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.

SourceResearch connectionFrom yearThrough year

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

PubMed2001

PSD95 protein level rises in murine somatosensory cortex after sensory training.

Proteins of the postsynaptic density are implicated in mechanisms of synaptic plasticity. We examined involvement of PSD95 and alphaCaMkII in learning-dependent plastic changes of representational maps in somatosensory cortex of mice. The barrel cortex of mice was examined following a 3 day long classical conditioning training, in which activation of facial vibrissae was linked to an aversive stimulus. This procedure produced expansion of cortical representations of vibrissae involved in the training. In subcellular fraction enriched in postsynaptic densities from the barrel cortex, it was estimated by Western blotting that the level of PSD95 increased after the training by about 50%, while the level of CaMkII remained unchanged. The results indicate involvement of PSD95 in learning-dependent cortical plasticity.

Afferent Pathways↗

PubMed2003

A motor learning strategy reflects neural circuitry for limb control.

During motor skill acquisition, the brain learns a mapping between intended limb motion and requisite muscular forces. We propose that regions where sensory and motor representations overlap are crucial for motor learning. In primary motor cortex, for example, cells that modulate their activity for motor actions at a joint tend to receive input from that same portion of the periphery. We predict that this correspondence reflects a default strategy--a Bayesian prior--in which subjects tend to associate loads at a joint with motion at that joint (local sensorimotor association) when there is ambiguity regarding the nature of the load. As predicted, we found that in the presence of uncertainty, humans inappropriately generalized elbow loads as though they were based on elbow velocity. Generalization improved when we reduced uncertainty by decreasing coupling between elbow velocity and load during training. These results illustrate a key link between motor learning and the underlying neural circuitry.

Extremities↗

PubMed2003

Context-sensitive elemental theory.

My theories of associative learning, like those of N. J. Mackintosh and almost all learning theorists, have employed elemental representations of the stimuli involved. We must take notice when two important contributors to elemental theory, J. M. Pearce and W. K. Estes, find sufficient problems with the theory type to cause them to defect from it. I will describe some of the essential problems, concerning the substantial influence of context on learning and retrieval, characterize the different responses of Pearce and Estes, and, then, propose a variation on a recently developed elemental model that was similarly inspired. The resulting elemental theory has a close quantitative relationship to the product-rule of Estes and D. L. Medin, and may help us to rationalize how the same formal experimental design can sometimes produce results that favour the configural interpretation of Pearce and at other times the elemental interpretation of R. A. Rescorla and A. R. Wagner, as these have often been pitted against each other.

Association Learning↗

PubMed2006

Are people with high and low mental rotation abilities differently susceptible to the alignment effect?

We investigated whether the alignment effect (Levine et al, 1982 Journal of Experimental Psychology: General 111 157-175) is influenced by mental rotation abilities. In two experiments, groups of undergraduate students with high and low performance in mental rotation tasks were required to study either schematic (experiment 1) or more complex (experiment 2) maps, and to perform a number of pointing tasks adopting a perspective which could be aligned, misaligned (45 degrees, 135 degrees), or counteraligned (180 degrees) with the perspective assumed during learning. Cognitive styles in spatial representation have also been considered. Results of experiment 1 show that people with low performance in mental rotation tasks prefer to adopt a representation of space focused more on landmarks. Their performance in the pointing tasks depends on the alignment conditions, with more errors in the counteraligned condition followed by the two misaligned and aligned ones. In contrast to this, high-ability mental rotators prefer survey and route spatial representations and are affected only by the aligned and non-aligned conditions. In the second experiment, practice was studied as a function of mental rotation and alignment. The group high in mental rotation ability was found to be free from the alignment effect in the pointing tasks performed after the final of four learning phases.

Cognition↗

PubMed1993

Cognitive aspects of classical conditioning.

Cognitive processes have been increasingly implicated in Pavlovian conditioning. Research in the past year has focused on questions of stimulus selection and the internal representation of events and the relations between them. Recent data support negative feedback models of selection that assume conditioning-dependent changes in processing of conditioned and unconditioned stimulus events, and suggest potential neural mechanisms that may underlie these processes. New models of conditioning propose a more detailed representation of individual conditioning episodes than traditionally assumed. The results of investigations into conditional discrimination learning imply a hierarchical organization of event representations, and illustrate the importance of conditioned modulatory processes as distinct from response elicitation.

Animals↗

PubMed2000

Neural correlates of topographic mental exploration: the impact of route versus survey perspective learning.

There are two major sources of information to build a topographic representation of an environment, namely actual navigation within the environment (route perspective) and map learning (survey perspective). The aim of the present work was to use positron emission tomography (PET) to compare the neural substrate of the topographic representation built from these two modes. One group of subjects performed a mental exploration task in an environment learned from actual navigation (mental navigation task). Another group of subjects performed exploration in the same environment learned from a map (mental map task). A right hippocampal activation common to both mental navigation and mental map tasks was evidenced and may correspond the neural substrate of a "dual-perspective" representation. The parahippocampal gyrus was additionally activated bilaterally during mental navigation only. These results suggest that the right hippocampus involvement would be sufficient when the representation incorporates essentially survey information while the bilateral parahippocampal gyrus would be involved when the environment incorporates route information and includes "object" landmarks. The activation of a parietofrontal network composed of the intraparietal sulcus, the superior frontal sulcus, the middle frontal gyrus, and the pre-SMA was observed in common for both mental navigation and mental map and is likely to reflect the spatial mental imagery components of the tasks.

Adult↗

PubMed1995

Learning surgical technical skills.

Training issues raised by the recent introduction of laparoscopic surgical techniques led to this analysis of motor-skill learning principles as they apply specifically to the learning of technical surgical skills. The most accepted theories of motor-skill learning are presented, not as opposing views, but as complementary constructs. The behaviourist school of thought's main contribution is the executive routine or knowledge of the steps of a procedure. Schmidt's schema theory and MacKay's node theory suggest that perceptual information may play an important role in the quality of the performance. The conclusions reached from neuropsychologic testing experiments on surgeons are that visuospatial perceptual skills (the ability to represent mentally the physical environment and the movement to be performed) are the major determinants of surgical technical performance. Learners should make use of learning strategies that improve mental representation of a skill and the corresponding anatomy. Specific strategies discussed include imagery, mental practice and a systematic review of performance that focuses on the perceptual feedback received by the learner.

General Surgery↗

PubMed1999

A model of probabilistic category learning.

A new connectionist model (named RASHNL) accounts for many "irrational" phenomena found in nonmetric multiple-cue probability learning, wherein people learn to utilize a number of discrete-valued cues that are partially valid indicators of categorical outcomes. Phenomena accounted for include cue competition, effects of cue salience, utilization of configural information, decreased learning when information is introduced after a delay, and effects of base rates. Experiments 1 and 2 replicate previous experiments on cue competition and cue salience, and fits of the model provide parameter values for making qualitatively correct predictions for many other situations. The model also makes 2 new predictions, confirmed in Experiments 3 and 4. The model formalizes 3 explanatory principles: rapidly shifting attention with learned shifts, decreasing learning rates, and graded similarity in exemplar representation.

Cues↗

PubMed2004

Transitive behavior in hippocampal-lesioned pigeons.

The hippocampus of birds and mammals is critical for the learning of map-like memory representations of environmental space. It has been suggested that the hippocampus of rats also participates in non-spatial relational learning, including the learning of non-spatial transitive relationships among odor stimuli [Bunsey and Eichenbaum, Nature 1996]. Although transitive-like learning has been demonstrated in a variety of vertebrate species, from a comparative perspective the role of the hippocampus in this form of learning has not been tested in other amniote groups. We trained control and hippocampal-lesioned homing pigeons on a series of visual, non-spatial, go/no-go conditional discriminations and then tested them on novel transitivity probe trials. The hippocampal-lesioned pigeons were as successful as control pigeons in responding appropriately to correct and incorrect transitivity pairs. The finding that the homing pigeon hippocampal formation is not necessary for solving this serial, conditional discrimination task is important for further understanding hippocampal function across species, and represents one of the few studies that have attempted to localize a brain region responsible for the phenomenon of transitive behavior learning.

Animals↗

PubMed2000

Implicit and explicit learning of event sequences: evidence for distinct coding of perceptual and motor representations.

Event-related brain potentials (ERPs) of 21 subjects were recorded in a choice reaction time task with a repeating eight-element long stimulus sequence. The regular event sequence was sometimes interrupted by 'perceptual' or by 'motor deviants' which both replaced an expected stimulus but either preserved or violated the sequence of motor responses. Response times confirmed that all subjects had acquired some knowledge of the sequential dependencies. By means of a post-experimental free recall and recognition test, subjects were classified as having either explicit or implicit knowledge of the event sequence. The ERPs showed different effects for different types of stimuli and the two groups. In the group of explicit learners, a larger N200 component was evoked by both types of deviants and a larger P300 by motor deviants only. In the group of implicit learners these 'perceptual components' remained unaffected. In contrast, in both groups of subjects the lateralized readiness potential (LRP) which accompanied motor deviants revealed a partial activation of the to be expected but incorrect response, i.e. motor learning. These results suggest that explicit learners acquire knowledge about both, stimulus and response dependencies while implicit learners acquire knowledge about response dependencies only.

Adult↗

PubMed1997

Distributed Learning, Recognition, and Prediction by ART and ARTMAP Neural Networks.

A class of adaptive resonance theory (ART) models for learning, recognition, and prediction with arbitrarily distributed code representations is introduced. Distributed ART neural networks combine the stable fast learning capabilities of winner-take-all ART systems with the noise tolerance and code compression capabilities of multilayer perceptrons. With a winner-take-all code, the unsupervised model dART reduces to fuzzy ART and the supervised model dARTMAP reduces to fuzzy ARTMAP. With a distributed code, these networks automatically apportion learned changes according to the degree of activation of each coding node, which permits fast as well as slow learning without catastrophic forgetting. Distributed ART models replace the traditional neural network path weight with a dynamic weight equal to the rectified difference between coding node activation and an adaptive threshold. Thresholds increase monotonically during learning according to a principle of atrophy due to disuse. However, monotonic change at the synaptic level manifests itself as bidirectional change at the dynamic level, where the result of adaptation resembles long-term potentiation (LTP) for single-pulse or low frequency test inputs but can resemble long-term depression (LTD) for higher frequency test inputs. This paradoxical behavior is traced to dual computational properties of phasic and tonic coding signal components. A parallel distributed match-reset-search process also helps stabilize memory. Without the match-reset-search system, dART becomes a type of distributed competitive learning network.

Journal Article↗

PubMed2004

Stimulus generalization in two associative learning processes.

Recent studies involving nonlinear discrimination problems suggest that stimuli in human associative learning are represented configurally with narrow generalization, such that presentation of stimuli that are even slightly dissimilar to stored configurations weakly activate these configurations. The authors note that another well-known set of findings in human associative learning, cue-interaction phenomena, suggest relatively broad generalization. Three experiments show that current models of human associative learning, which try to model both nonlinear discrimination and cue interaction as the result of 1 process, fail because they cannot simultaneously account for narrow and broad generalization. Results suggest that human associative learning involves (a) an exemplar-based process with configural stimulus representation and narrow generalization and (b) an adaptive learning process characterized by broad generalization and cue interaction.

Cues↗

PubMed2006

Are sparse-coding simple cell receptive field models physiologically plausible?

Olshausen and Field (1996) developed a simple cell receptive field model for natural scene processing in V1, based on unsupervised learning and non-orthogonal basis function optimization of an overcomplete representation of visual space. The model was originally tested with an ensemble of whitened natural scenes, simulating pre-cortical filtering in the retinal ganglia and lateral geniculate nucleus, and the basis functions qualitatively resembled the orientation-specific responses of V1 simple cells in the spatial domain. In this study, the quantitative tuning responses of the basis functions in the spectral domain are estimated using a Gaussian model, to determine their goodness-of-fit to the known bandwidths of simple cells in primate V1. Five simulation experiments which examined key features of the model are reported: changing the size of the basis functions; using a complete versus over-complete representation; changing the sparseness factor; using a variable learning rate; and mapping the basis functions with a whitening spatial function. The key finding of this study is that across all image themes, basis function sizes, number of basis functions, sparseness factors and learning rates, the spatial-frequency tuning did not closely resemble that of primate area 17 -- the model results more closely resembled the unclassified cat neurones of area 19 with a single exception, and not area 17 as predicted.

Action Potentials↗

PubMed2004

Context-independent directional cue learning by hippocampal place cells.

In a symmetrical environment possessing no other polarizing visual cues, the spatially localized firing of hippocampal place cells can be primarily orientated by a reliable distal visual stimulus, such as a white cue card. However, if such a directional cue is made unreliable by being frequently moved in full view of the rat, the rat's internal sense of direction comes, over the course of a few days, to control the orientation of place fields instead. We investigated whether this simple form of 'cue-instability' learning would transfer to a new context, in which the firing patterns of the place cells become reorganized and in which a new spatial representation is thus active. We found that after cue-instability learning, the 'remapped' place field representation in the new environment was also orientated by the internal sense of direction of the rat rather than by the cue card, showing that the cue learning generalized from one context (and hence spatial representation) to another. This contrasts with another kind of place cell learning, in which the cells can acquire the ability to discriminate two spatial locations in one context but do not transfer this discrimination to a new context. We discuss the different effects of context changes on learned place cell activity in terms of the possible architecture of the inputs to place cells.

Animals↗

PubMed2004

Learning multiview face subspaces and facial pose estimation using independent component analysis.

An independent component analysis (ICA) based approach is presented for learning view-specific subspace representations of the face object from multiview face examples. ICA, its variants, namely independent subspace analysis (ISA) and topographic independent component analysis (TICA), take into account higher order statistics needed for object view characterization. In contrast, principal component analysis (PCA), which de-correlates the second order moments, can hardly reveal good features for characterizing different views, when the training data comprises a mixture of multiview examples and the learning is done in an unsupervised way with view-unlabeled data. We demonstrate that ICA, TICA, and ISA are able to learn view-specific basis components unsupervisedly from the mixture data. We investigate results learned by ISA in an unsupervised way closely and reveal some surprising findings and thereby explain underlying reasons for the emergent formation of view subspaces. Extensive experimental results are presented.

Algorithms↗

PubMed2005

Acquired dyslexia in a Turkish-English speaker.

The Turkish script is characterised by completely transparent bidirectional mappings between orthography and phonology. To date, there has been no reported evidence of acquired dyslexia in Turkish speakers leading to the naïve view that reading and writing problems in Turkish are probably rare. We examined the extent to which phonological impairment and orthographic transparency influence reading disorders in a native Turkish speaker. BRB is a bilingual Turkish-English speaker with deep dysphasia accompanied by acquired dyslexia in both languages. The main findings are an effect of imageability on reading in Turkish coincident with surface dyslexia in English and preserved nonword reading. BRB's acquired dyslexia suggests that damage to phonological representations might have a consequence for learning to read in Turkish. We argue that BRB's acquired dyslexia has a common locus in chronic underactivation of phonological representations in Turkish and English. Despite a common locus, reading problems manifest themselves differently according to properties of the script and the type of task.

Aged↗

PubMed2004

Statistical shape modeling of low level visual area borders.

This paper proposes a statistical modeling of functional landmarks delimiting low level visual areas which are highly variable across individuals. Low level visual areas are first precisely delineated by fMRI retinotopic mapping which provides detailed information about the correspondence between the visual field and its cortical representation. The model is then built by learning the variability within a given training set. It relies on an appropriate data representation and on the definition of an intrinsic coordinate system common to all visual maps. This allows to build a consistent training set on which a principal component analysis is eventually applied. Our approach constitutes a first step toward a functional landmark-based probabilistic atlas of low level visual areas.

Adult↗