University of Cagliari

About: University of Cagliari is a education organization based out in Cagliari, Italy. It is known for research contribution in the topics: Population & Dopamine. The organization has 11029 authors who have published 29046 publications receiving 771023 citations. The organization is also known as: Università degli Studi di Cagliari & Universita degli Studi di Cagliari.

Evolution of public attitudes about mental illness: a systematic review and meta-analysis

Abstract: Schomerus G, Schwahn C, Holzinger A, Corrigan PW, Grabe HJ, Carta MG, Angermeyer MC. Evolution of public attitudes about mental illness: a systematic review and meta-analysis. Objective: To explore whether the increase in knowledge about the biological correlates of mental disorders over the last decades has translated into improved public understanding of mental illness, increased readiness to seek mental health care and more tolerant attitudes towards mentally ill persons. Method: A systematic review of all studies on mental illness-related beliefs and attitudes in the general population published before 31 March 2011, examining the time trends of attitudes with a follow-up interval of at least 2 years and using national representative population samples. A subsample of methodologically homogeneous studies was further included in a meta-regression analysis of time trends. Results: Thirty-three reports on 16 studies on national time trends met our inclusion criteria, six of which were eligible for a meta-regression analysis. Two major trends emerged: there was a coherent trend to greater mental health literacy, in particular towards a biological model of mental illness, and greater acceptance of professional help for mental health problems. In contrast, however, no changes or even changes to the worse were observed regarding the attitudes towards people with mental illness. Conclusion: Increasing public understanding of the biological correlates of mental illness seems not to result in better social acceptance of persons with mental illness.

Mutations in GPC3, a glypican gene, cause the Simpson-Golabi-Behmel overgrowth syndrome

Abstract: Simpson-Golabi-Behmel syndrome (SGBS) is an X-linked condition characterized by pre- and postnatal overgrowth with visceral and skeletal anomalies. To identify the causative gene, breakpoints in two female patients with X;autosome translocations were identified. The breakpoints occur near the 5' and 3' ends of a gene, GPC3, that spans more than 500 kilobases in Xq26; in three families, different microdeletions encompassing exons cosegregate with SGBS. GPC3 encodes a putative extracellular proteoglycan, glypican 3, that is inferred to play an important role in growth control in embryonic mesodermal tissues in which it is selectively expressed. Initial western- and ligand-blotting experiments suggest that glypican 3 forms a complex with insulin-like growth factor 2 (IGF2), and might thereby modulate IGF2 action.

Poisoning Attacks against Support Vector Machines

Abstract: We investigate a family of poisoning attacks against Support Vector Machines (SVM). Such attacks inject specially crafted training data that increases the SVM's test error. Central to the motivation for these attacks is the fact that most learning algorithms assume that their training data comes from a natural or well-behaved distribution. However, this assumption does not generally hold in security-sensitive settings. As we demonstrate, an intelligent adversary can, to some extent, predict the change of the SVM's decision function due to malicious input and use this ability to construct malicious data. The proposed attack uses a gradient ascent strategy in which the gradient is computed based on properties of the SVM's optimal solution. This method can be kernelized and enables the attack to be constructed in the input space even for non-linear kernels. We experimentally demonstrate that our gradient ascent procedure reliably identifies good local maxima of the non-convex validation error surface, which significantly increases the classifier's test error.

Volatility clustering in financial markets: a microsimulation of interacting agents

Abstract: The finding of clustered volatility and ARCH effects is ubiquitous in financial data. This paper presents a possible explanation for this phenomenon within a multi-agent framework of speculative activity. In the model, both chartist and fundamentalist strategies are considered with agents switching between both behavioural variants according to observed differences in pay-offs. Price changes are brought about by a market maker reacting to imbalances between demand and supply. Most of the time, a stable and efficient market results. However, its usual tranquil performance is interspersed by sudden transient phases of destabilisation. An outbreak of volatility occurs if the fraction of agents using chartist techniques surpasses a certain threshold value, but such phases are quickly brought to an end by stabilising tendencies. Formally, this pattern can be understood as an example of a new type of dynamic behaviour known as "on-off intermittency" in physics literature. Statistical analysis of simulated time ...

Poisoning Attacks against Support Vector Machines

Abstract: We investigate a family of poisoning attacks against Support Vector Machines (SVM). Such attacks inject specially crafted training data that increases the SVM's test error. Central to the motivation for these attacks is the fact that most learning algorithms assume that their training data comes from a natural or well-behaved distribution. However, this assumption does not generally hold in security-sensitive settings. As we demonstrate, an intelligent adversary can, to some extent, predict the change of the SVM's decision function due to malicious input and use this ability to construct malicious data. The proposed attack uses a gradient ascent strategy in which the gradient is computed based on properties of the SVM's optimal solution. This method can be kernelized and enables the attack to be constructed in the input space even for non-linear kernels. We experimentally demonstrate that our gradient ascent procedure reliably identifies good local maxima of the non-convex validation error surface, which significantly increases the classifier's test error.