Urban image classification with semisupervised multiscale cluster kernels

This paper presents a semisupervised support vector machine (SVM) that integrates the information of both labeled and unlabeled pixels efficiently. Method's performance is illustrated in the relevant problem of very high resolution image classification of urban areas. The SVM is trained with the linear combination of two kernels: a base kernel working only with labeled examples is deformed by a likelihood kernel encoding similarities between labeled and unlabeled examples. Results obtained on very high resolution (VHR) multispectral and hyperspectral images show the relevance of the method in the context of urban image classification. Also, its simplicity and the few parameters involved make the method versatile and workable by unexperienced users.

Structured output SVM for remote sensing image classification

In the recent years, kernel methods have revealed very powerful tools in many application domains in general and in remote sensing image classification in particular. The special characteristics of remote sensing images (high dimension, few labeled samples and different noise sources) are efficiently dealt with kernel machines. In this paper, we propose the use of structured output learning to improve remote sensing image classification based on kernels. Structured output learning is concerned with the design of machine learning algorithms that not only implement input-output mapping, but also take into account the relations between output labels, thus generalizing unstructured kernel methods. We analyze the framework and introduce it to the remote sensing community. Output similarity is here encoded into SVM classifiers by modifying the model loss function and the kernel function either independently or jointly. Experiments on a very high resolution (VHR) image classification problem shows promising results and opens a wide field of research with structured output kernel methods.

Evidence-based guidelines on the therapeutic use of repetitive transcranial magnetic stimulation (rTMS).

A group of European experts was commissioned to establish guidelines on the therapeutic use of repetitive transcranial magnetic stimulation (rTMS) from evidence published up until March 2014, regarding pain, movement disorders, stroke, amyotrophic lateral sclerosis, multiple sclerosis, epilepsy, consciousness disorders, tinnitus, depression, anxiety disorders, obsessive-compulsive disorder, schizophrenia, craving/addiction, and conversion. Despite unavoidable inhomogeneities, there is a sufficient body of evidence to accept with level A (definite efficacy) the analgesic effect of high-frequency (HF) rTMS of the primary motor cortex (M1) contralateral to the pain and the antidepressant effect of HF-rTMS of the left dorsolateral prefrontal cortex (DLPFC). A Level B recommendation (probable efficacy) is proposed for the antidepressant effect of low-frequency (LF) rTMS of the right DLPFC, HF-rTMS of the left DLPFC for the negative symptoms of schizophrenia, and LF-rTMS of contralesional M1 in chronic motor stroke. The effects of rTMS in a number of indications reach level C (possible efficacy), including LF-rTMS of the left temporoparietal cortex in tinnitus and auditory hallucinations. It remains to determine how to optimize rTMS protocols and techniques to give them relevance in routine clinical practice. In addition, professionals carrying out rTMS protocols should undergo rigorous training to ensure the quality of the technical realization, guarantee the proper care of patients, and maximize the chances of success. Under these conditions, the therapeutic use of rTMS should be able to develop in the coming years.

Being old-for-grade and drug use: not such a clear connection.

AIM: To assess whether repeating a grade was associated with drug use among adolescents after controlling for personal, family and school-related variables, and whether there were differences between students in mandatory and post-mandatory school. METHODS: Data were drawn from the Catalonia Adolescent Health Survey, a cross-sectional study of in-school adolescents aged 14-19 y. The index group included 366 subjects who were repeating a grade at the time the survey was carried out (old-for-grade, OFG). A control group matched by gender, school and being one grade ahead was randomly chosen among all the subjects who had never repeated a grade. All statistically significant variables in the bivariate analysis were included in a multivariate analysis. In a second step, all analyses were repeated for students in mandatory (14-16 y) and post-mandatory (17-19 y) school. RESULTS: After controlling for background variables, subjects in the index group were more likely to perceive that most of their peers were using synthetic drugs and to have ever used them, to have bad grades and a worse relationship with their teachers. OFG students in mandatory school were more likely to have divorced parents, bad grades and have ever used synthetic drugs, whereas they were less likely to be regular drinkers. OFG students in post-mandatory school were more likely to have below average grades, to be regular smokers and to perceive that most of their peers used synthetic drugs. CONCLUSIONS: When background variables are taken into consideration, the relationship between repeating a grade and drug use is not so clear. By increasing the familial and academic support of adolescents with academic underachievement, we could reduce their drug consumption.

Sex, drugs and chronic illness: health behaviours among chronically ill youth.

BACKGROUND: A growing body of literature indicates that adolescents with chronic conditions are as likely, or more likely, to take risky behaviours than their healthy peers. The objective of this research was to assess whether adolescents with chronic illness in Catalonia differ from their healthy peers in risk-taking behaviour. METHODS: Data were drawn from the Catalonia Adolescent Health database, a survey including a random school-based sample of 6952 young people, aged 14-19 years. The index group (IG) included 665 adolescents (450 females) reporting several chronic conditions. The comparison group (CG) comprised 6287 healthy adolescents (3306 females). Personal, family and school-related variables were analysed to ensure comparability between groups. Sexual behaviour, drug use (tobacco, alcohol, cannabis, cocaine and synthetic drugs) and perception of drug use among peers and in school were compared. Analysis was carried out separately by gender. chi-square, Fisher's and Student's tests were used to compare categorical and continuous variables. RESULTS: The prevalence of chronic conditions was 9.6%, with females showing a higher prevalence than males. The IG showed similar or higher rates of sexual intercourse and risky sexual behaviour. For most studied drugs, IG males reported slightly lower rates of use than CG males, while IG females showed higher rates for every drug studied. No differences were found in the perceptions of drug use among peers or in their school. CONCLUSIONS: Similar to previous research, chronically ill adolescents in our sample are as likely, or more likely, to take risky behaviours than their healthy counterparts and should receive the same anticipatory guidance.

Cluster kernels for semisupervised classification of VHR urban images

In this paper, we present and apply a semisupervised support vector machine based on cluster kernels for the problem of very high resolution image classification. In the proposed setting, a base kernel working with labeled samples only is deformed by a likelihood kernel encoding similarities between unlabeled examples. The resulting kernel is used to train a standard support vector machine (SVM) classifier. Experiments carried out on very high resolution (VHR) multispectral and hyperspectral images using very few labeled examples show the relevancy of the method in the context of urban image classification. Its simplicity and the small number of parameters involved make it versatile and workable by unexperimented users.