Chaos and Neural Network Learning. Some Observations

In this paper, we investigate the impact of chaos on the learning process of the XOR-boolean function by backpropagation neural networks. It has been shown previously that such networks exhibit chaotic behavior but it has never been studied whether chaos enhances or prohibits learning. We show that chaos (when learning the XOR-boolean function) does indeed allow learning but our findings do not indicate any positive role of chaos for learning. In particular, we found that the temperature parameter in the backpropagation algorithm causes the parameter regime, as represented by means of a bifurcation diagram, to shift to the right. We furthermore found that as less chaos appears during the learning process, the faster, on the average, a neural network learned the XOR-function.     

Modellversuche und numerische Simulationen mit der Diskrete‐Elemente‐Methode zum räumlichen passiven Erddruck

Model tests and numerical simulations with the Distinct‐Element‐Method to spatial passive earth pressure. The growing urban development in connection with a closer approach of excavations to existing buildings requires a more careful dimensioning of the excavation walls which specially refers to the movement of retaining walls. For more realistic investigations own model experiments were planned, made and analysed. New testing and measuring methods were developed which for the first time made it possible to record stress distributions along girders. The Distinct‐Element‐Method (DEM) was extended and improved to enable the numerical simulation of load tests. This method allows to describe the experiments with an adequate accuracy. Based on this knowledge, conditions not covered by experiments could be simulated by means of the DEM. Accordingly, the test matrix could be extended.     

The goal of forming accurate impressions during social interactions: Attenuating the impact of negative expectancies.

Investigated the idea that impression formation goals may regulate the impact that perceiver expectancies have on social interactions. In simulated interviews, interviewer Ss were given a negative expectancy about one applicant S and no expectancy about another. Half the interviewers were encouraged to form accurate impressions; the others were not. As predicted, no-goal interviewers exhibited a postinteraction impression bias against the negative-expectancy applicants, whereas the accuracy-goal interviewers did not. Moreover, the ability of the accuracy goal to reduce this bias was apparently mediated by more extensive and less biased interviewer information-gathering, which in turn elicited an improvement in negative-expectancy applicants' performances. These findings stress the theoretical and practical importance of considering the motivational context within which expectancy-tinged social interactions occur. (PsycINFO Database Record (c) 2010 APA, all rights reserved)     

On dimensionality, discriminant validity, and the role of psychometric analyses in personality theory and measurement: Reply to Kruglanski et al.'s (1997) defense of the Need for Closure Scale.

D. M. Webster and A. W. Kruglanski's (1994) Need for Closure Scale (NFCS) confounds multiple dimensions and is redundant with existing instruments (S. L. Neuberg, T. N. Judice, & S. G. West, 1997). A. W. Kruglanski and his colleagues (1997) dismissed these findings as "psychometric naysaying," although they presented no data that refute them. Moreover, Kruglanski et al (1997) suggested that researchers (a) be unconcerned with the NFCS's lack of discriminant validity and (b) use the scale as if it were unidimensional. These recommendations are problematic. Using the NFCS in this manner invites interpretational ambiguity and theoretical confusion. In contrast to the Kruglanski et al (1997) position, proper psychometric analyses play a critical role in theory testing and in the development of conceptually coherent measures of individual differences. (PsycINFO Database Record (c) 2010 APA, all rights reserved)     

Heterozygous peripheral myelin protein 22-deficient mice are affected by a progressive demyelinating tomaculous neuropathy

Hereditary neuropathy with liability to pressure palsy (HNPP) is associated with a heterozygous 1.5 megabase deletion on chromosome 17 that includes the peripheral myelin protein (PMP) gene PMP22. We show that heterozygous PMP22 knock-out mice, which carry only one functional pmp22 allele and thus genetically mimic HNPP closely, display similar morphological and electrophysiological features as observed in HNPP nerves. As reported previously, focal hypermyelinating structures called tomacula, the pathological hallmarks of HNPP, develop progressively in young PMP22(+/0) mice. By following the fate of tomacula during aging, we demonstrate now that these mutant animals are also interesting models for examining HNPP disease mechanisms. Subtle electrophysiological abnormalities are detected in PMP22(+/0) mice >1 year old, and a significant number of abnormally swollen and degenerating tomacula are present. Thinly myelinated axons and supernumerary Schwann cells forming onion bulbs as fingerprints of repeated cycles of demyelination and remyelination are also encountered frequently. Quantitative analyses using electron microscopy on cross sections and light microscopy on single teased nerve fibers suggest that tomacula are intrinsically unstable structures that are prone to degeneration; however, the severity of morphological and electrophysiological abnormalities in PMP22(+/0) mice is variable. These combined findings are reminiscent of the disease progression in HNPP and offer a possible explanation about why some HNPP patients develop a chronic motor and sensory neuropathy later in life that resembles demyelinating forms of Charcot-Marie-Tooth disease by both morphological and clinical criteria.     

Prophylaxis and therapy with factor VII concentrate (human) immuno, vapor heated in patients with congenital factor VII deficiency: a summary of case reports

Hereditary factor VII deficiency is a rare autosomal recessive condition, usually associated with normal or reduced levels of a functionally defective molecule. The available means of treating this condition in North America presents serious health risks to the patient. Transfusion with fresh frozen plasma carries a risk of volume overload and a significant risk for viral transmission. Sustained prothrombin complex therapy is associated with a high risk for thrombogenic complications. This communication describes the use of Factor VII Concentrate (Human) Immuno, Vapor Heated--an intermediate purity factor VII concentrate from Immuno A.G.--for the treatment of 13 patients with factor VII deficiency. Treatment regimens described include those for long-term prophylaxis (three children), acute hemorrhages (two children, one adult), peripartum prophylaxis (one patient), and surgical coverage (two children, four adults). Prophylaxis and therapy were successful in all cases, the medication was well-tolerated, and there were no complications. In the three cases of long-term prophylaxis in children, doses of 10-50 IU/kg were given one to three times a week; one patient has undergone long-term prophylaxis for approximately 8 years, one patient for 1 year, and one patient for 1 1/2 years. Three cases in which Factor VII Concentrate was principally used for treatment of acute episodes of bleeding are described. One infant received Factor VII Concentrate on about 50 occasions for treatment of mucosal bleeding; a correction to 40-100% resulted in cessation of bleeding within 15 min in all cases. For treatment of an episode of intracranial bleeding, an 8-year-old boy received a dose of 37 IU/kg Factor VII Concentrate every 6 hr for peak factor VII levels of approximately 100% and troughs as low as 4% over the 11-day treatment period. A 37-year-old adult male with intracranial bleeding received alternating doses of 16 IU/kg and 8 IU/kg every 6 hr for 10 days with peak factor VII levels in the upper thirties (%). The peak favor VII level during surgical coverage with Factor VII Concentrate (neurosurgery, open reduction of ankle bones, dental surgery, pituitary adenoma surgery, closed liver biopsy) was approximately 100% in all cases, with trough levels ranging from 8 to 65% over treatment periods of 24 hr to 16 days using treatment intervals of 6-12 hr.     

Different Emotional Reactions to Different Groups: A Sociofunctional Threat-Based Approach to "Prejudice".

The authors suggest that the traditional conception of prejudice--as a general attitude or evaluation--can problematically obscure the rich texturing of emotions that people feel toward different groups. Derived from a sociofunctional approach, the authors predicted that groups believed to pose qualitatively distinct threats to in-group resources or processes would evoke qualitatively distinct and functionally relevant emotional reactions. Participants' reactions to a range of social groups provided a data set unique in the scope of emotional reactions and threat beliefs explored. As predicted, different groups elicited different profiles of emotion and threat reactions, and this diversity was often masked by general measures of prejudice and threat. Moreover, threat and emotion profiles were associated with one another in the manner predicted: Specific classes of threat were linked to specific, functionally relevant emotions, and groups similar in the threat profiles they elicited were also similar in the emotion profiles they elicited. (PsycINFO Database Record (c) 2010 APA, all rights reserved)