explanation for the observed cranial differences than the two single-origin theories. Of these, single southwest Asian origins seems the more likely interpretation of the data.     

Recapture Heterogeneity in Cliff Swallows: Increased Exposure to Mist Nets Leads to Net Avoidance

Ecologists often use mark-recapture to estimate demographic variables such as abundance, growth rate, or survival for samples of wild animal populations. A common assumption underlying mark-recapture is that all animals have an equal probability of detection, and failure to meet or correct for this assumption–as when certain members of the population are either easier or more difficult to capture than other animals–can lead to biased and inaccurate demographic estimates. We built within-year and among-years Cormack-Jolly-Seber recaptures-only models to identify causes of capture heterogeneity for a population of colonially nesting cliff swallows (Petrochelidon pyrrhonota) caught using mist-netting as a part of a 20-year mark-recapture study in southwestern Nebraska, U.S.A. Daily detection of cliff swallows caught in stationary mist nets at their colony sites declined as the birds got older and as the frequency of netting at a site within a season increased. Experienced birds’ avoidance of the net could be countered by sudden disturbances that startled them into a net, such as when we dropped a net over the side of a bridge or flushed nesting cliff swallows into a stationary net positioned at a colony entrance. Our results support the widely held, but seldom tested, belief that birds learn to avoid stationary mist nets over time, but also show that modifications of traditional field methods can reduce this source of recapture heterogeneity.     

Comparison of high-performance liquid chromatography with electrochemical detection and gas chromatography—mass fragmentography for the assay of salsolinol, dopamine and dopamine metabolites in food and beverage samples

High-performance liquid chromatography with electrochemical detection (HPLC—ED) and combined gas chromatograph—mass spectrometry in the single-ion monitoring mode (GC—MS-SIM) have been used for the determination of salsolinol, dopamine, 3,4-dihydroxyphenylacetic acid, 3,4-dihydorxyphenylethanol and norepinephrine in a selection of food and beverage samples. The unique specificity of the SIM mode allows a simple one-step extraction to be used even for complex sample matrices. We have been able to demonstrate the quantitative and qualitative advantages offered by GC—MS over HPLC—ED by direct comparison of the chromatographic data obtained. We demostrate that the specificity of SIM and the benefits offered by the incorporation of deuterated internal standards make GC—MS-SIM the method of choice for valid identification and precise quantitation of salsolinol, dopamine and dopamine metabolites in a complex sample matrix.     

Condition Number Estimation of Preconditioned Matrices

The present paper introduces a condition number estimation method for preconditioned matrices. The newly developed method provides reasonable results, while the conventional method which is based on the Lanczos connection gives meaningless results. The Lanczos connection based method provides the condition numbers of coefficient matrices of systems of linear equations with information obtained through the preconditioned conjugate gradient method. Estimating the condition number of preconditioned matrices is sometimes important when describing the effectiveness of new preconditionerers or selecting adequate preconditioners. Operating a preconditioner on a coefficient matrix is the simplest method of estimation. However, this is not possible for large-scale computing, especially if computation is performed on distributed memory parallel computers. This is because, the preconditioned matrices become dense, even if the original matrices are sparse. Although the Lanczos connection method can be used to calculate the condition number of preconditioned matrices, it is not considered to be applicable to large-scale problems because of its weakness with respect to numerical errors. Therefore, we have developed a robust and parallelizable method based on Hager’s method. The feasibility studies are curried out for the diagonal scaling preconditioner and the SSOR preconditioner with a diagonal matrix, a tri-daigonal matrix and Pei’s matrix. As a result, the Lanczos connection method contains around 10% error in the results even with a simple problem. On the other hand, the new method contains negligible errors. In addition, the newly developed method returns reasonable solutions when the Lanczos connection method fails with Pei’s matrix, and matrices generated with the finite element method.     

A Novel Tool to Mitigate By-Catch Mortality of Baltic Seals in Coastal Fyke Net Fishery

Developing methods to reduce the incidental catch of non-target species is important, as by-catch mortality poses threats especially to large aquatic predators. We examined the effectiveness of a novel device, a “seal sock”, in mitigating the by-catch mortality of seals in coastal fyke net fisheries in the Baltic Sea. The seal sock developed and tested in this study was a cylindrical net attached to the fyke net, allowing the seals access to the surface to breathe while trapped inside fishing gear. The number of dead and live seals caught in fyke nets without a seal sock (years 2008–2010) and with a sock (years 2011–2013) was recorded. The seals caught in fyke nets were mainly juveniles. Of ringed seals (Phoca hispida botnica) both sexes were equally represented, while of grey seals (Halichoerus grypus) the ratio was biased (71%) towards males. All the by-caught seals were dead in the fyke nets without a seal sock, whereas 70% of ringed seals and 11% of grey seals survived when the seal sock was used. The seal sock proved to be effective in reducing the by-catch mortality of ringed seals, but did not perform as well with grey seals.     

Estimating errors and confidence intervals for branch lengths in phylogenetic trees by a bootstrap approach

A method, based on the bootstrap procedure, is proposed for the estimation of branch-length errors and confidence intervals in a phylogenetic tree for which equal rates of substitution among lineages do not necessarily hold. The method can be used to test whether an estimated internodal distance is significantly greater than zero. In the application of the method, any estimator of genetic distances, as well as any tree reconstruction procedure (based on distance matrices), can be used. Also the method is not limited by the number of species involved in the phylogenetic tree. An example of the application of the method in the reconstruction of the phylogenetic tree for the four hominoid species—human, chimpanzee, gorilla, and orangutan—is shown. Correspondence to: J. Dopazo     

A Small-Scale Field Trial of Pyriproxyfen-Impregnated Bed Nets against Pyrethroid-Resistant Anopheles gambiae s.s. in Western Kenya

Pyrethroid resistance is becoming a major problem for vector control programs, because at present, there are few suitable chemical substitutes for pyrethroids, as when used on bed nets the insecticide must have low mammalian toxicity as well as high activity to mosquitoes. Pyriproxyfen (PPF) is one of the most active chemicals among the juvenile hormone mimic (JHM) group. Sterilizing mosquitoes by using PPF could be a potential control measure for pyrethroid-resistant malaria vectors. We investigated the sterilizing effects of two types of PPF-impregnated bed nets – a 1% PPF-impregnated net and a 1% PPF +2% permethrin-impregnated net (Olyset Duo) – to pyrethroid-resistant wild population of Anopheles gambiae s.s. in western Kenya. High mortality of blood-fed mosquitos was observed 3 days post-collection, in the houses where PPF-impregnated nets were used, indicating the effect of PPF on the longevity of mosquitos that came in contact with the net. Reduction in the number of ovipositing females, number of eggs, and number of progeny per female were also observed in the houses in which both Olyset Duo and PPF-impregnated nets were used. This is the first field study showing the high sterilizing efficacy of PPF against wild pyrethroid-resistant An. gambiae s.s. population. In addition, we recognized the necessity of combined use of permethrin with PPF, in order to reduce the risk of mosquito bites and provide a level of personal protection. Further studies on wild pyrethroid-resistant mosquito populations such as An. arabiensis and An. funestus s.s. would provide more information on the practical use of the PPF-impregnated bed nets.     

Do Size and Insecticide Treatment Matter? Evaluation of Different Nets against Phlebotomus argentipes,the Vector of Visceral Leishmaniasis in Nepal

In the Indian subcontinent, Leishmania donovani, the parasite causing visceral leishmaniasis (VL) is transmitted by the sand fly vector Phlebotomus argentipes. Long lasting insecticide treated nets (LN) have been postulated as alternative or complement to Indoor Residual Spraying but there are few field studies evaluating the entomological efficacy of different nets against this vector. We conducted two crossover trials in a VL endemic area in Nepal to compare the barrier effect of (1) LN with different mesh sizes (156 holes/inch² vs 625 holes/inch²) and (2) alpha-cypermethrin treated LN and untreated nets having the same mesh size (156 holes/inch²). Each crossover trial had two arms consisting of a sequence of two different nets for 8 nights. We used 10 cattle sheds per trial. A cow placed under the net was used as bait. CDC light traps placed inside the nets were used to evaluate the number of P. argentipes crossing the net barrier. Negative binomial generalized estimating equation (GEE) population-averaged models adjusted by night and sequence were used to estimate the barrier effect of the different nets. The crossover trials conducted in a rural village in Morang district (South-eastern Nepal) demonstrated that reducing the size of the holes in treated nets (625 holes/inch²) increased the barrier effect of LN by 77% (95% confidence interval (CI): 56%–88%) compared with treated nets with larger holes (156 holes/inch²). Treating nets with alpha-cypermethrin reduced the number of P. argentipes captured inside the nets by 77% (95% CI: 27%–93%) compared with untreated nets. The effectiveness and acceptability of finer mesh pyrethroid treated LN should be tested for VL prevention in a randomized controlled trial.     

Evaluation of Two Statistical Methods Provides Insights into the Complex Patterns of Alternative Polyadenylation Site Switching

Switching between different alternative polyadenylation (APA) sites plays an important role in the fine tuning of gene expression. New technologies for the execution of 3’-end enriched RNA-seq allow genome-wide detection of the genes that exhibit significant APA site switching between different samples. Here, we show that the independence test gives better results than the linear trend test in detecting APA site-switching events. Further examination suggests that the discrepancy between these two statistical methods arises from complex APA site-switching events that cannot be represented by a simple change of average 3’-UTR length. In theory, the linear trend test is only effective in detecting these simple changes. We classify the switching events into four switching patterns: two simple patterns (3’-UTR shortening and lengthening) and two complex patterns. By comparing the results of the two statistical methods, we show that complex patterns account for 1/4 of all observed switching events that happen between normal and cancerous human breast cell lines. Because simple and complex switching patterns may convey different biological meanings, they merit separate study. We therefore propose to combine both the independence test and the linear trend test in practice. First, the independence test should be used to detect APA site switching; second, the linear trend test should be invoked to identify simple switching events; and third, those complex switching events that pass independence testing but fail linear trend testing can be identified.     

Informational properties of neural nets performing algorithmic and logical tasks

It is argued that the genetic information necessary to encode an algorithmic neural processor tutoring an otherwise randomly connected biological neural net is represented by the entropy of the analogous minimal Turing machine. Such a near-minimal machine is constructed performing the whole range of bivalent propositional logic in variables. Neural nets computing the same task are presented; their informational entropy can be gauged with reference to the analogous Turing machine. It is also shown that nets with one hidden layer can be trained to perform algorithms solving propositional logic by error back-propagation. Received: 30 June 1995 / Accepted in revised form: 9 January 1996     

A simple vertical gill net system for variable current conditions

A new technique for suspending vertical gill nets is described. The system is designed to minimize the effects of currents, which otherwise could lift the lead line off the bottom or pull the floating line down.For more than two years, 30 m deep gill nets have been used in a Baltic Sea coastal region and have been found to work well in current, wave and wind conditions characteristic of this area. Setting a net takes about 5 minutes and retrieving it 10 minutes, even under relatively windy conditions (10–15 ms^–1). No special equipment is required to set or to haul a net, and we have used a variety of boats, from a small rowing boat to a 16 m long research vessel. The total material cost per net is about US $100.     

Use of Bed Nets and Factors That Influence Bed Net Use among Jinuo Ethnic Minority in Southern China

Background

Insecticide-treated nets (ITNs) are an integral part of vector control recommendations for malaria elimination in China. This study investigated the extent to which bed nets were used and which factors influence bed net use among Jinuo Ethnic Minority in China-Myanmar-Laos border areas.

Methods and Findings

This study combined a quantitative household questionnaire survey and qualitative semi-structured in-depth interviews (SDI). Questionnaires were administered to 352 heads of households. SDIs were given to 20 key informants. The bed net to person ratio was 1∶2.1 (i.e., nearly one net for every two people), however only 169 (48.0%) households owned at least one net and 623 (47.2%) residents slept under bed nets the prior night. The percentages of residents who regularly slept under nets (RSUN) and slept under nets the prior night (SUNPN) were similar (48.0% vs. 47.2%, P>0.05), however the percentage correct use of nets (CUN) was significantly lower (34.5%, P<0.0001). The annual cash income per person (ACIP) was an independent factor that influenced bed net use (P<0.0001), where families with an ACIP of {"type":"entrez-protein","attrs":{"text":"CNY10000","term_id":"895491745","term_text":"CNY10000"}}CNY10000 or more were much more likely to use nets. House type was strongly associated with bed net use (OR: 4.71, 95% CI: 2.81, 7.91; P<0.0001), where those with traditional wood walls and terracotta roofs were significantly more likely to use nets, and the head of household''s knowledge was an independent factor (OR: 5.04, 95% CI: 2.72, 9.35; P<0.0001), where those who knew bed nets prevent malaria were significantly more likely to use nets too.

Conclusions

High bed net availability does not necessarily mean higher coverage or bed net use. Household income, house type and knowledge of the ability of bed nets to prevent malaria are all independent factors that influence bed net use among Jinuo Ethnic Minority.     

On the character of and distance between states in a binary switching net

Behavioral examination of binary switching net models has typically concerned itself with an examination of their cyclical character. This article considers two less frequently discussed behavioral variables-the density of 1's in net states and the (Hamming) distance between net states. These variables are studied under fully random nets and under nets controlled at levels of internal homogeneity, forcibility or threshold. A collection of theoretical and simulated results is presented.     

Random field models for fitness landscapes

 In many cases fitness landscapes are obtained as particular instances of random fields by randomly assigning a large number of parameters. Models of this type are often characterized reasonably well by their covariance matrices. We characterize isotropic random fields on finite graphs in terms of their Fourier series expansions and investigate the relation between the covariance matrix of the random field model and the correlation structure of the individual landscapes constructed from this random field. Correlation measures are a good characteristic of “rugged landscapes” models as they are closely related to quantities like the number of local optima or the length of adaptive walks. Our formalism suggests to approximate landscape with known autocorrelation function by a random field model that has the same correlation structure. Received: 10 November 1995 / Revised version: 19 February 1996     

Bioanalysis of erythromycin 2'-ethylsuccinate in plasma using phase-system switching continuous-flow fast atom bombardment liquid chromatography—mass spectrometry

A specific method for the determination of erythromycin 2'-ethylsuccinate (EM-ES) in plasma is described. The method involves a liquid—liquid extraction procedure followed by the analysis of extracts using phase-system switching (PSS) continuous-flow fast atom bombardment (CF-FAB) liquid chromatography—mass spectrometry (LC—MS). In PSS EM-ES is enriched after analytical separation on a short trapping column, from which it is desorbed to the LC—MS interface. In this way, favourable mobile phases can be used for the LC separation and for the MS detection. Using the PSS approach a flow-rate reduction from 1.0 ml/min in the LC system to 15 μl/min going into the mass spectrometer was achieved without splitting. The determination limit for EM-ES was 0.1 μg/ml.     

On some distribution functions for non-linear switching elements with finite dead time

Summary The output distribution function of a non-linear switching element with a Poissonian sequence of impulses at the input is calculated by a straightforward method. The dead time of the element is taken into account. Some limiting properties of this distribution and its mean value are studied.The present research has been sponsored in part by the Air Force Avionics Lab., Research and Technology Div., Wright-Patterson Air Force, Air Force System Command — U.S.A.F. — Government of United States of America, Contract no. AF 33(615)-2786.