Search PubMed⌕ Search

SEARCH · Search PubMed

Results for “GENERALIZATION 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 451 records · Page 25Linked to original sources
PubMed2000

Nonmonotonic generalization bias of Gaussian mixture models.

Theories of learning and generalization hold that the generalization bias, defined as the difference between the training error and the generalization error, increases on average with the number of adaptive parameters. This article, however, shows that this general tendency is violated for a gaussian mixture model. For temperatures just below the first symmetry breaking point, the effective number of adaptive parameters increases and the generalization bias decreases. We compute the dependence of the neural information criterion on temperature around the symmetry breaking. Our results are confirmed by numerical cross-validation experiments.

Computer Simulation↗
PubMed1996

Radial Basis Function Networks: Generalization in Over-realizable and Unrealizable Scenarios.

Learning and generalization in a two-layer radial basis function network, with fixed centres of the basis functions, is examined within a stochastic training paradigm. Employing a Bayesian approach, expressions for generalization error are derived under the assumption that the generating mechanism (teacher) for the training data is also a radial basis function network, but one for which the basis function centres and widths need not correspond to those of the student network. The effects of regularization, via a weight decay term, are examined. The cases in which the student has greater representational power than the teacher (over-realizable), and in which the teacher has greater power than the student (unrealizable) are studied. Dependence on knowing the centres of the teacher is eliminated by introducing a single degree-of-confidence parameter. Finally, simulations are performed which validate the analytic results. Copyright 1996 Elsevier Science Ltd.

Journal Article↗
PubMed1983

Evaluation of a tactile vocoder for work recognition.

Normal subjects learned to identify words through a tactile vocoder. The vocoder employed 16 filter channels, each with a bandwidth of 1/3 octave, with center frequencies ranging from 200-8000 Hz. The output of each filter was detected and after logarithmic amplification the resulting outputs were transmitted to a 16-channel solenoid array placed on the subject's ventromedial forearm. Words, spoken "live voice," were used as stimuli; the subject was able to feel both the vocalization of the reader and herself and, most importantly, extensive training was provided. In 40.5 h one subject learned 70 words and a second subject reached criterion on 150 words in the comparatively short time of 55 h. Words that were poorly identified initially were identified more readily with increased experience. Phonetic identification tests showed that the features of voicing, nasality, and frication were reliably recognized, indicating the tactile vocoder will be useful in providing information to complement lipreading. Finally, subjects learned rapidly to generalize word-learning to unfamiliar readers.

Communication Devices for People with Disabilities↗
PubMed1994

Problem-based medical education in general practice: experience from Linköping, Sweden.

Medical services which are better suited to the needs of the community are increasingly demanded, for example by national governments, and depend to a great extent on education of the new generations of doctors. Problem-based learning in general practice at the community level represents a method of providing appropriate education. This paper describes some of the experiences of undergraduate and continuing medical education at the Faculty of Health Sciences, Linköping University, Sweden, which since 1986 have been entirely problem-based. General practice forms the backbone of the curriculum and is its largest subject. Problem-based learning in general practice and quality assurance have much in common. This applies over the spectrum of lifelong education, from the undergraduate curriculum through vocational training to later stages of continuing medical education. Involvement and understanding, a feeling of purpose, sharing and cooperation and a wish for self-improvement are all stimulated by the problem-based method. This approach has helped in the creation of a curriculum with general practice as the largest contributor, with early contact with patients in the community, and with ordinary general practitioners as teachers, tutors and examiners in spite of a relatively poorly developed primary care organization.

Curriculum↗
PubMed1986

Principles of learning.

This article discusses some general principles of learning as well as possible constraints and how such principles can apply to horses. A brief review is presented of experiments that were designed to assess learning in horses. The use of behavior modification techniques to treat behavior problems in horses is discussed and several examples of the use of these techniques are provided.

Animals↗
PubMed1964

GENERALIZATION GRADIENTS FOLLOWING TWO-RESPONSE DISCRIMINATION TRAINING.

Stimulus generalization was investigated using institutionalized human retardates as subjects. A baseline was established in which two values along the stimulus dimension of auditory frequency differentially controlled responding on two bars. The insertion of the test probes disrupted the control established to the two S(D)s during training. The discrimination was recovered between each test probe and the resulting gradients were stable across 10 test sessions. These gradients, supported by other two-response generalization studies, indicate that this type of two-response discrimination training divides the stimulus dimension into two functional classes separated by a region of transition from one class to the other. Each stimulus value in a class, which extends from an S(D) outward to the functional limit of the dimension, controls a similar proportion of the two responses as each other value in the class. All values on the stimulus dimension control identical response rates with an absence of the usual generalization decrement. The latency of the initial response, however, shows a bimodal gradient with the modes at the S(D) values.

Adolescent↗
PubMed1997

An anatomically structured sensory-motor sequence learning system displays some general linguistic capacities.

The capacity in primates to master temporal-sequential constraints from the external world might provide a basis for accommodating similar constraints in language. While the neural specialization required for language clearly distinguishes man from the other primates, it is less clear to what extent this specialization constitutes a drastic neurophysiological divergence versus a variation on an existing sequencing capability. In an effort to address this issue, an anatomically structured neural network model, previously developed to reproduce complex sensory-motor sequences and the corresponding single-unit recordings from primate prefrontal cortex, is studied in a simple linguistic context. The model is presented sentences from a small language and demonstrates a simple capacity to "understand" and generalize at different levels. Interactions between variations on (a) the model architecture and (b) the target language structure agree with data from crosslinguistic aphasia studies. These results support the hypothesis that a brain architecture for nonlinguistic cognitive functions (in this case sensory-motor sequencing) can provide a basis for a general sequence processing component of linguistic function.

Brain↗
PubMed2005

Supervised learning of postural tasks in patients with poststroke hemiparesis, Parkinson's disease or cerebellar ataxia.

Supervised learning of different postural tasks in patients with lesions of the motor cortex or pyramidal system (poststroke hemiparesis: 20 patients), nigro-striatal system (Parkinson's disease: 33 patients) and cerebellum (spinocerebellar ataxia: 37 patients) was studied. A control group consisted of 13 healthy subjects. The subjects stood on a force platform and were trained to change the position of the center of pressure (CP) presented as a cursor on a monitor screen in front of the patient. Subjects were instructed to align the CP with the target and then move the target by shifting the CP in the indicated direction. Two different tasks were used. In "Balls", the target (a ball) position varied randomly, so the subject learned a general strategy of voluntary CP control. In "Bricks", the subject had to always move the target in a single direction (downward) from the top to the bottom of the screen, so that a precise postural coordination had to be learned. The training consisted of 10 sessions for each task. The number of correctly performed trials for a session (2 min for each task) was scored. The voluntary control of the CP position was initially impaired in all groups of patients in both tasks. In "Balls", there were no differences between the groups of the patients on the first day. The learning course was somewhat better in hemiparetic patients than in the other groups. In "Bricks", the initial deficit was greater in the groups of parkinsonian and cerebellar patients than in hemiparetic patients. However, learning was more efficient in parkinsonian than in hemiparetic and cerebellar patients. After 10 days of training, the hemiparetic and cerebellar patients completed the acquisition at a certain level whereas the parkinsonian patients showed the ability for further improvement. The results suggest that motor cortex, cerebellum, and basal ganglia are involved in voluntary control of posture and learning different postural tasks. However, these structures play different roles in postural control and learning: basal ganglia are mainly involved in learning a general strategy of CP control while the function of the motor cortex chiefly concerns learning a specific CP trajectory. The cerebellum is involved in both kinds of learning.

Adult↗
PubMed1996

Attitudes of general practitioners to caring for people with learning disability.

BACKGROUND: The views of general practitioners on their increasing role in caring for people with learning disability in the community are not known. AIM: A study was carried out to assess the views of general practitioners with regard to providing routine care, organizing health promotion and specific health checks for people with learning disability and the role of specialists. METHOD: A postal questionnaire was sent to all 242 general practitioners in Gwent, south Wales. Participants had to mark their level of agreement with 20 attitude statements regarding learning disability. RESULTS: A total of 126 general practitioners (52%) responded. Respondents generally agreed that general practitioners were responsible for the medical care of people with learning disability. Respondents tended to be opposed to providing regular structured health promotion for people with learning disability, such as annual health checks and assessing hearing and eyesight. Specialist services were generally valued by respondents. CONCLUSION: General practitioners largely accepted their role as primary health care providers for people with learning disability. In contrast, their role as providers of health promotion for this patient group was not generally accepted. Further research into the appropriateness and opportunity costs of health screening for people with learning disability is needed.

Attitude of Health Personnel↗
PubMed2006

Evaluation of the errorless learning technique in children with traumatic brain injury.

OBJECTIVE: To compare errorless learning with trial-and-error (T&E) learning of declarative facts in children with memory disorders secondary to traumatic brain injury (TBI). DESIGN: Retrospective within-subjects concurrent treatment design. SETTING: Participants' school or home. PARTICIPANTS: Thirty-four children, ages 6 to 18 years, with mild, moderate, or severe postacute TBI who met criteria for memory impairment. INTERVENTION: Conditions consisted of an errorless learning method and a T&E method. Within a session, half the items were taught with the errorless learning method and half with the T&E method. Each child received two 1-hour sessions a week for 7 weeks. MAIN OUTCOME MEASURES: Relative effectiveness of errorless learning and T&E methods for (1) initial learning and (2) retention over time for learned items. RESULTS: There was an advantage for T&E on initial learning. In children with mild, but not moderate or severe TBI, 2-day retention was better with the errorless learning technique; 7-day retention was better with errorless learning in young children with mild TBI. Seventy-seven-day retention revealed an advantage for errorless learning in younger children with severe TBI. CONCLUSIONS: Findings did not support errorless learning as a generalized intervention for learning difficulties after TBI or identify specific age- or injury-severity groups that benefited from this technique.

Adolescent↗
PubMed1990

Conceptual learning in capuchin monkeys.

This paper reviews concept learning in Cebus monkeys, focussing on their ability to use the identity relation, oddity and natural concepts. Capuchins are similar to other primate genera in their use of these concepts. The extant data on learning in primates generally reflect historical concerns with general processes of learning. An alternative approach which considers the tasks the animal faces in its natural environment may be better suited to the discovery of species-unique characteristics of learning. This approach has not yet been applied to Cebus.

Animals↗
PubMed1979

Serial learning in the pigeon.

Three pigeons learned to peck four colors in a particular sequence, regardless of how these colors were positioned on four response keys and without feedback following each response. This demonstrates that serial learning is possible in subprimate animals.

Animals↗
PubMed2001

A hybrid model of categorization.

Category learning is often modeled as either an exemplar-based or a rule-based process. This paper shows that both strategies can be combined in a cognitive architecture that was developed to model other task domains. Variations on the exemplar-based random walk (EBRW) model of Nosofsky and Palmeri (1997b) and the rule-plus-exception (RULEX) rule-based model of Nosofsky, Palmeri, and McKinley (1994) were implemented in the ACT-R cognitive architecture. The architecture allows the two strategies to be mixed to produce classification behavior. The combined system reproduces latency, learning, and generalization data from three category-learning experiments--Nosofsky and Palmeri (1997b), Nosofsky et al., and Erickson and Kruschke (1998). It is concluded that EBRW and ACT-R have different but equivalent means of incorporating similarity and practice. In addition, ACT-R brings a theory of strategy selection that enables the exemplar and the rule-based strategies to be mixed.

Cognition↗
PubMed1994

Assessing learning abilities, readiness for education.

This article gives a brief description of the history of perioperative education, as well as an overview of the theories of teaching and learning. It discusses general principles of learning, factors facilitating learning, and factors hindering learning. Learning needs and interventions for special populations such as the pediatric patient, the elderly patient, and the ethnic patient are also identified and briefly discussed.

Humans↗
PubMed1991

A demonstration of the graded nature of the generalization function of drug discrimination learning within the conditioned taste aversion procedure.

Although control of discriminative performance will often generalize to different doses of the training drug or to drugs from the same class as the training drug, the nature of this generalization is unknown. Prior work has suggested that the generalization is primarily quantal in nature with animals displaying either vehicle-appropriate or drug-appropriate responding, depending upon their detection of the drug stimulus. It has been questioned whether this quantal nature of generalization reflects a characteristic response to drug stimuli or whether such responding is a function of the specific training and testing procedures used to establish and measure drug discrimination learning. The present paper evaluated this issue by analyzing the generalization functions of individual subjects trained and tested within one specific drug discrimination procedure, i.e. the conditioned taste aversion design. Responding within this design was generally graded. It is clear that quantal responding is not a necessary outcome of drug generalization assessments and that the nature of generalization in drug discrimination learning is a function of the specific procedure utilized in training and testing the discrimination. The results of the present analysis are discussed in terms of other recent work reporting graded functions.

Journal Article↗