Evaluating retraining rules for semi-supervised learning in neural network based cursive word recognition

Training a system to recognize handwritten words is a task that requires a large amount of data with their correct transcription. However, the creation of such a training set, including the generation of the ground truth, is tedious and costly. One way of reducing the high cost of labeled training data acquisition is to exploit unlabeled data, which can be gathered easily. Making use of both labeled and unlabeled data is known as semi-supervised learning. One of the most general versions of semi-supervised learning is self-training, where a recognizer iteratively retrains itself on its own output on new, unlabeled data. In this paper we propose to apply semi-supervised learning, and in particular self-training, to the problem of cursive, handwritten word recognition. The special focus of the paper is on retraining rules that define what data are actually being used in the retraining phase. In a series of experiments it is shown that the performance of a neural network based recognizer can be significantly improved through the use of unlabeled data and self-training if appropriate retraining rules are applied.

A triplet spike-timing–dependent plasticity model generalizes the Bienenstock–Cooper–Munro rule to higher-order spatiotemporal correlations

Synaptic strength depresses for low and potentiates for high activation of the postsynaptic neuron. This feature is a key property of the Bienenstock–Cooper–Munro (BCM) synaptic learning rule, which has been shown to maximize the selectivity of the postsynaptic neuron, and thereby offers a possible explanation for experience-dependent cortical plasticity such as orientation selectivity. However, the BCM framework is rate-based and a significant amount of recent work has shown that synaptic plasticity also depends on the precise timing of presynaptic and postsynaptic spikes. Here we consider a triplet model of spike-timing–dependent plasticity (STDP) that depends on the interactions of three precisely timed spikes. Triplet STDP has been shown to describe plasticity experiments that the classical STDP rule, based on pairs of spikes, has failed to capture. In the case of rate-based patterns, we show a tight correspondence between the triplet STDP rule and the BCM rule. We analytically demonstrate the selectivity property of the triplet STDP rule for orthogonal inputs and perform numerical simulations for nonorthogonal inputs. Moreover, in contrast to BCM, we show that triplet STDP can also induce selectivity for input patterns consisting of higher-order spatiotemporal correlations, which exist in natural stimuli and have been measured in the brain. We show that this sensitivity to higher-order correlations can be used to develop direction and speed selectivity.

Improving Pediatric Basic Life Support Performance Through Blended Learning With Web-Based Virtual Patients: Randomized Controlled Trial

BACKGROUND E-learning and blended learning approaches gain more and more popularity in emergency medicine curricula. So far, little data is available on the impact of such approaches on procedural learning and skill acquisition and their comparison with traditional approaches. OBJECTIVE This study investigated the impact of a blended learning approach, including Web-based virtual patients (VPs) and standard pediatric basic life support (PBLS) training, on procedural knowledge, objective performance, and self-assessment. METHODS A total of 57 medical students were randomly assigned to an intervention group (n=30) and a control group (n=27). Both groups received paper handouts in preparation of simulation-based PBLS training. The intervention group additionally completed two Web-based VPs with embedded video clips. Measurements were taken at randomization (t0), after the preparation period (t1), and after hands-on training (t2). Clinical decision-making skills and procedural knowledge were assessed at t0 and t1. PBLS performance was scored regarding adherence to the correct algorithm, conformance to temporal demands, and the quality of procedural steps at t1 and t2. Participants' self-assessments were recorded in all three measurements. RESULTS Procedural knowledge of the intervention group was significantly superior to that of the control group at t1. At t2, the intervention group showed significantly better adherence to the algorithm and temporal demands, and better procedural quality of PBLS in objective measures than did the control group. These aspects differed between the groups even at t1 (after VPs, prior to practical training). Self-assessments differed significantly only at t1 in favor of the intervention group. CONCLUSIONS Training with VPs combined with hands-on training improves PBLS performance as judged by objective measures.

Shallow Dialogue Processing Using Machine Learning Algorithms (or not)

This paper presents a shallow dialogue analysis model, aimed at human-human dialogues in the context of staff or business meetings. Four components of the model are defined, and several machine learning techniques are used to extract features from dialogue transcripts: maximum entropy classifiers for dialogue acts, latent semantic analysis for topic segmentation, or decision tree classifiers for discourse markers. A rule-based approach is proposed for solving cross-modal references to meeting documents. The methods are trained and evaluated thanks to a common data set and annotation format. The integration of the components into an automated shallow dialogue parser opens the way to multimodal meeting processing and retrieval applications.

Imagery of a moving object: the role of occipital cortex and human MT/V5+

Visual imagery – similar to visual perception – activates feature-specific and category-specific visual areas. This is frequently observed in experiments where the instruction is to imagine stimuli that have been shown immediately before the imagery task. Hence, feature-specific activation could be related to the short-term memory retrieval of previously presented sensory information. Here, we investigated mental imagery of stimuli that subjects had not seen before, eliminating the effects of short-term memory. We recorded brain activation using fMRI while subjects performed a behaviourally controlled guided imagery task in predefined retinotopic coordinates to optimize sensitivity in early visual areas. Whole brain analyses revealed activation in a parieto-frontal network and lateral–occipital cortex. Region of interest (ROI) based analyses showed activation in left hMT/V5+. Granger causality mapping taking left hMT/V5+ as source revealed an imagery-specific directed influence from the left inferior parietal lobule (IPL). Interestingly, we observed a negative BOLD response in V1–3 during imagery, modulated by the retinotopic location of the imagined motion trace. Our results indicate that rule-based motion imagery can activate higher-order visual areas involved in motion perception, with a role for top-down directed influences originating in IPL. Lower-order visual areas (V1, V2 and V3) were down-regulated during this type of imagery, possibly reflecting inhibition to avoid visual input from interfering with the imagery construction. This suggests that the activation in early visual areas observed in previous studies might be related to short- or long-term memory retrieval of specific sensory experiences.

Coupling biomechanics to a cellular level model: An approach to patient-specific image driven multi-scale and multi-physics tumor simulation

Modeling of tumor growth has been performed according to various approaches addressing different biocomplexity levels and spatiotemporal scales. Mathematical treatments range from partial differential equation based diffusion models to rule-based cellular level simulators, aiming at both improving our quantitative understanding of the underlying biological processes and, in the mid- and long term, constructing reliable multi-scale predictive platforms to support patient-individualized treatment planning and optimization. The aim of this paper is to establish a multi-scale and multi-physics approach to tumor modeling taking into account both the cellular and the macroscopic mechanical level. Therefore, an already developed biomodel of clinical tumor growth and response to treatment is self-consistently coupled with a biomechanical model. Results are presented for the free growth case of the imageable component of an initially point-like glioblastoma multiforme tumor. The composite model leads to significant tumor shape corrections that are achieved through the utilization of environmental pressure information and the application of biomechanical principles. Using the ratio of smallest to largest moment of inertia of the tumor material to quantify the effect of our coupled approach, we have found a tumor shape correction of 20\% by coupling biomechanics to the cellular simulator as compared to a cellular simulation without preferred growth directions. We conclude that the integration of the two models provides additional morphological insight into realistic tumor growth behavior. Therefore, it might be used for the development of an advanced oncosimulator focusing on tumor types for which morphology plays an important role in surgical and/or radio-therapeutic treatment planning.

Natural Language Checking with Program Checking Tools

Written text is an important component in the process of knowledge acquisition and communication. Poorly written text fails to deliver clear ideas to the reader no matter how revolutionary and ground-breaking these ideas are. Providing text with good writing style is essential to transfer ideas smoothly. While we have sophisticated tools to check for stylistic problems in program code, we do not apply the same techniques for written text. In this paper we present TextLint, a rule-based tool to check for common style errors in natural language. TextLint provides a structural model of written text and an extensible rule-based checking mechanism.

Potential of information technology in dental education

The use of information technology (IT) in dentistry is far ranging. In order to produce a working document for the dental educator, this paper focuses on those methods where IT can assist in the education and competence development of dental students and dentists (e.g. e-learning, distance learning, simulations and computer-based assessment). Web pages and other information-gathering devices have become an essential part of our daily life, as they provide extensive information on all aspects of our society. This is mirrored in dental education where there are many different tools available, as listed in this report. IT offers added value to traditional teaching methods and examples are provided. In spite of the continuing debate on the learning effectiveness of e-learning applications, students request such approaches as an adjunct to the traditional delivery of learning materials. Faculty require support to enable them to effectively use the technology to the benefit of their students. This support should be provided by the institution and it is suggested that, where possible, institutions should appoint an e-learning champion with good interpersonal skills to support and encourage faculty change. From a global prospective, all students and faculty should have access to e-learning tools. This report encourages open access to e-learning material, platforms and programs. The quality of such learning materials must have well defined learning objectives and involve peer review to ensure content validity, accuracy, currency, the use of evidence-based data and the use of best practices. To ensure that the developers' intellectual rights are protected, the original content needs to be secure from unauthorized changes. Strategies and recommendations on how to improve the quality of e-learning are outlined. In the area of assessment, traditional examination schemes can be enriched by IT, whilst the Internet can provide many innovative approaches. Future trends in IT will evolve around improved uptake and access facilitated by the technology (hardware and software). The use of Web 2.0 shows considerable promise and this may have implications on a global level. For example, the one-laptop-per-child project is the best example of what Web 2.0 can do: minimal use of hardware to maximize use of the Internet structure. In essence, simple technology can overcome many of the barriers to learning. IT will always remain exciting, as it is always changing and the users, whether dental students, educators or patients are like chameleons adapting to the ever-changing landscape.

Jazz drummers recruit language-specific areas for the processing of rhythmic structure

Rhythm is a central characteristic of music and speech, the most important domains of human communication using acoustic signals. Here, we investigated how rhythmical patterns in music are processed in the human brain, and, in addition, evaluated the impact of musical training on rhythm processing. Using fMRI, we found that deviations from a rule-based regular rhythmic structure activated the left planum temporale together with Broca's area and its right-hemispheric homolog across subjects, that is, a network also crucially involved in the processing of harmonic structure in music and the syntactic analysis of language. Comparing the BOLD responses to rhythmic variations between professional jazz drummers and musical laypersons, we found that only highly trained rhythmic experts show additional activity in left-hemispheric supramarginal gyrus, a higher-order region involved in processing of linguistic syntax. This suggests an additional functional recruitment of brain areas usually dedicated to complex linguistic syntax processing for the analysis of rhythmical patterns only in professional jazz drummers, who are especially trained to use rhythmical cues for communication.

A serial scheme for minimizing the duration of resource-constrained projects within Microsoft Project

Microsoft Project is one of the most-widely used software packages for project management. For the scheduling of resource-constrained projects, the package applies a priority-based procedure using a specific schedule-generation scheme. This procedure performs relatively poorly when compared against other software packages or state-of-the-art methods for resource-constrained project scheduling. In Microsoft Project 2010, it is possible to work with schedules that are infeasible with respect to the precedence or the resource constraints. We propose a novel schedule-generation scheme that makes use of this possibility. Under this scheme, the project tasks are scheduled sequentially while taking into account all temporal and resource constraints that a user can define within Microsoft Project. The scheme can be implemented as a priority-rule based heuristic procedure. Our computational results for two real-world construction projects indicate that this procedure outperforms the built-in procedure of Microsoft Project