Mind change complexity of learning logic programs

The present paper motivates the study of mind change complexity for learning minimal models of length-bounded logic programs. It establishes ordinal mind change complexity bounds for learnability of these classes both from positive facts and from positive and negative facts. Building on Angluin’s notion of finite thickness and Wright’s work on finite elasticity, Shinohara defined the property of bounded finite thickness to give a sufficient condition for learnability of indexed families of computable languages from positive data. This paper shows that an effective version of Shinohara’s notion of bounded finite thickness gives sufficient conditions for learnability with ordinal mind change bound, both in the context of learnability from positive data and for learnability from complete (both positive and negative) data. Let Omega be a notation for the first limit ordinal. Then, it is shown that if a language defining framework yields a uniformly decidable family of languages and has effective bounded finite thickness, then for each natural number m >0, the class of languages defined by formal systems of length <= m: • is identifiable in the limit from positive data with a mind change bound of Omega (power)m; • is identifiable in the limit from both positive and negative data with an ordinal mind change bound of Omega × m. The above sufficient conditions are employed to give an ordinal mind change bound for learnability of minimal models of various classes of length-bounded Prolog programs, including Shapiro’s linear programs, Arimura and Shinohara’s depth-bounded linearly covering programs, and Krishna Rao’s depth-bounded linearly moded programs. It is also noted that the bound for learning from positive data is tight for the example classes considered.

Multi-linear strategies for (re)presenting the complexity of young people in research

Within the current climate of unpredictability and constant change, young people at school are faced with a multitude of choices and contradictory influences. In this article, I argue that (re)presentations of young people in youth research need to reflect the complexity and multiplicity of their lives and changing priorities, and I attempt to (re)present a small group of young people in this particular milieu. I illustrate some of the competing influences in their lives, and I outline some specific strategies that are useful for (re)presenting these contextual worlds. The strategies I advocate disrupt the homogenous representations of ‘youth’ as a developmental phase and instead reflect the diverse spheres of influence which shape their subjectivities and practices.

Unraveling the complexity of driving while intoxicated: a study into the prevalence of psychiatric and substance abuse comorbidity

Objective Research is beginning to provide an indication of the co-occurring substance abuse and mental health needs for the driving under the influence (DUI) population. This study aimed to examine the extent of such psychiatric problems among a large sample size of DUI offenders entering treatment in Texas. Methods This is a study of 36,373 past year DUI clients and 308,714 non-past year DUI clients admitted to Texas treatment programs between 2005 and 2008. Data were obtained from the State's administrative dataset. Results Analysis indicated that non-past year DUI clients were more likely to present with more severe illicit substance use problems, while past year DUI clients were more likely to have a primary problem with alcohol. Nevertheless, a cannabis use problem was also found to be significantly associated with DUI recidivism in the last year. In regards to mental health status, a major finding was that depression was the most common psychiatric condition reported by DUI clients, including those with more than one DUI offence in the past year. This cohort also reported elevated levels of Bipolar Disorder compared to the general population, and such a diagnosis was also associated with an increased likelihood of not completing treatment. Additionally, female clients were more likely to be diagnosed with mental health problems than males, as well as more likely to be placed on medications at admission and more likely to have problems with methamphetamine, cocaine, and opiates. Conclusions DUI offenders are at an increased risk of experiencing comorbid psychiatric disorders, and thus, corresponding treatment programs need to cater for a range of mental health concerns that are likely to affect recidivism rates.

Genetic Composition and Complexity of Virus Populations at Tungro-Endemic and Outbreak Rice Sites

We have recently demonstrated the geographic isolation of rice tungro bacilliform virus (RTBV) populations in the tungro-endemic provinces of Isabela and North Cotabato, Philippines. In this study, we examined the genetic structure of the virus populations at the tungro-outbreak sites of Lanao del Norte, a province adjacent to North Cotabato. We also analyzed the virus populations at the tungro-endemic sites of Subang, Indonesia, and Dien Khanh, Vietnam. Total DNA extracts from 274 isolates were digested with EcoRV restriction enzyme and hybridized with a full-length probe of RTBV. In the total population, 22 EcoRV-restricted genome profiles (genotypes) were identified. Although overlapping genotypes could be observed, the outbreak sites of Lanao del Norte had a genotype combination distinct from that of Subang or Dien Khanh but a genotype combination similar to that identified earlier from North Cotabato, the adjacent endemic province. Sequence analysis of the intergenic region and part of the ORF1 RTBV genome from randomly selected genotypes confirms the geographic clustering of RTBV genotypes and, combined with restriction analysis, the results suggest a fragmented spatial distribution of RTBV local populations in the three countries. Because RTBV depends on rice tungro spherical virus (RTSV) for transmission, the population dynamics of both tungro viruses were then examined at the endemic and outbreak sites within the Philippines. The RTBV genotypes and the coat protein RTSV genotypes were used as indicators for virus diversity. A shift in population structure of both viruses was observed at the outbreak sites with a reduced RTBV but increased RTSV gene diversity

The sample complexity of pattern classification with neural networks: the size of the weights is more important than the size of the network

Sample complexity results from computational learning theory, when applied to neural network learning for pattern classification problems, suggest that for good generalization performance the number of training examples should grow at least linearly with the number of adjustable parameters in the network. Results in this paper show that if a large neural network is used for a pattern classification problem and the learning algorithm finds a network with small weights that has small squared error on the training patterns, then the generalization performance depends on the size of the weights rather than the number of weights. For example, consider a two-layer feedforward network of sigmoid units, in which the sum of the magnitudes of the weights associated with each unit is bounded by A and the input dimension is n. We show that the misclassification probability is no more than a certain error estimate (that is related to squared error on the training set) plus A3 √((log n)/m) (ignoring log A and log m factors), where m is the number of training patterns. This may explain the generalization performance of neural networks, particularly when the number of training examples is considerably smaller than the number of weights. It also supports heuristics (such as weight decay and early stopping) that attempt to keep the weights small during training. The proof techniques appear to be useful for the analysis of other pattern classifiers: when the input domain is a totally bounded metric space, we use the same approach to give upper bounds on misclassification probability for classifiers with decision boundaries that are far from the training examples.

Overcoming the complexity of diagnostic problems due to sensor network architecture

In fault detection and diagnostics, limitations coming from the sensor network architecture are one of the main challenges in evaluating a system’s health status. Usually the design of the sensor network architecture is not solely based on diagnostic purposes, other factors like controls, financial constraints, and practical limitations are also involved. As a result, it quite common to have one sensor (or one set of sensors) monitoring the behaviour of two or more components. This can significantly extend the complexity of diagnostic problems. In this paper a systematic approach is presented to deal with such complexities. It is shown how the problem can be formulated as a Bayesian network based diagnostic mechanism with latent variables. The developed approach is also applied to the problem of fault diagnosis in HVAC systems, an application area with considerable modeling and measurement constraints.

The rademacher complexity of coregularized kernel classes

In the multi-view approach to semisupervised learning, we choose one predictor from each of multiple hypothesis classes, and we co-regularize our choices by penalizing disagreement among the predictors on the unlabeled data. We examine the co-regularization method used in the co-regularized least squares (CoRLS) algorithm, in which the views are reproducing kernel Hilbert spaces (RKHS's), and the disagreement penalty is the average squared difference in predictions. The final predictor is the pointwise average of the predictors from each view. We call the set of predictors that can result from this procedure the co-regularized hypothesis class. Our main result is a tight bound on the Rademacher complexity of the co-regularized hypothesis class in terms of the kernel matrices of each RKHS. We find that the co-regularization reduces the Rademacher complexity by an amount that depends on the distance between the two views, as measured by a data dependent metric. We then use standard techniques to bound the gap between training error and test error for the CoRLS algorithm. Experimentally, we find that the amount of reduction in complexity introduced by co regularization correlates with the amount of improvement that co-regularization gives in the CoRLS algorithm.

Investigating the complexity of judgement practice

Underlying all assessments are human judgements regarding the quality of students’ understandings. Despite their ubiquity, those judgements are conceptually elusive. The articles selected for inclusion in this issue explore the complexity of judgement practice raising critical questions that challenge existing views and accepted policy and practice.

How difficult are exams? A framework for assessing the complexity of introductory programming exams

Student performance on examinations is influenced by the level of difficulty of the questions. It seems reasonable to propose therefore that assessment of the difficulty of exam questions could be used to gauge the level of skills and knowledge expected at the end of a course. This paper reports the results of a study investigating the difficulty of exam questions using a subjective assessment of difficulty and a purpose-built exam question complexity classification scheme. The scheme, devised for exams in introductory programming courses, assesses the complexity of each question using six measures: external domain references, explicitness, linguistic complexity, conceptual complexity, length of code involved in the question and/or answer, and intellectual complexity (Bloom level). We apply the scheme to 20 introductory programming exam papers from five countries, and find substantial variation across the exams for all measures. Most exams include a mix of questions of low, medium, and high difficulty, although seven of the 20 have no questions of high difficulty. All of the complexity measures correlate with assessment of difficulty, indicating that the difficulty of an exam question relates to each of these more specific measures. We discuss the implications of these findings for the development of measures to assess learning standards in programming courses.