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 577 records · Page 32Linked to original sources
PubMed1994

A shape-based machine learning tool for drug design.

Building predictive models for iterative drug design in the absence of a known target protein structure is an important challenge. We present a novel technique, Compass, that removes a major obstacle to accurate prediction by automatically selecting conformations and alignments of molecules without the benefit of a characterized active site. The technique combines explicit representation of molecular shape with neural network learning methods to produce highly predictive models, even across chemically distinct classes of molecules. We apply the method to predicting human perception of musk odor and show how the resulting models can provide graphical guidance for chemical modifications.

Algorithms↗
PubMed1992

Classification of crossed immunoelectrophoretic patterns using digital image processing and artificial neural networks.

A method is presented which makes it possible to present crossed immunoelectrophoretic patterns to an artificial neural network. The electrophoretic patterns are presented for the artificial neural network as three-dimensional vectors and it is shown that it is possible with this representation to train the network to learn the patterns and classify them. It was found that the ability to generalize was substantially increased by the addition of noise to the input patterns during training. Furthermore, the addition of noise decreased the number of presentations needed to reach the predetermined error level. The trained neural network was able to classify all distorted patterns correctly within an error range of 1%.

Image Processing, Computer-Assisted↗
PubMed2005

What went where? Impaired object-location learning in patients with right hippocampal lesions.

The role of the right medial temporal-lobe structures in memory for object location was investigated in three studies. In the first two studies, 118 patients with selective amygdalo-hippocampectomy or with anterior temporal lobectomy (either invading or largely sparing the hippocampal region) and 33 healthy participants were tested on array learning. Groups with extensive right hippocampal lesions were impaired on immediate and delayed recall and on learning to criterion. In the third study, magnetic resonance imaging (MRI) was used in 75 of these patients to measure the extent of tissue remaining in the various medial temporal-lobe structures. The extent of right hippocampus remaining was found to be the best predictor of array-learning performance, underlining its critical role in building a representation of objects in space.

Adult↗
PubMed2001

The development of ordinal numerical competence in young children.

Two experiments assessed ordinal numerical knowledge in 2- and 3-year-old children and investigated the relationship between ordinal and verbal numerical knowledge. Children were trained on a 1 vs 2 comparison and then tested with novel numerosities. Stimuli consisted of two trays, each containing a different number of boxes. In Experiment 1, box size was held constant. In Experiment 2, box size was varied such that cumulative surface area was unrelated to number. Results show children as young as 2 years of age make purely numerical discriminations and represent ordinal relations between numerosities as large as 6. Children who lacked any verbal numerical knowledge could not make ordinal judgments. However, once children possessed minimal verbal numerical competence, further knowledge was entirely unrelated to ordinal competence. Number may become a salient dimension as children begin to learn to count. An analog magnitude representation of number may underlie success on the ordinal task.

Attention↗
PubMed2004

The effects of practice and delay on motor skill learning and retention.

The present study assessed the effects of amount of practice and length of delay on the learning and retention of a timed motor sequence task. Participants learned to reproduce ten-element visual sequences by tapping in synchrony with the stimulus. Participants were randomly assigned to a varied-practice condition or a varied-delay condition. In the varied-practice condition, participants received either one, three, or six blocks of practice followed by a fixed 4-week delayed-recall. In the varied-delay condition, participants received three blocks of practice followed by a varied delay of either 3 days, or 2, 4, or 8 weeks. Learning was assessed by changes in accuracy, response variance, and percent response asynchrony. Our results showed that amount of practice per se did not affect learning and retention of the task. Rather, distribution of practice over several days was the most important factor affecting learning and retention. We hypothesize that passage of time is essential for a maximum benefit of practice to be gained, as the time delay may allow for consolidation of learning, possibly reflecting plastic changes in motor cortical representations of the skill. With regards to delay, our findings suggest that explicit and motoric components of a motor sequence are likely to be learned and maintained in separate but interacting systems. First, only the longest delay group showed decrements in percent correct, indicating that longer lengths of delay might hinder retrieval of explicit aspects of the task. Second, all groups showed a decrement in percent response asynchrony, suggesting that synchronization may be a more difficult parameter to maintain because it relies heavily on sensorimotor integration.

Adolescent↗
PubMed2000

Intact negative patterning in rats with fornix or combined perirhinal and postrhinal cortex lesions.

It has been proposed that the hippocampal formation is necessary for the acquisition of tasks that require the use of configural representations for their solution, including spatial learning and negative patterning. Tests of this influential view have, however, yielded conflicting results. For example fornix or hippocampal lesions, which reliably impair spatial learning, do not reliably impair negative patterning. A problem in interpreting these results has been the lack of controls for factors such as over-responding, excitatory effects of reward, and the possibility of non-configural solutions. At the same time, other studies have pointed to a role in configural learning for parahippocampal regions such as the perirhinal cortex. The present experiments controlled for the above factors and revealed that neither lesions of the fornix nor of the perirhinal/postrhinal cortex in the rat had any effect on negative patterning, although subsequent tests of object and spatial memory demonstrated the functional efficacy of the lesions.

Amygdala↗
PubMed2004

A design and application of a multi-agent system for simulation of multi-actor spatial planning.

Multi-agent Systems (MAS) offer a conceptual approach to include multi-actor decision making into models of land use change. The main goal is to explore the use of MAS to simulate spatial scenarios based on modelling multi-actor decision-making within a spatial planning process. We demonstrate MAS that consists of agents representing organizations and interest groups involved in an urban allocation problem during a land use planning process. The multi-actor based decision-making is modelled by generating beliefs and preferences of actors about the location of and relation between spatial objects. This allows each agent to confront these beliefs and preferences with it's own desires and with that of other agents. The MAS loosely resembles belief, desire and intentions architecture. Based on a case study for a hypothetical land use planning situation in a study area in the Netherlands we discuss the potential and limitations of the MAS to build models that enable spatial planners to include the 'actor factor' in their analysis and design of spatial scenarios. In addition, our experiments revealed the need for further research on the representation of spatial objects and reasoning, learning and communication about allocation problems using MAS.

Communication↗
PubMed2006

Brain-behavior relations across the lifespan: a commentary.

This final article provides an overview and commentary on some of the methods, findings and interpretations discussed in the preceding papers. Some major commonalities among them are highlighted and possible questions for future research projects are suggested. Topics discussed include plasticity, compensation and dedifferentiation, representation and control, memory and learning-all considered under the general heading of brain-behavior relations, and how they change across the lifespan.

Age Factors↗
PubMed2004

ARTSTREAM: a neural network model of auditory scene analysis and source segregation.

Multiple sound sources often contain harmonics that overlap and may be degraded by environmental noise. The auditory system is capable of teasing apart these sources into distinct mental objects, or streams. Such an 'auditory scene analysis' enables the brain to solve the cocktail party problem. A neural network model of auditory scene analysis, called the ARTSTREAM model, is presented to propose how the brain accomplishes this feat. The model clarifies how the frequency components that correspond to a given acoustic source may be coherently grouped together into a distinct stream based on pitch and spatial location cues. The model also clarifies how multiple streams may be distinguished and separated by the brain. Streams are formed as spectral-pitch resonances that emerge through feedback interactions between frequency-specific spectral representations of a sound source and its pitch. First, the model transforms a sound into a spatial pattern of frequency-specific activation across a spectral stream layer. The sound has multiple parallel representations at this layer. A sound's spectral representation activates a bottom-up filter that is sensitive to the harmonics of the sound's pitch. This filter activates a pitch category which, in turn, activates a top-down expectation that is also sensitive to the harmonics of the pitch. Resonance develops when the spectral and pitch representations mutually reinforce one another. Resonance provides the coherence that allows one voice or instrument to be tracked through a noisy multiple source environment. Spectral components are suppressed if they do not match harmonics of the top-down expectation that is read-out by the selected pitch, thereby allowing another stream to capture these components, as in the 'old-plus-new heuristic' of Bregman. Multiple simultaneously occurring spectral-pitch resonances can hereby emerge. These resonance and matching mechanisms are specialized versions of Adaptive Resonance Theory, or ART, which clarifies how pitch representations can self-organize during learning of harmonic bottom-up filters and top-down expectations. The model also clarifies how spatial location cues can help to disambiguate two sources with similar spectral cues. Data are simulated from psychophysical grouping experiments, such as how a tone sweeping upwards in frequency creates a bounce percept by grouping with a downward sweeping tone due to proximity in frequency, even if noise replaces the tones at their intersection point. Illusory auditory percepts are also simulated, such as the auditory continuity illusion of a tone continuing through a noise burst even if the tone is not present during the noise, and the scale illusion of Deutsch whereby downward and upward scales presented alternately to the two ears are regrouped based on frequency proximity, leading to a bounce percept. Since related sorts of resonances have been used to quantitatively simulate psychophysical data about speech perception, the model strengthens the hypothesis that ART-like mechanisms are used at multiple levels of the auditory system. Proposals for developing the model to explain more complex streaming data are also provided.

Acoustic Stimulation↗
PubMed1987

Representation of semantic and spatial knowledge in Alzheimer's patients: implications for models of preserved learning in amnesia.

Patients with Alzheimer's disease may exhibit severe, but variable patterns of cognitive deficits which can complicate and confound investigations of memory. Three patients are described with comparable levels of intellectual and memory abilities as assessed by the Wechsler scales. One patient (W) had a relatively circumscribed impairment of word-finding ability concurrent with intact visuospatial and constructional skill. Another patient (C) showed the opposite profile of abilities, while the third patient (G) exhibited deficits in both domains. All three patients were severely and globally amnesic when tested with traditional recall and recognition procedures. Patient W was able to demonstrate preserved semantic knowledge on a superordinate and category level but not for specific object attributes. Patient C demonstrated a remarkably preserved ability to generate and copy complex, meaningful, but not meaningless material. Corresponding profiles of preserved learning were revealed, but only under conditions that limited the demands on encoding and retrieval processes. Models were offered to account for these contrasting patterns of impairment. In addition, the possibility that medial temporal structures contribute to the ability to consciously reconstruct prior experiences was discussed.

Alzheimer Disease↗
PubMed2005

Spatial-temporal modeling of malware propagation in networks.

Network security is an important task of network management. One threat to network security is malware (malicious software) propagation. One type of malware is called topological scanning that spreads based on topology information. The focus of this work is on modeling the spread of topological malwares, which is important for understanding their potential damages, and for developing countermeasures to protect the network infrastructure. Our model is motivated by probabilistic graphs, which have been widely investigated in machine learning. We first use a graphical representation to abstract the propagation of malwares that employ different scanning methods. We then use a spatial-temporal random process to describe the statistical dependence of malware propagation in arbitrary topologies. As the spatial dependence is particularly difficult to characterize, the problem becomes how to use simple (i.e., biased) models to approximate the spatially dependent process. In particular, we propose the independent model and the Markov model as simple approximations. We conduct both theoretical analysis and extensive simulations on large networks using both real measurements and synthesized topologies to test the performance of the proposed models. Our results show that the independent model can capture temporal dependence and detailed topology information and, thus, outperforms the previous models, whereas the Markov model incorporates a certain spatial dependence and, thus, achieves a greater accuracy in characterizing both transient and equilibrium behaviors of malware propagation.

Algorithms↗
PubMed2001

Functional magnetic resonance imaging of working memory impairment after traumatic brain injury.

OBJECTIVES: To examine patterns of brain activation while performing a working memory task in persons with moderate to severe traumatic brain injury (TBI) and healthy controls. It is well established that working memory is an area of cognition that is especially vulnerable to disruption after TBI. Although much has been learned about the system of cerebral representation of working memory in healthy people, little is known about how this system is disrupted by TBI. METHODS: Functional magnetic resonance imaging (fMRI) was used to assess brain activation during a working memory task (a modified version of the paced auditory serial addition test) in nine patients with TBI and seven healthy controls. RESULTS: Patients with TBI were able to perform the task, but made significantly more errors than healthy controls. Cerebral activation in both groups was found in similar regions of the frontal, parietal, and temporal lobes, and resembled patterns of activation found in previous neuroimaging studies of working memory in healthy persons. However, compared with the healthy controls, the TBI group displayed a pattern of cerebral activation that was more regionally dispersed and more lateralised to the right hemisphere. Differences in lateralisation were particularly evident in the frontal lobes. CONCLUSIONS: Impairment of working memory in TBI seems to be associated with alterations in functional cerebral activity.

Adult↗
PubMed2002

The cerebellum: it's about time! But timing is not everything--new insights into the role of the cerebellum in timing motor and cognitive tasks.

Converging evidence from different research studies supports a role for the cerebellum in timing neural processes. The cerebellum is part of a distributed system for motor control. The timing hypothesis provides a specific functional role for the unique contribution of the cerebellum. The timing capabilities of the cerebellum appear to extend beyond motor control into tasks focusing on perceptual processing that require the precise representation of temporal information and sensorimotor learning. Behavioral and modeling studies suggest that the cerebellar timing system is best characterized as providing a near-infinite set of interval-type timers rather than as a single clock with pacemaker or oscillatory properties, but this is controversial. In addition to learning precisely timed motor responses, the cerebellum is involved in on-line processing using feed-forward systems for which sensory input is used prior to movement execution to improve movement accuracy. This would be a mechanism for triggering accurate "time." The cerebellum continues to fascinate scientists, and although survival is possible without the cerebellum, the resultant quality of life is significantly compromised with clumsiness, ataxia, hypotonia, dysarthria, slowing of various cognitive perceptual processes, and impaired fine motor and ocular-motor coordination. The last three decades have seen the development of research that has focused on how the cerebellum functions. Further neurophysiologic research in cerebellar cortical neurotransmission is likely to further our understanding of the cerebellar contribution to timing sensorimotor processes.

Animals↗
PubMed2006

Strength of British English accents in altered listening conditions.

This work is concerned with the processing or representational level at which accent forms learned early in life can change and with whether alteration to a speaker's auditory environment can elicit an original accent. In Experiment 1, recordings were made of an equal number of (1) speakers living in the home counties (HC) of Britain (around the London conurbation) who claimed to have retained the accent of the region that they originally had come from, (2) speakers who stated that they had lost their regional accent and acquired an HC accent, and (3) native HC speakers. They read two texts in a normal listening environment. Listeners rated the similarity in accent between each of these texts and all the other texts. The results showed that in the normal listening conditions, the speakers who had lost their accent were rated as being more similar to HC English speakers than to those speakers from the same region who had retained their accent. In Experiment 2, recordings of the same speakers under frequency-shifted and delayed auditory feedback, as well as the normal listening conditions used earlier, were rated in order to see whether the manipulations of listening environment would elicit the speaker's original accent. Listeners rated similarity of accent in a sample of speech recorded under normal listening against a sample read by another speaker in one of the altered listening conditions. When listening condition was altered, the speakers who had lost their original accent were rated as more similar to those who had retained their accent. It is concluded that accent differences can be elicited by altering listening environment because the speech systems of speakers who have lost their accent are more vulnerable than those of speakers who have not changed their original accent.

Adult↗
PubMed1999

Dealing with the fear of mutilation in the donation discussion.

Fear of mutilation is a significant barrier to organ and tissue donation. It constitutes an example of Mystical Thinking and may be seen as an exemplar of animal learning or, more specifically, as a representation of the "blood phobia." As such the fear is not amenable to conventional public education efforts. Cognitive and behavioral techniques used in treating other types of phobias should be studied as a way to remove this barrier to donation.

Animals↗
PubMed1994

ERISA estoppel: say what you mean and mean what you say.

Estoppel is rapidly becoming part of ERISA litigation. However, as employers, TPAs and other entities associated with ERISA plans learn to take precautions against erroneous representations, estoppel litigation may subside.

Health Benefit Plans, Employee↗
PubMed1990

Rule abstraction, item memory, and chunking in rat serial-pattern tracking.

Two studies used a stimulus tracking paradigm to test whether rats are sensitive to the rule-based formal properties of structured serial patterns. Hooded rats tracked 16-element patterns of flashing lights by pressing levers under an array of 6 indicator lights. In Experiment 1, rats tracked a pattern similar to one previously used with human subjects and yielded remarkably similar results. More errors and response omissions occurred at boundaries of structural "chunks" than within chunks, and errors often reflected anticipation of the next chunk or extrapolation of the preceding chunk. In Experiment 2, temporal "phrasing" cues encouraged different groups to encode a pattern as a series of either "runs" or "trills." Differential placement of pauses induced rats to encode different rule-based representations of the pattern. Results indicate that under appropriate conditions rats may encode a representation of formal structure when they learn organized response patterns.

Animals↗