Search PubMed⌕ Search

SEARCH · Search PubMed

Results for “representation learning”

Search indexed PubMed citations on genomics, clinical trials, systematic reviews and public health. Explore titles, authors and supplied subject terms, then open the PubMed record.

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 1,153 records · Page 64Linked to original sources
PubMed2004

An expressive three-mode principal components model for gender recognition.

We present a three-mode expressive-feature model for recognizing gender (female, male) from point-light displays of walking people. Prototype female and male walkers are initially decomposed into a subspace of their three-mode components (posture, time, and gender). We then apply a weight factor to each point-light trajectory in the basis representation to enable adaptive, context-based gender estimations. The weight values are automatically learned from labeled training data. We present experiments using physical (actual) and perceived (from perceptual experiments) gender labels to train and test the system. Results with 40 walkers demonstrate greater than 90% recognition for both physically and perceptually labeled training examples. The approach has a greater flexibility over standard squared-error gender estimation to successfully adapt to different matching contexts.

Adult↗
PubMed1999

Interactive educational diabetes simulators: future possibilities.

In 1996 an interactive educational diabetes simulator called AIDA was released without charge on the Internet as a non-commercial contribution to continuing diabetes education. Over the past 3 years over 30,000 people have visited the AIDA Web site-- http://www.diabetic.org.uk/aida.htm--and over 10,000 copies of the program have been downloaded from there free-of-charge. This review builds on the experience gained from the AIDA development and the World Wide Web distribution of the software, and looks to the future, highlighting features which users might expect to see in future generations of such interactive educational diabetes programs. Novel functions already described in the literature are overviewed, and possible applications using personal computers and the Internet are discussed. The importance of the user interface is stressed. The concept of a "virtual diabetic patient" that provides an electronic representation of a patient with diabetes--and which can be used for self-learning/teaching/demonstration purposes--is highlighted.

Blood Glucose↗
PubMed2000

Spatial-temporal distribution of whisker-evoked activity in rat somatosensory cortex and the coding of stimulus location.

Rats use their facial vibrissae ("whiskers") to locate and identify objects. To learn about the neural coding of contact between whiskers and objects, we investigated the representation of single-vibrissa deflection by populations of cortical neurons. Microelectrode arrays, arranged in a geometric 10 x 10 grid, were inserted into the thalamo-recipient layers of "barrel cortex" (the vibrissal region of somatosensory cortex) in urethane-anesthetized rats, and neuronal activity across large sets of barrel-columns was measured. Typically, 5 msec after deflection of a whisker a 0.2 mm(2) focus of activity emerged. It rapidly expanded, doubling in size by 7 msec, before retracting and disappearing 28-59 msec after stimulus onset. The total territory engaged by the stimulus ranged from 0.5 to 2.9 mm(2) (2-11 barrels). Stimulus site dictated the domain of activity. To quantify the coding of whisker location, we applied the population d' measure of discriminability. Activity patterns elicited by two whiskers were highly discriminable at the initial cortical response; peak discriminability typically occurred within 16 msec of stimulus onset. To determine how widely information about stimulus location was distributed, we measured population d' while excluding response data from the on-center electrodes of the two tested whiskers. Response patterns remained discriminable, indicating that information about stimulus location was distributed across barrel cortex. Taken together, these results show that single-whisker deflections are represented in a multicolumn region constrained by barrel cortex map topography. The nature of this coding allows information about stimulus location to be coded extremely rapidly and unambiguously by one to two spikes per neuron.

Action Potentials↗
PubMedDate not supplied

The freeware AIDA interactive educational diabetes simulator--http://www.2aida.org--(2). Simulating glycosylated haemoglobin (HbA1c) levels in AIDA v4.3.

In 1996 an interactive educational diabetes simulator called AIDA was released without charge on the Internet as a non-commercial contribution to continuing diabetes education. Over the past 4+ years over 74,000 people have visited the AIDA Web pages at http://www.2aida.org and over 20,000 copies of the program have been downloaded from there free-of-charge. This article builds on the experience gained from the AIDA development, and the World Wide Web distribution of the software, and highlights some of the problems which users have reported with the program. An updated release of the software (AIDA v4.3) is described and the method applied for modelling glycosylated haemoglobin (HbA1c) levels within this new version of AIDA is documented. An overview is provided of the trialling and beta-testing of this latest release of the program, and the general concept of a 'virtual diabetic patient' that provides an electronic representation of a patient with diabetes--and which can be used for self-learning/teaching/demonstration purposes--is highlighted.

Computer Simulation↗
PubMed1986

Medical libraries and computers. The role of medical libraries in medical informatics.

The classic function of health sciences libraries is to build and maintain a knowledge base and to provide timely access to that collective memory for the purpose of learning, teaching, caring for patients, conducting research or managing an organization. The formats and representation of that knowledge base are changing rapidly, as are the methods and techniques for gaining access to information. Medical libraries have long used computers for cataloging and controlling records but are now shifting to acquiring, managing and distributing bibliographic and full-text information to local library "networks."

Computer Communication Networks↗
PubMedDate not supplied

Continuing professional development in public health medicine.

BACKGROUND: A survey of public health doctors was undertaken in the South Thames region to support coordination of continuing professional development (CPD) and to guide appropriate provision of educational activities for those working in the specialty. METHODS: A postal survey of 130 public health doctors in service, academic and military posts in the South Thames region of the UK. RESULTS: The response rate was 79% with good representation of grades and geographical areas. Public health doctors preferred personal and small group learning particularly focused on problem solving. Epidemiology as a basic discipline was rated highly by respondents, but senior doctors also showed preferences for topics beyond the textbooks. Respondents demonstrated that they had acquired such knowledge and skills, which could be shared in CPD sessions with colleagues. There was support for learning with and from non-medical colleagues. In the service posts structural change was ongoing, expectations of performance by management were high, and appropriate CPD was an expressed need, although time was seriously limited for this activity. CONCLUSIONS: Appropriate CPD could support the effectiveness of public health physicians in challenging circumstances. However, a critical evaluation of the relevance of CPD programmes to public health practice is essential if CPD is to support doctors in delivering the public health agenda in the changing structures within which they work.

Adult↗
PubMed2006

Understanding consolidation through the architecture of memories.

Following its encoding, a memory undergoes consolidation. It may be possible to deepen our understanding of the mechanisms supporting consolidation by considering the complex architecture of a memory. Any behavior can be split into multiple components. For example, when learning a new skill we simultaneously learn the movement and the goal of that movement. Each of these components has a distinct representation within a memory. The "off-line" processing of each component may follow different rules, providing an explanation for the variety of performance changes supported by consolidation. By viewing a memory as a representation with multiple components, it is possible to bridge the gap between the behavioral changes, which define consolidation, and the biological mechanisms that support those changes. This is partly because different memory components can be mapped onto different neural circuits. With an increased understanding of consolidation, it may become possible to modulate these off-line processes to improve psychiatric and neurological rehabilitation.

Animals↗
PubMed2000

Across the great divide: bridging the gap between understanding of toddlers' and older children's thinking.

Research on very young children's cognitive development differs greatly from research on cognitive development in older children. The differences include the questions that are asked, the methods that are used to address them, the measures that are employed to provide relevant evidence, and the level of detail at which children's knowledge is represented. The research approaches are so different that they create an impression that infants' and toddlers' thinking differs qualitatively from that of preschoolers and older children. This impression, however, may reflect differences in research approaches rather than differences in children's thinking. In the present study, we attempted to bridge this gap by applying to toddlers a type of process analysis that has proved fruitful in studies of older children. Overlapping waves theory, trial-by-trial strategy assessments, and microgenetic methods were used to analyze 1.5- and 2.5-year-olds' problem solving and learning. The results demonstrated that changes in toddlers' strategies could be assessed reliably on a trial-by-trial basis, that the changes followed the basic form predicted by the overlapping waves model, and that analyses of toddlers' strategies could tell us a great deal about both qualitative and quantitative aspects of their learning. A componential analysis of learning that previously had been applied to older children also proved useful for understanding toddlers' learning. The analysis specified that cognitive change frequently involves five components: acquisition of new strategies; strengthening of the strategies in their original context; improved mapping of strategies onto novel problems; increasingly refined choices among variants of the strategies; and increasingly skillful execution of the strategies. Independent measures of these components indicated that strategic development in toddlers involves improvements in all five components. Analyses of individual differences in learning showed that the effects of distal variables, such as age and sex, could be partially explained in terms of their influence on mastery of the components, but that the distal variables exercised additional direct effects as well. The process of learning in toddlers closely resembled that of older children in other ways as well. Like older children, toddlers use multiple strategies over the course of learning; their choices among strategies are quite adaptive from early on; their choices become progressively more adaptive as they gain experience with the task; they switch strategies not only from trial to trial but within a single trial; their transfer of learning from one problem to the next is primarily influenced by structural relations between problems but also is influenced by superficial features; they show utilization deficiencies early in learning that they gradually overcome; and they show individual differences in learning that fall into a few qualitatively distinct categories. Perhaps most striking, the 1.5- and 2.5-year-olds emerged as active learners, who continued to work out the lessons of previous instruction in the absence of further instruction. That is, they integrated the lessons of their own problem-solving efforts with the previous instruction in ways that magnified the initial effects of the instruction. Overall, the findings indicated that the gap can be bridged; that theories, methods, measures, and representations of knowledge typically used with older children can improve our understanding of toddlers' problem solving and learning as well.

Child Behavior↗
PubMed2003

Accuracy-based learning classifier systems: models, analysis and applications to classification tasks.

Recently, Learning Classifier Systems (LCS) and particularly XCS have arisen as promising methods for classification tasks and data mining. This paper investigates two models of accuracy-based learning classifier systems on different types of classification problems. Departing from XCS, we analyze the evolution of a complete action map as a knowledge representation. We propose an alternative, UCS, which evolves a best action map more efficiently. We also investigate how the fitness pressure guides the search towards accurate classifiers. While XCS bases fitness on a reinforcement learning scheme, UCS defines fitness from a supervised learning scheme. We find significant differences in how the fitness pressure leads towards accuracy, and suggest the use of a supervised approach specially for multi-class problems and problems with unbalanced classes. We also investigate the complexity factors which arise in each type of accuracy-based LCS. We provide a model on the learning complexity of LCS which is based on the representative examples given to the system. The results and observations are also extended to a set of real world classification problems, where accuracy-based LCS are shown to perform competitively with respect to other learning algorithms. The work presents an extended analysis of accuracy-based LCS, gives insight into the understanding of the LCS dynamics, and suggests open issues for further improvement of LCS on classification tasks.

Algorithms↗
PubMed2006

Implementing tutoring strategies into a patient simulator for clinical reasoning learning.

OBJECTIVE: This paper describes an approach for developing intelligent tutoring systems (ITS) for teaching clinical reasoning. MATERIALS AND METHODS: Our approach to ITS for clinical reasoning uses a novel hybrid knowledge representation for the pedagogic model, combining finite state machines to model different phases in the diagnostic process, production rules to model triggering conditions for feedback in different phases, temporal logic to express triggering conditions based upon past states of the student's problem solving trace, and finite state machines to model feedback dialogues between the student and TeachMed. The expert model is represented by an influence diagram capturing the relationship between evidence and hypotheses related to a clinical case. RESULTS: This approach is implemented into TeachMed, a patient simulator we are developing to support clinical reasoning learning for a problem-based learning medical curriculum at our institution; we demonstrate some scenarios of tutoring feedback generated using this approach. CONCLUSION: Each of the knowledge representation formalisms that we use has already been proven successful in different applications of artificial intelligence and software engineering, but their integration into a coherent pedagogic model as we propose is unique. The examples we discuss illustrate the effectiveness of this approach, making it promising for the development of complex ITS, not only for clinical reasoning learning, but potentially for other domains as well.

Computer Simulation↗
PubMed2005

NMDA antagonist MK-801 does not interfere with the use of spatial representation in a familiar environment.

There is disagreement among researchers concerning whether glutamatergic N-methyl-D-aspartate (NMDA) receptors play a role in constructing spatial representations. Therefore, the authors reexamined the effects of the NMDA antagonist on a spatial discrimination task using rats in a water pool. The authors confirmed that MK-801 impaired acquisition of the spatial discrimination task (Experiment 1). When rats were pretrained before drug treatment, MK-801 induced learning deficits in the novel environment but not in the familiar environment (Experiment 2). Moreover, in a familiar environment, MK-801 did not impair spatial learning, even when the task was completely novel for the rats (Experiment 3). These results suggest that NMDA receptors play an important role in the construction of spatial representations but not in the use of them.

Animals↗
PubMed2002

Hippocampus lesions impair landmark array spatial learning in homing pigeons: a laboratory study.

Hippocampal (HF)-lesioned pigeons display impaired homing ability when flying over familiar terrain, where they are presumably relying on a map-like representation of familiar landmarks to navigate. However, research carried out in the field precludes a direct test of whether hippocampal lesions compromise the ability of homing pigeons to navigate by familiar landmarks. To examine more thoroughly the relationship between hippocampus and landmark spatial learning, control, neostriatum-lesioned, and HF-lesioned homing pigeons were trained on two open field, laboratory, conditional discrimination tasks. One was a visual landmark array task, and the other was a room color discrimination task. For the tasks, the correct of three differently colored food bowls was determined by the spatial relationship among a group of five landmarks and room color, respectively. Intact control birds successfully learned both tasks, while neostriatum-lesioned birds successfully learned the landmark array task-the only task on which they were trained. By contrast, HF-lesioned birds successfully learned the room color task but were unable to learn the landmark array task. The data support the hypothesis that homing performance deficits observed in the field following hippocampal lesions are in part a consequence of an impairment in the ability of lesioned pigeons to use familiar visual landmarks for navigation.

Animals↗
PubMedDate not supplied

An inductive algorithm approach to knowledge acquisition for expert system development. A pilot study.

Knowledge acquisition, which consists of knowledge elicitation and knowledge representation, often is considered the weakest link in the design of expert systems. Systems frequently are built on the knowledge of one expert and require extensive use of knowledge engineering techniques to elicit this knowledge from the expert. Inductive algorithms are a potential alternative method of knowledge acquisition for expert system development. The aim of this pilot study was to examine the feasibility of applying machine learning techniques, specifically, inductive algorithms, to an existing research database as a method for knowledge elicitation and knowledge representation for expert system development. Two inductive algorithms (C4 and Classification and Regression Trees [CART]) that generate decision trees were selected for the analysis using a data set of 201 patients hospitalized for Pneumocystis carinii pneumonia. Neither C4 nor CART produced trees with an accuracy that was significantly better than the baseline accuracy (71.3%) for prediction of outcome in the data set. The mean accuracy of the C4 decision trees was below baseline and the mean accuracy of CART decision trees was 74.6%. The experts found both algorithms comprehensible, but not adequate, and identified important missing predictor variables. The study findings suggest that additional research is needed to examine the appropriate use of inductive algorithms in the transformation of nursing data and information into nursing knowledge.

Algorithms↗
PubMed2002

Final consonant discrimination in children: effects of phonological disorder, vocabulary size, and articulatory accuracy.

Preschool-age children with phonological disorders were compared to their typically developing age peers on their ability to discriminate CVC words that differed only in the identity of the final consonant in whole-word and gated conditions. The performance of three age groups of typically developing children and adults was also assessed on the same task. Children with phonological disorders performed more poorly than age-matched peers, and younger typically developing children performed more poorly than older children and adults, even when the entire CVC word was presented. Performance in the whole-word condition was correlated with receptive vocabulary size and a measure of articulatory accuracy across all children. These results suggest that there is a complex relationship among word learning skills, the ability to attend to fine phonetic detail, and the acquisition of articulatory-acoustic and acoustic-auditory representations.

Adult↗
PubMed2006

Student perceptions about the characteristics of an effective discussion during the reporting phase in problem-based learning.

To explore student perceptions of factors contributing to the effectiveness of discussions in the reporting phase of the problem-based learning (PBL) process, where students report and synthesise the results of self-study. Forty-eight Year 1 and 2 medical students participated in 6 focus group interviews about the characteristics of effective group discussions and possible improvements. The data were analysed qualitatively in several stages. The analysis yielded 4 main characteristics of effective discussions: asking for, giving and receiving explanations; integrating and applying knowledge; discussing differences with regard to learning content, and guiding and monitoring the content and the group process of the discussion. Integrating and applying knowledge included structuring, relating and summarising information and providing examples from practice. Discussing different opinions included discussing a variety of literature resources and disagreements. The main learning effects mentioned by the students were retention, understanding, integration and application of knowledge. Students have clear ideas about what promotes effective discussions during the reporting phase. Their PBL experience has provided them with some insights that are in line with theory and research on collaborative learning. Future research should examine differences between student and tutor perceptions of the quality of discussions. Introductions to PBL for students and tutors should include training in asking open but focused questions, supporting explanations with arguments and dealing with conflicts about learning content. Tutors should be trained in giving effective and personal feedback. Collaborative creation of external knowledge representations (i.e. concept maps) should be advocated, as should variety of literature resources.

Adolescent↗
PubMed1991

Contextual dependencies in motor skills.

The development of contextual dependencies during motor skill acquisition was examined. Environmental context was varied along intentional and incidental dimensions. Intentional stimuli were defined as essential for achieving skilled performance, whereas incidental stimuli were defined as those that have the potential to become associated with specific tasks due to their selective presence in the learning environment. Experiment 1 demonstrated the occurrence of contextual dependencies for the learning of four-key typing sequences. Contextual dependencies were diminished in Experiment 2 when the number of keys used in the sequences was reduced. In Experiment 3, a retention condition was incorporated, in which both the intentional and the incidental stimuli were not available; this confirmed that task difficulty mediated the development of contextual dependencies. These findings are discussed with respect to the incorporation of environmental contextual stimuli with memorial representations of movement information.

Adult↗
PubMed2003

The role of motor imagery in learning a totally novel movement.

The aim of the present study is to gain more insight into the mechanisms underlying mental practice. The question of whether a totally novel movement may be learned by mental practice was investigated. Healthy young adults had to learn the abduction of the big toe (dominant right foot) without moving the other toes or the foot. The subjects were divided into two groups: subjects who were absolutely unable to abduct their big toe ("absolute zero" group) and subjects who were able to abduct their toe to some extent but showed clear room for improvement ("already doing it" group). Two separate experiments were executed. In the first experiment, 37 absolute-zero subjects had to practice, mentally or physically, the target movement. In the second experiment 40 already-doing-it subjects had to improve their toe-abduction skill. The results showed that absolute-zero subjects could not acquire the toe-abduction movement by means of mental practice. Only subjects who physically practiced the target movement improved significantly. Subjects who had some experience in the task (already-doing-it subjects) improved significantly after mental practice as well as after physical practice. The results seem to indicate that it is more plausible to explain the learning effects of mental practice in terms of a "top-down" mechanism based on the activation of a central representation of the movement than in terms of a peripheral "bottom-up" mechanism based on the activation of muscles.

Adult↗