On the optimal usage of labelled examples in semi-supervised multi-class classification problems

In recent years, the performance of semi-supervised learning has been theoretically investigated. However, most of this theoretical development has focussed on binary classification problems. In this paper, we take it a step further by extending the work of Castelli and Cover [1] [2] to the multi-class paradigm. Particularly, we consider the key problem in semi-supervised learning of classifying an unseen instance x into one of K different classes, using a training dataset sampled from a mixture density distribution and composed of l labelled records and u unlabelled examples. Even under the assumption of identifiability of the mixture and having infinite unlabelled examples, labelled records are needed to determine the K decision regions. Therefore, in this paper, we first investigate the minimum number of labelled examples needed to accomplish that task. Then, we propose an optimal multi-class learning algorithm which is a generalisation of the optimal procedure proposed in the literature for binary problems. Finally, we make use of this generalisation to study the probability of error when the binary class constraint is relaxed.

Investigations on phytoplankton with special reference to water usage

Experiments and observations on the phytoplankton of certain lakes in the English Lake District were made from early 1973 to the end of March, 1974. They included laboratory and lake bioassays and observations on the quantity and quality of the phytoplankton in six lakes. The introductory sections of the report are about algae, the ecology of phytoplankton and the scope of the contracted work. Laboratory bioassays on water from one lake, Blelham Tarn, showed that phosphorus, silicon (for diatoms) and organic substances forming complexes with iron were the major substances limiting the growth of the algae tested. The growth of the test algae was limited to different degrees by those substances and, to some extent, to a greater or lesser degree at different times of year. It is suggested that a relatively simple form of bioassay could give valuable information to water undertakings. Lake bioassays and other experiments were carried out by using large in situ tubular plastic enclosures. Two such investigations are described. The effects of a change in sewerage in two drainage basins on the phytoplankton of three lakes is described and some data given about changes since 1945 in three other lakes in the same overall drainage basin. These latter lakes have been affected too by changes in sewerage and by increasing inputs of domestic and agricultural wastes. Throughout, the relevance of the work done to practical problems of water usage is kept in mind and discussed. In the last section special reference is made to the largely unpredictable results of water transfers. The report ends with a note on river phytoplankton.

Combination of web usage, content and structure information for diverse web mining applications in the tourism context and the context of users with disabilities

188 p.

Usage of substances and their dosage regimes in pharmacological animal models of schizophrenia

Background: Contrary to what is generally thought schizophrenia is a very common mental health issue. For this, several animal models are used to assess the illness in order to develop a definitive. The most widely spread paradigm is the use of pharmacological models. Aim: The aim of this review is to display which are the most used insults for the assessment of social behaviour related negative symptoms in animal models as well as to ascertain which is the most adequate regime. Design: Literature review. Methods: PubMed database was used for this article by the search of the indexed “schizophrenia”, “animal models”, “social behaviour” and “negative symptoms” descriptors. With the exception of a single article due to it value this review is based on articles from 10 years onwards. Besides, only clinical trials and reviews written in English or Spanish and that had laboratory rodents as target population were accepted. Results: The studies assessed agree that pharmacological models (specially those regarding the NMDA receptor antagonists) are a valuable means for the experimental investigation of negative symptoms in schizophrenia with the necessity to emphasise that only some negative symptoms (anhedonia and social interaction, mainly) can be experimentally assessed. Conclusions: There is not enough evidence regarding the fours aspects of this review. PCP, Ketamine or MK-801 in sub-acute dosage regimes are currently the most indicated insults to mimic schizophrenic symptoms in rodents, although further research in needed, albeit other substances are valuable as well. (In English language exclusively)

Gender effect on the right-left discrimination task in a sample of heroin-dependent patients

Rationale: Discriminating right from left is an everyday cognitive ability. Repeated exposure to certain drugs, such as heroin, can produce poor performance on many cognitive tasks. However, it is yet unclear whether drug abuse impairs the ability of right-left discrimination. Objectives: The aim of the present study is to examine whether the spatial ability measured by the right-left discrimination task can be affected by heroin abuse and whether such drug effect, if it exists, is gender related. Methods: A paper-and-pen test was used. The test consists of line drawings of a person with no arm, one arm, or both arms crossing the vertical body axis of the figure. The line drawings are viewed from the back, from the front, or randomly alternating between the back and front drawings. The subjects task is to mark which is the right or left hand in the figure as fast as possible. Results: A main finding in this study was that the ability to discriminate between left and right in visual space was impaired in heroin-dependent patients. Especially, heroin-dependent females performed poorer than control females in all conditions but heroin-dependent males only performed poorly in part of conditions. Conclusions: Recent heroin abuse impairs the ability of right-left discrimination and such impairment is gender related: heroin-dependent females demonstrated greater performance deficits than males.

Heroin impairs map-picture-following and memory tasks dependent on gender and orientation of the tasks

Male and female heroin-dependent patients (HDPs) matched with "normal" people were tested on 4 topographical orientation tasks: schematic map-following, map-memory, schematic picture-following, and picture-memory tasks. The results showed that, in general