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 1,207 records · Page 67Linked to original sources
PubMed1995

Specificity of perceptual processing in rereading spatially transformed materials.

While most studies using the task of reading spatially transformed text do not reveal evidence of specific perceptual transfer, a study by Masson (1986, Experiment 3) provides clear evidence of such effects. Several experiments were designed to identify the basis for this empirical discrepancy. The only substantive evidence of specific perceptual transfer occurred when the words were presented in an unfamiliar typography, although each study suggested a trend toward perceptual specificity effects. The results are discussed in terms of Graf and Ryan's (1990) ideas about the role of distinctive memory representations.

Adult↗
PubMed1997

Features and their configuration in face recognition.

Tanaka and Farah (1993) have proposed a holistic approach to face recognition in which information about the features of a face and their configuration are combined together in the face representation. An implication of the holistic hypothesis is that alterations in facial configuration should interfere with retrieval of features. In four experiments, the effect of configuration on feature recognition was investigated by creating two configurations of a face, one with eyes close together and one with eyes far apart. After subjects studied faces presented in one of the two configurations (eyes-close or eyes-far), they were tested for their recognition of features shown in isolation, in a new face configuration, and in the old face configuration. It was found that subjects recognized features best when presented in the old configuration, next best in the new configuration, and poorest in isolation. Moreover, subjects were not sensitive to configural information in inverted faces (Experiment 2) or nonface stimuli (i.e., houses; Experiments 3 and 4). Importantly, for normal faces, altering the spatial location of the eyes not only impaired subjects' recognition of the eye features but also impaired their recognition of the nose and mouth features-features whose spatial locations were not directly altered. These findings emphasize the interdependency of featural and configural information in a holistic face representation.

Adult↗
PubMed1999

Constructing biological knowledge bases by extracting information from text sources.

Recently, there has been much effort in making databases for molecular biology more accessible and interoperable. However, information in text form, such as MEDLINE records, remains a greatly underutilized source of biological information. We have begun a research effort aimed at automatically mapping information from text sources into structured representations, such as knowledge bases. Our approach to this task is to use machine-learning methods to induce routines for extracting facts from text. We describe two learning methods that we have applied to this task--a statistical text classification method, and a relational learning method--and our initial experiments in learning such information-extraction routines. We also present an approach to decreasing the cost of learning information-extraction routines by learning from "weakly" labeled training data.

Artificial Intelligence↗
PubMed1984

Relationship between physical and conceptual color codes.

The experiment was designed to determine whether the efficacy of the conceptual color code increases as the availability of the physical color code decreases. On each trial subjects judged whether two color chips were physically identical. On primed trials the prototype color chip was presented for 2 s prior to the test pair. After exposure to test pairs involving moderate goodness levels, all subjects responded to test pairs consisting of either good or poor members of a category. For some subjects the test pairs were presented immediately after the termination of the prototype chip. For other subjects there was either a 1- or 2-s delay between viewing the prototype chip and the presentation of the test pair. Under delay conditions of both 0 s and 1 s, priming with the prototype chip facilitated responses to good members of the category and inhibited responses to poor members of the category. However, after a 2-s delay, priming facilitated responses to both good and poor category members. It was argued that under the 0- and 1-s conditions priming effects were mediated by the physical color code. When the availability of the physical code was reduced by the 2-s delay, priming effects were determined by the modified conceptual representation.

Color Perception↗
PubMed1980

Constructing cognitive maps from sentences.

Three experiments investigated the construction of one- and two-dimensional cognitive maps from presented sentences containing information about adjacent pairs of items in the map. Sentences were of the form The item is X of the item or The item has the item on its X side when items were high-frequency nouns and X was the adjective north, south, east, or west. Hypotheses derived from three-term series and linear ordering work were tested. Increasing the number of dimensions and end anchors did not affect performance, although increasing the number of items lowered performance. Construction was best when the ordering was complete rather than partial, when each incoming sentence presented new information that could be immediately integrated with previously presented information, when new items could be added to the end rather than the beginning of a stored representation, and when the sentence adjective was congruent with the placement of the new item in the map. This last result appears to depend to some extent on the representational strategy adopted by participants.

Cognition↗
PubMed1980

The barren prospect of a representation world.

This is a criticism of the view that the term "representation" has a peculiarly tangible clinical referent, a unique phenomeriological meaning, and a special power to reveal subjective reality undistorted by theoretical constructs. Such a view is illegitimately encouraged by a tendency to think of "image" when we say "representation", even though we mean something more like "understanding", and by a tendency to treat understanding as though they were discrete experiences of the subject, rather than abstractions made by the observer. Because the reliance on representational terms mixed method with data, its covert principles of abstraction cannot be used to learn from new data.

Freudian Theory↗
PubMed1991

Response specificity in animal timing.

The stimuli that control responding in the peak procedure were investigated by training rats, in separate sessions, to make two different responses for food reinforcement. During one type of session, lever pressing was normally reinforced 32 s after the onset of a light. During the other type of session, chain pulling was normally reinforced either 8 s after the onset of one auditory cue or 128 s after the onset of a different auditory cue. For both types of sessions, only the appropriate manipulandum was available, and 20% of the trials lasted 240 s and involved no response-contingent consequences. Rats were then tested with the auditory cues in the presence of the lever and the light in the presence of the chain. If the time of reinforcement associated with each stimulus was learned, response rates should peak at these times during transfer testing. However, if a specific response pattern was learned for each stimulus, little transfer should occur. The results did not clearly support either prediction, leading to the conclusion that both a representation of the time of reinforcement and the rat's own behavior may control responding in this situation.

Animals↗
PubMed1998

Learning attractors in an asynchronous, stochastic electronic neural network.

LANN27 is an electronic device implementing in discrete electronics a fully connected (full feedback) network of 27 neurons and 351 plastic synapses with stochastic Hebbian learning. Both neurons and synapses are dynamic elements, with two time constants--fast for neurons and slow for synapses. Learning, synaptic dynamics, is analogue and is driven in a Hebbian way by neural activities. Long-term memorization takes place on a discrete set of synaptic efficacies and is effected in a stochastic manner. The intense feedback between the nonlinear neural elements, via the learned synaptic structure, creates in an organic way a set of attractors for the collective retrieval dynamics of the neural system, akin to Hebbian learned reverberations. The resulting structure of the attractors is a record of the large-scale statistics in the uncontrolled, incoming flow of stimuli. As the statistics in the stimulus flow changes significantly, the attractors slowly follow it and the network behaves as a palimpsest--old is gradually replaced by new. Moreover, the slow learning creates attractors which render the network a prototype extractor: entire clouds of stimuli, noisy versions of a prototype, used in training, all retrieve the attractor corresponding to the prototype upon retrieval. Here we describe the process of studying the collective dynamics of the network, before, during and following learning, which is rendered complex by the richness of the possible stimulus streams and the large dimensionality of the space of states of the network. We propose sampling techniques and modes of representation for the outcome.

Animals↗
PubMed2002

Experience modifies olfactory acuity: acetylcholine-dependent learning decreases behavioral generalization between similar odorants.

Perceptual learning has been demonstrated in several thalamocortical sensory systems wherein experience enhances sensory acuity for trained stimuli. This perceptual learning is believed to be dependent on changes in sensory cortical receptive fields. Sensory experience and learning also modifies receptive fields and neural response patterns in the mammalian olfactory system; however, to date there has been little reported evidence of learned changes in behavioral olfactory acuity. The present report used a bradycardial orienting response and cross-habituation paradigm that allowed assessment of behavioral discrimination of nearly novel odorants, and then used the same paradigm to examine odorant discrimination after associative olfactory conditioning with similar or dissimilar odorants. The results demonstrate that associative conditioning can enhance olfactory acuity for odors that are the same as or similar to the learned odorant, but not for odors dissimilar to the learned odorant. Furthermore, scopolamine injected before associative conditioning can block the acquisition of this learned enhancement in olfactory acuity. These results could have important implications for mechanisms of olfactory perception and memory, as well as for correlating behavioral olfactory acuity with observed spatial representations of odorant features in the olfactory system.

Acetylcholine↗
PubMed1995

[Deficiency in the reproduction and learning proper names after left tubero-thalamic ischemic lesion].

Predominant impairment or preservation of category-specific naming and comprehension is not rare in aphasics. Much less frequent is a selective inability to generate proper names. To our knowledge, only one such case has been reported after a left thalamic lesion, located in the ventral anterior nucleus, the mamillo-thalamic tractus and the genu of the internal capsule. We report a new case of selective inability to generate proper names after a left tubero-thalamic infarct. A 65-year old right-handed man presented with a selective impairment in producing proper names, both from photographs or descriptions and on tests of verbal fluency. The deficit was obvious both for persons names and for geographical names. The rest of the neuropsychological testing was remarkable only for a mild verbal amnesia, affecting only serial material (list of words), a reduced fluency for flowers, fruits and musical instruments, difficulties in learning of new words, and a dissociation between preserved learning for words on a test of learning of words and occupations (Cohen, 1990). This anomia for proper names could result from an indirect frontal-lobe dysfunction, preventing voluntary activation of the phonological representation of proper names.

Aged↗
PubMedDate not supplied

Strategies of shape representation in macaque visual area V2.

Contours and surface textures provide powerful cues used in image segmentation and the analysis of object shape. To learn more about how the visual system extracts and represents these visual cues, we studied the responses of V2 neurons in awake, fixating monkeys to complex contour stimuli (angles, intersections, arcs, and circles) and texture patterns such as non-Cartesian gratings, along with conventional bars and sinusoidal gratings. Substantial proportions of V2 cells conveyed information about many contour and texture characteristics associated with our stimuli, including shape, size, orientation, and spatial frequency. However, the cells differed considerably in terms of their degree of selectivity for the various stimulus characteristics. On average, V2 cells responded better to grating stimuli but were more selective for contour stimuli. Metric multidimensional scaling and principal components analysis showed that, as a population, V2 cells show strong correlations in how they respond to different stimulus types. The first two and five principal components accounted for 69% and 85% of the overall response variation, respectively, suggesting that the response correlations simplified the population representation of shape information with relatively little loss of information. Moreover, smaller random subsets of the population carried response correlation patterns very similar to the population as a whole, indicating that the response correlations were a widespread property of V2 cells. Thus, V2 cells extract information about a number of higher order shape cues related to contours and surface textures and about similarities among many of these shape cues. This may reflect an efficient strategy of representing cues for image segmentation and object shape using finite neuronal resources.

Animals↗
PubMed2006

Sparse representation for coarse and fine object recognition.

This paper offers a sparse, multiscale representation of objects. It captures the object appearance by selection from a very large dictionary of Gaussian differential basis functions. The learning procedure results from the matching pursuit algorithm, while the recognition is based on polynomial approximation to the bases, turning image matching into a problem of polynomial evaluation. The method is suited for coarse recognition between objects and, by adding more bases, also for fine recognition of the object pose. The advantages over the common representation using PCA include storing sampled points for recognition is not required, adding new objects to an existing data set is trivial because retraining other object models is not needed, and significantly in the important case where one has to scan an image over multiple locations in search for an object, the new representation is readily available as opposed to PCA projection at each location. The experimental result on the COIL-100 data set demonstrates high recognition accuracy with real-time performance.

Algorithms↗
PubMed1986

Prefrontal unit activity during delayed conditional Go/No-Go discrimination in the monkey. I. Relation to the stimulus.

Three monkeys learned a delayed conditional Go/No-go discrimination task. The sequential events of the task were the instructional color stimulus presentation (S1), the first delay (D1), discriminative pattern stimulus presentation (S2), the second delay (D2), and the response (Go or No-go). Single units in the principal sulcus, arcuate area and the inferior convexity of the prefrontal cortex were studied during this task. Of 512 task-related units recorded, 29% had stimulus-related differential activity. Type I units (n = 46) responded differentially to the two S1 stimuli. Type II units (n = 7) responded differentially to the two S2 patterns. Type III units (n = 65) also responded following the S2 presentation, but their activity depended upon whether Go or No-go responses were being signaled, rather than the particular S2 pattern. Type IV units (n = 54) appeared to code both the stimulus properties and behavioral significance of S2. A few premotor units (Area 6) were also examined; they showed either type III (n = 12) or type IV (n = 2) activity. In summary, these results indicate that primate prefrontal cortex participates in visual information processing and may code several aspects of visual stimuli including behavioral significance and mnemonic representations.

Animals↗
PubMed1989

A distributed, developmental model of word recognition and naming.

A parallel distributed processing model of visual word recognition and pronunciation is described. The model consists of sets of orthographic and phonological units and an interlevel of hidden units. Weights on connections between units were modified during a training phase using the back-propagation learning algorithm. The model simulates many aspects of human performance, including (a) differences between words in terms of processing difficulty, (b) pronunciation of novel items, (c) differences between readers in terms of word recognition skill, (d) transitions from beginning to skilled reading, and (e) differences in performance on lexical decision and naming tasks. The model's behavior early in the learning phase corresponds to that of children acquiring word recognition skills. Training with a smaller number of hidden units produces output characteristic of many dyslexic readers. Naming is simulated without pronunciation rules, and lexical decisions are simulated without accessing word-level representations. The performance of the model is largely determined by three factors: the nature of the input, a significant fragment of written English; the learning rule, which encodes the implicit structure of the orthography in the weights on connections; and the architecture of the system, which influences the scope of what can be learned.

Dyslexia↗
PubMed1996

Learning theory and intrapsychic conflict.

This paper is one in a series (Gillett, 1990, 1994) attempting to explore the implications of modern ideas about learning for psychoanalytic theories of treatment and pathogenesis. The key concept is that of learned expectations, which establishes links with Freud's 1926 theory of neurotic anxiety as caused by the expectation of danger. The new understanding of classical Pavlovian conditioning entails changes in the basic theory of intrapsychic conflict described in previous papers (1990, 1994). The relationship of learning theory to Freud's 1926 theory of intrapsychic conflict has received insufficient attention in the psychoanalytic literature because of insufficient familiarity with the repudiation of behaviourism by psychologists in favour of a representational theory of the mind.

Behaviorism↗
PubMed1998

A second-order learning algorithm for multilayer networks based on block Hessian matrix.

This article proposes a new second-order learning algorithm for training the multilayer perceptron (MLP) networks. The proposed algorithm is a revised Newton's method. A forward-backward propagation scheme is first proposed for network computation of the Hessian matrix, H, of the output error function of the MLP. A block Hessian matrix, H(b), is then defined to approximate and simplify H. Several lemmas and theorems are proved to uncover the important properties of H and H(b), and verify the good approximation of H(b) to H; H(b) preserves the major properties of H. The theoretic analysis leads to the development of an efficient way for computing the inverse of H(b) recursively. In the proposed second-order learning algorithm, the least squares estimation technique is adopted to further lessen the local minimum problems. The proposed algorithm overcomes not only the drawbacks of the standard backpropagation algorithm (i.e. slow asymptotic convergence rate, bad controllability of convergence accuracy, local minimum problems, and high sensitivity to learning constant), but also the shortcomings of normal Newton's method used on the MLP, such as the lack of network implementation of H, ill representability of the diagonal terms of H, the heavy computation load of the inverse of H, and the requirement of a good initial estimate of the solution (weights). Several example problems are used to demonstrate the efficiency of the proposed learning algorithm. Extensive performance (convergence rate and accuracy) comparisons of the proposed algorithm with other learning schemes (including the standard backpropagation algorithm) are also made.

Journal Article↗
PubMed1989

Glycoprotein Synthesis Is Necessary for Memory of Sickness-Induced Learning in Chicks.

Can current biochemical models of memory account for sickness-induced learning? We show that chicks can form an association between pecking a coloured but tasteless lure and becoming ill (LiCl, i.p.) 30 min later. We go on to demonstrate amnesia for this association, induced by intracranial administration of 2-deoxygalactose (10 micromole per hemisphere, in a 10 microl vol), an inhibitor of the synthesis of glycoproteins of the synaptic membrane, 10 min before pecking. Further, we show that this 2-deoxygalactose-induced amnesia is not state dependent. Thus the brain representation of the lure must be held, and require macromolecular syntheses, similar to those found in other forms of learning, for a considerable time before it can be associated with new significant experience. This is incompatible with contiguous synaptic firing views of memory.

Journal Article↗
PubMed1996

Motor imagery--anatomical representation and electrophysiological characteristics.

Experience and results of neuropsychological studies have shown that motor imagery can improve motor performance and enhance motor learning. In recent years several electro-physiological and functional imaging studies have investigated the physiological basis for this observation. In the present essay we review two of our recent studies, in which we compared motor imagery with motor preparation and motor execution. In the first we used positron emission tomography to describe their functional anatomy and in the second we employed electromyography, H-reflexes and transcranial magnetic stimulation to delineate their electrophysiological characteristics. Both studies demonstrated that motor imagery shares some characteristics with motor preparation and other, additional ones with motor execution. Thus it can be seen as a special form of motor behaviour, similar but distinct from both motor preparation and execution. This combination of mutual and distinct characteristics may be the key to its successful role in motor learning.

Adult↗