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 379 records · Page 21Linked to original sources
PubMed2026

Scalable, generalizable and uncertainty-aware integration of spatial multiomics across diverse modalities and platforms with SCIGMA.

Recent advances in spatial omics technologies have enabled simultaneous profiling of transcriptomic, proteomic, epigenomic, metabolomic and imaging data at high spatial resolution, offering unprecedented opportunities to dissect tissue complexity. However, integrating these diverse and large-scale spatial multimodal datasets remains a major computational challenge. We present SCIGMA, a scalable and generalizable deep learning framework for spatial multiomics integration. SCIGMA introduces an uncertainty-aware contrastive learning objective and multiview graph neural networks to preserve modality-specific signals while learning biologically meaningful joint representations. Unlike previous methods, SCIGMA provides spatially resolved uncertainty estimates, interpretably identifying regions of biological or technical heterogeneity. SCIGMA supports integration of up to five modalities, and its modular framework is extensible to future technologies with even more modalities. It also scales to more than 1 million spatial locations, enabling analysis of high-resolution datasets such as Visium HD and Xenium Prime. We evaluated SCIGMA across 19 datasets spanning 8 modalities, 10 tissues and 9 platforms. On benchmarkable datasets, SCIGMA outperformed other methods in spatial domain detection, modality preservation, feature reconstruction and reproducibility. SCIGMA identifies biologically meaningful structures, refined spatial domains and modality-specific regulatory programs, providing a robust, flexible and future-ready solution for scalable spatial multimodal integration.

Multiomics↗
PubMed2003

Role of sensory information in updating internal models of the effector during arm tracking.

This chapter is divided into three main parts. Firstly, on the basis of the literature, we will shortly discuss how the recent introduction of the concept of internal models by Daniel Wolpert and Mitsuo Kawato contributes to a better understanding of what is motor learning and what is motor adaptation. Then, we will present a model of eye-hand co-ordination during self-moved target tracking, which we used as a way to specifically address these topics. Finally, we will show some evidence about the use of proprioceptive information for updating the internal models, in the context of eye-hand co-ordination. Motor and afferent information appears to contribute to the parametric adjustment (adaptation) between arm motor command and visual information about arm motion. The study reported here was aimed at assessing the contribution of arm proprioception in building (learning) and updating (adaptation) these representations. The subjects (including a deafferented subject) had to make back and forth movements with their forearm in the horizontal plane, over learned amplitude and at constant frequency, and to track an arm-driven target with their eyes. The dynamical conditions of arm movement were altered (unexpectedly or systematically) during the movement by changing the mechanical properties of the manipulandum. The results showed a significant change of the latency and the gain of the smooth pursuit system, before and after the perturbation for the control subjects, but not for the deafferented subject. Moreover, in control subjects, vibrations of the arm muscles prevented adaptation to the mechanical perturbation. These results suggest that in a self-moved target tracking task, the arm motor system shares with the smooth pursuit system an internal representation of the arm dynamical properties, and that arm proprioception is necessary to build this internal model. As suggested by Ghez et al. (1990) (Cold Spring Harbor Symp. Quant. Biol., 55: 837-8471), proprioception would allow control subjects to learn the inertial properties of the limb.

Adaptation, Physiological↗
PubMed2002

[The learning curve in surgery: possibilities and limits of this method].

The learning curve is a graphic representation of the relationship between the experience of a surgeon and one or more performance indicators. The operation time alone is an insufficient indicator to assess the performance of a surgeon. The procedure time has to be set in relation to the complication rate and in laparoscopic surgery to the conversion rate. Techniques to visualize the changes over time are the moving average method for the operation time and the Cusum method for dichotomous outcomes, like the conversion and the complication rates. At the time when the learning curve reaches the plateau phase this representation can be used to assess the quality of a surgeon or a team in one hospital. As far as there is no validated complexity scale for laparoscopic procedures available it is difficult to compare patient populations between different hospitals. Out of this reason the learning curve is no legitimate instrument to rank surgeons or different hospitals.

Clinical Competence↗
PubMed2002

Learning to recognize three-dimensional objects.

A learning account for the problem of object recognition is developed within the probably approximately correct (PAC) model of learnability. The key assumption underlying this work is that objects can be recognized (or discriminated) using simple representations in terms of syntactically simple relations over the raw image. Although the potential number of these simple relations could be huge, only a few of them are actually present in each observed image, and a fairly small number of those observed are relevant to discriminating an object. We show that these properties can be exploited to yield an efficient learning approach in terms of sample and computational complexity within the PAC model. No assumptions are needed on the distribution of the observed objects, and the learning performance is quantified relative to its experience. Most important, the success of learning an object representation is naturally tied to the ability to represent it as a function of some intermediate representations extracted from the image. We evaluate this approach in a large-scale experimental study in which the SNoW learning architecture is used to learn representations for the 100 objects in the Columbia Object Image Library. Experimental results exhibit good generalization and robustness properties of the SNoW-based method relative to other approaches. SNoW's recognition rate degrades more gracefully when the training data contains fewer views, and it shows similar behavior in some preliminary experiments with partially occluded objects.

Artificial Intelligence↗
PubMed2004

Ventral pallidal representation of pavlovian cues and reward: population and rate codes.

We recorded neural activity in the ventral pallidum (VP) while rats learned a pavlovian reward association. Rats learned to distinguish a tone that predicted sucrose pellets (CS+) from a different tone that predicted nothing (CS-). Many VP units became responsive to CS+, but few units responded to CS-. When two CS+ were encountered sequentially, the earliest predictor of reward became most potent. Many VP units were also activated when the sucrose reward was received [unconditioned stimulus (UCS)]. These VP units for UCS remained responsive to sucrose reward after learning, even when sucrose was already predicted by CS+. Neural representation of reward learning and reward itself was characterized by population codes. The population of units that responded to CS+ increased with learning, whereas the population that responded to UCS did not change. A relative firing rate code also represented the identities of conditioned stimuli and UCS. Firing rate differences among stimuli were acquired early and remained stable during subsequent training, whereas population codes and behavioral conditioned responses continued to develop during subsequent training. Thus, the VP makes use of dynamic CS population and rate codes to encode pavlovian reward cues in reward learning and uses stable UCS population and firing codes to encode sucrose reward itself.

Acoustic Stimulation↗
PubMed1996

Hebbian learning of context in recurrent neural networks.

Single electrode recording in the inferotemporal cortex of monkeys during delayed visual memory tasks provide evidence for attractor dynamics in the observed region. The persistent elevated delay activities could be internal representations of features of the learned visual stimuli shown to the monkey during training. When uncorrelated stimuli are presented during training in a fixed sequence, these experiments display significant correlations between the internal representations. Recently a simple model of attractor neural network has reproduced quantitatively the measured correlations. An underlying assumption of the model is that the synaptic matrix formed during the training phase contains in its efficacies information about the contiguity of persistent stimuli in the training sequence. We present here a simple unsupervised learning dynamics that produces such a synaptic matrix if sequences of stimuli are repeatedly presented to the network at fixed order. The resulting matrix is then shown to convert temporal correlations during training into spatial correlations between attractors. The scenario is that, in the presence of selective delay activity, at the presentation of each stimulus, the activity distribution in the neural assembly contain information of both the current stimulus and the previous one (carried by the attractor). Thus the recurrent synaptic matrix can code not only for each of the stimuli presented to the network but also for their context. We combine the idea that for learning to be effective, synaptic modification should be stochastic, with the fact that attractors provide learnable information about two consecutive stimuli. We calculate explicitly the probability distribution of synaptic efficacies as a function of training protocol, that is, the order in which stimuli are presented to the network. We then solve for the dynamics of a network composed of integrate-and-fire excitatory and inhibitory neurons with a matrix of synaptic collaterals resulting from the learning dynamics. The network has a stable spontaneous activity, and stable delay activity develops after a critical learning stage. The availability of a learning dynamics makes possible a number of experimental predictions for the dependence of the delay activity distributions and the correlations between them, on the learning stage and the learning protocol. In particular it makes specific predictions for pair-associates delay experiments.

Animals↗
PubMed2002

A further investigation of category learning by inference.

Categories are learned in many ways besides by classification, for example, by making inferences about classified items. One hypothesis is that classifications lead to the learning of features that distinguish categories, whereas inferences promote the learning of the internal structure of categories, such as the typical features. Experiment 1 included single-feature and full-feature classification tests following either classification or inference learning. Consistent with predictions, inference learners did better on the single tests but worse on the full tests. Experiment 2 further showed that inference learners, unlike classification learners, were no better at classifying items that they had seen at study compared with equally typical items they had not seen at study. Experiment 3 showed that features queried about during inference learning were classified better than ones not queried about, although even the latter features showed some learning on single-feature tests. The discussion focuses on how different types of category learning lead to different category representations.

Adult↗
PubMed2000

Designing a computer-based simulator for interventional cardiology training.

Interventional cardiology training traditionally involves one-on-one experience following a master-apprentice model, much as other procedural disciplines. Development of a realistic computer-based training system that includes hand-eye coordination, catheter and guide wire choices, three-dimensional anatomic representations, and an integrated learning system is desirable, in order to permit learning to occur safely, without putting patients at risk. Here we present the first report of a PC-based simulator that incorporates synthetic fluoroscopy, real-time three-dimensional interactive anatomic display, and selective right- and left-sided coronary catheterization and angiography using actual catheters. Significant learning components also are integrated into the simulator.

Cardiac Catheterization↗
PubMed2002

Task switching: a PDP model.

When subjects switch between a pair of stimulus-response tasks, reaction time is slower on trial N if a different task was performed on trial N - 1. We present a parallel distributed processing (PDP) model that simulates this effect when subjects switch between word reading and color naming in response to Stroop stimuli. Reaction time on "switch trials" can be slowed by an extended response selection process which results from (a) persisting, inappropriate states of activation and inhibition of task-controlling representations; and (b) associative learning, which allows stimuli to evoke tasks sets with which they have recently been associated (as proposed by Allport & Wylie, 2000). The model provides a good fit to a large body of empirical data, including findings which have been seen as problematic for this explanation of switch costs, and shows similar behavior when the parameters are set to random values, supporting Allport and Wylie's proposal.

Humans↗
PubMed1990

Trajectory formation of arm movement by cascade neural network model based on minimum torque-change criterion.

We proposed that the trajectory followed by human subject arms tended to minimize the time integral of the square of the rate of change of torque (Uno et al. 1987). This minimum torque-change model predicted and reproduced human multi-joint movement data quite well (Uno et al. 1989). Here, we propose a neural network model for trajectory formation based on the minimum torque-change criterion. Basic ideas of information representation and algorithm are (i) spatial representation of time, (ii) learning of forward dynamics and kinetics model and (iii) relaxation computation based on the acquired model. The model can resolve ill-posed inverse kinematics and inverse dynamics problems for redundant controlled object as well as ill-posed trajectory formation problems. By computer simulation, we show that the model can produce a multi-joint arm trajectory while avoiding obstacles or passing through viapoints.

Algorithms↗
PubMed1990

Acquisition of spatial knowledge under conditions of temporospatial discontinuity in young and elderly adults.

Young and elderly adults acquired route information from a sequence of slides depicting a walk through an actual environment. The accuracy of their distance knowledge after viewing the slides was compared for a normal presentation and a presentation with temporospatial discontinuity. No differences between age groups were noted under normal presentation conditions, but young adults were more accurate under conditions of temporospatial discontinuity. Results were interpreted in terms of an age-related decrement in the operational capacity of working memory. They were also viewed as supportive of a constructivist-representational theory of spatial learning.

Adult↗
PubMed2001

Cerebellar lesions impair context-dependent adaptation of reaching movements in primates.

To produce accurate movements when conditions change suddenly, the brain must be capable of learning multiple versions of a given motor task and must be able to access the appropriate program using sensory information linked to the context of the movement. The neural basis for context-dependent motor learning is uncertain, but the cerebellum is thought to play a fundamental role. In this study, we examined the effect of lesions of the dorsal vermal and paravermal cerebellar cortex on the adaptation of reaching movements produced by modified visual feedback and accessed with a visual cue. Two rhesus monkeys were trained to point to targets displayed on a video monitor while viewing monocularly with either eye. During the experimental sessions, visual information received by one eye (the "modified" eye) was displaced horizontally, while the information received by the other ("normal") eye remained unaltered. In the first set of experiments (noncontextual paradigm), the animals pointed to targets while viewing with the modified eye. This paradigm resulted in a gradual improvement in pointing accuracy when viewing with that eye, but also produced a shift in pointing responses of equivalent size when viewing with the normal eye. In the second set of experiments (contextual paradigm), the animals alternated six blocks of reaches while viewing monocularly with the modified eye with six blocks viewing with the normal eye. This paradigm improved the pointing accuracy when viewing with the modified eye, but produced only a small shift in pointing responses when viewing with the normal eye. After the dorsal vermal and paravermal cerebellar cortex were resected, no change occurred in the pattern of adaptation produced by the noncontextual paradigm. The contextual paradigm, however, no longer selectively adapted pointing responses for each eye, but rather produced a pointing shift of equivalent size when viewing with either eye. The results indicate that pointing responses can be differentially adapted for each viewing eye, which is a form of context-dependent motor learning. This capability was lost after focal lesions of the dorsal vermal and paravermal cerebellar cortex, suggesting that these regions of cerebellar cortex are required to learn or store multiple representations of a movement, or to retrieve the appropriate motor program in a given sensory context.

Adaptation, Physiological↗
PubMed1989

First experiences with the modelling and simulation package MIRACLES applied to a picture archiving and communication system (PACS) in a clinical environment.

Since the construction of picture archiving and communication systems (PACS) appears to be extremely difficult, computer modelling and simulation are used as decision support tools. The package MIRACLES (Medical Image Representation, Archiving and Communication Learned from Extensive Simulation) has been developed at BAZIS in order to support the construction of simulation models of image information systems. This article discusses the application of MIRACLES to a prototypical PACS as being installed in a clinical environment. Attention is focussed to the required system analysis and difficulties which arose during the construction of the simulation model. The emphasis is on the presentation of the results of the simulation study, which show that simulation can be fruitfully used to predict, to analyse and to assist in solving performance problems. The simulation study confirmed assumptions and suppositions concerning both the system performance itself and strategies to improve the performance. The study also resulted in a number of concrete recommendations which might be useful for the set-up of the prototypical PACS.

Computer Simulation↗
PubMed2004

Cue valence representation studied by Fos immunocytochemistry after acquisition of a discrimination learning task.

The piriform cortex (PCx) and related structures such as hippocampus and frontal cortex could play an important role in olfactory memory. We investigated their involvement in learning the biological value of an odor cue, i.e. predicting reward or non-reward in a two-odor discrimination task. Rats were sacrificed after stimulation by either rewarded or non-rewarded odor and Fos immunocytochemistry was performed. The different experimental groups of rats did not show strongly differentiated Fos expression pattern in either the PCx or the hippocampus. A few differences were noted in frontal areas. In the ventro-lateral orbital cortex, rats, ramdomly rewarded during the conditionning had a higher Fos level in comparison with other groups. In infralimbic cortex, rats, which learned the reward value of the olfactory cue and were water-reinforced the day of sacrifice, showed a higher Fos expression. Data are discussed in view of the olfactory learning paradigm and of the accuracy of the control groups used in the present experimental design. The behavioural conditions leading to Fos expression are further discussed since Fos is a marker of learning-induced plasticity as well as a general activity marker which can be activated by a wide range of stimuli not directly linked to memory.

Animals↗
PubMed2000

Medical care-seeking for menstrual symptoms.

Fifty-six female undergraduates completed questionnaires regarding their menstrual symptoms, social learning experiences, and illness representations. The results showed that, compared to non-care-seekers, participants who had sought medical care for their menstrual symptoms reported more symptoms that had been problematic since menarche. Consistent with previous research, care-seekers reported more reinforcement for adolescent menstrual illness behaviours than non-care-seekers. Care-seekers also reported their symptoms as more serious and more difficult to ignore. The perceived seriousness and severity of symptoms were both correlated with reinforcement for adolescent menstrual symptoms. Lay referral was also a reported factor in care-seekers. The results of the present data are discussed with respect to previous research on care-seeking for menstrual and other symptoms.

Adaptation, Psychological↗
PubMed2001

S-TREE: self-organizing trees for data clustering and online vector quantization.

This paper introduces S-TREE (Self-Organizing Tree), a family of models that use unsupervised learning to construct hierarchical representations of data and online tree-structured vector quantizers. The S-TREE1 model, which features a new tree-building algorithm, can be implemented with various cost functions. An alternative implementation, S-TREE2, which uses a new double-path search procedure, is also developed. The performance of the S-TREE algorithms is illustrated with data clustering and vector quantization examples, including a Gauss-Markov source benchmark and an image compression application. S-TREE performance on these tasks is compared with the standard tree-structured vector quantizer (TSVQ) and the generalized Lloyd algorithm (GLA). The image reconstruction quality with S-TREE2 approaches that of GLA while taking less than 10% of computer time. S-TREE1 and S-TREE2 also compare favorably with the standard TSVQ in both the time needed to create the codebook and the quality of image reconstruction.

Algorithms↗
PubMed2001

The shape of ears to come: dynamic coding of auditory space.

In order to pinpoint the location of a sound source, we make use of a variety of spatial cues that arise from the direction-dependent manner in which sounds interact with the head, torso and external ears. Accurate sound localization relies on the neural discrimination of tiny differences in the values of these cues and requires that the brain circuits involved be calibrated to the cues experienced by each individual. There is growing evidence that the capacity for recalibrating auditory localization continues well into adult life. Many details of how the brain represents auditory space and of how those representations are shaped by learning and experience remain elusive. However, it is becoming increasingly clear that the task of processing auditory spatial information is distributed over different regions of the brain, some working hierarchically, others independently and in parallel, and each apparently using different strategies for encoding sound source location.

Journal Article↗
PubMed1999

Early fraction calculation ability.

Three- to 7-year-olds' ability to calculate with whole-number, fraction, and mixed-number amounts was tested using a nonverbal task in which an amount was displayed and then hidden (J. Huttenlocher, N. C. Jordan, & S. C. Levine, 1994). Next, an amount was added to or subtracted from the hidden amount. The child's task was to determine the hidden amount that resulted from the transformation. Although fraction problems were more difficult than whole-number problems, competence on all problem types emerged in the early childhood period. Furthermore, there were striking parallels between the development of whole-number and fraction calculation. This is inconsistent with the hypothesis that early representations of quantity promote learning about whole numbers but interfere with learning about fractions (e.g., R. Gelman, 1991; K. Wynn, 1995, 1997).

Age Factors↗