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 199 records · Page 11Linked to original sources
PubMed2002

Generating decision trees from otoneurological data with a variable grouping method.

When medical data sets are modelled by machine learning methods, wealth of variables may be available. This paper deals with variable selection for decision tree induction in the context of two otoneurological data sets: vertigo data, and postoperative nausea and vomiting data. First, a variable grouping method based on measures of association and graph theoretic techniques was used to gain insight into data. Then, representations of learning data were defined using the information from discovered variable groups, and decision trees were generated. The use of variable grouping method was beneficial by revealing interesting associations between variables and enabling generation of accurate and reasonable decision trees that modelled the application areas from different viewpoints.

Data Collection↗
PubMedDate not supplied

Extending models of hippocampal function in animal conditioning to human amnesia.

Although most analyses of amnesia have focused on the loss of explicit declarative and episodic memories following hippocampal-region damage, considerable insights into amnesia can also be realised by studying hippocampal function in simple procedural, or habit-based, associative learning tasks. Although many simple forms of associative learning are unimpaired by hippocampal damage, more complex tasks which require sensitivity to unreinforced stimuli, configurations of multiple stimuli, or contextual information are impaired by hippocampal damage. In several recent papers we have developed a computational theory of hippocampal function which argues that this brain region plays a critical role in the formation of new stimulus representations during learning (Gluck & Myers, 1993, 1995; Myers & Gluck, 1996; Myers, Gluck, & Granger, 1995). We have applied this theory to a broad range of empirical data from studies of classical conditioning in both intact and hippocampal-lesioned animals, and the model correctly accounts for these data. The classical conditioning paradigm can be adapted for use in humans, and similar results for acquisition are obtained in both normal and hippocampal-damaged humans. More recently, we have begun to address an important set of category learning studies in both normals and hippocampal-damaged amnesics. This work integrates experimental studies of amnesic category learning (Knowlton, Squire, & Gluck, 1994) with theoretical accounts of associative learning, and builds on previously established behavioural correspondences between animal conditioning and human category learning (Gluck & Bower, 1988a). Our work to date illustrates some initial progress towards a more integrative understanding of hippocampal function in both animal and human learning, which may be useful in guiding further empirical and theoretical research in human memory and amnesia.

Amnesia↗
PubMed2000

Dynamics and context dependence of visual category learning.

Visual category learning by humans is observed within a paradigm of supervised learning. Mental representations for recognition are reconstructed from the observed data structures by fitting to them predicted classification data obtained from similarity-based models of recognition on the one hand and machine vision systems for image understanding on the other hand. These principles are illustrated with examples concerning the dynamics and the dependence on context of processes of category learning.

Computer Simulation↗
PubMed1998

A self-organizing neural network architecture for navigation using optic flow.

This article describes a self-organizing neural network architecture that transforms optic flow and eye position information into representations of heading, scene depth, and moving object locations. These representations are used to navigate reactively in simulations involving obstacle avoidance and pursuit of a moving target. The network's weights are trained during an action-perception cycle in which self-generated eye and body movements produce optic flow information, thus allowing the network to tune itself without requiring explicit knowledge of sensor geometry. The confounding effect of eye movement during translation is suppressed by learning the relationship between eye movement outflow commands and the optic flow signals that they induce. The remaining optic flow field is due to only observer translation and independent motion of objects in the scene. A self-organizing feature map categorizes normalized translational flow patterns, thereby creating a map of cells that code heading directions. Heading information is then recombined with translational flow patterns in two different ways to form maps of scene depth and moving object locations. Most of the learning processes take place concurrently and evolve through unsupervised learning. Mapping the learned heading representations onto heading labels or motor commands requires additional structure. Simulations of the network verify its performance using both noise-free and noisy optic flow information.

Computer Simulation↗
PubMed2003

Place representation within hippocampal networks is modified by long-term potentiation.

In the brain, information is encoded by the firing patterns of neuronal ensembles and the strength of synaptic connections between individual neurons. We report here that representation of the environment by "place" cells is altered by changing synaptic weights within hippocampal networks. Long-term potentiation (LTP) of intrinsic hippocampal pathways abolished existing place fields, created new place fields, and rearranged the temporal relationship within the affected population. The effect of LTP on neuron discharge was rate and context dependent. The LTP-induced "remapping" occurred without affecting the global firing rate of the network. The findings support the view that learned place representation can be accomplished by LTP-like synaptic plasticity within intrahippocampal networks.

Animals↗
PubMed2003

Representational blending in human conditional learning: Implications for associative theory.

In two experiments, participants were presented with pictures of different foods (A, B, C, D, X,) and learned which combinations resulted in an allergic reaction in a fictitious patient, Mr X. In Problem 1, when A or B (but not C or D) was combined with food X an allergic reaction occurred, and when C or D (but not A or B) was combined with Y an allergic reaction occurred. In Experiment 1, participants also received Problem 2 in which A, B, C, and D interacted with foods V and W either in the same way as X and Y, respectively, or in a different way. Participants performed more proficiently in the former than in the latter condition. In Experiment 2, after training on Problem 1, participants judged whether or not novel combinations of foods (e.g., AB, CD, AD, CB) would cause an allergic reaction in Mr X. They were no more likely to indicate that AB or CD would cause an allergic reaction than AD or CB, but made their judgements more rapidly and with greater confidence on AB and CD trials than on AD and CB trials. These results (1) indicate that shared representations come to be addressed by the components of similar compounds (e.g., AX and BX) that have predicted the same outcome (an allergic reaction), and (2) are inconsistent with standard, associative theories of learning, but (3) are consistent with findings from nonhuman animals and with a connectionist interpretation of these findings.

Association↗
PubMed1997

Conceptual apraxia from lateralized lesions.

Models of praxis have posited two major components, production and conceptual. Conceptual praxis disorders may occur in two domains: associative knowledge (tool-action associations such as hammer pound; tool-object associations such as hammer nail) and mechanical knowledge such as knowing the advantage that tools afford. Patients with Alzheimer's disease not only have conceptual apraxia (CA) but can dissociate CA from language deficits and from praxis production deficits (ideomotor apraxia). These findings suggests that knowledge about tools (action semantics) is independent of verbal semantics as well as movement representations. To learn if conceptual praxis knowledge is stored in one hemisphere (right or left) and if associative and mechanical conceptual praxis knowledge can be dissociated, we studied 29 right-handed subjects with unilateral strokes. Ten had left-hemisphere damage with no ideomotor apraxia. Eleven had left-hemisphere damage with ideomotor apraxia. There were eight right-hemisphere-damaged controls and 10 normal controls. These subjects were given tests for conceptual apraxia. There was a significant difference between groups, the left-hemisphere group with ideomotor apraxia being most impaired on both the associative and mechanical CA tests. There was a trend for associative and mechanical knowledge to be dissociated. Although conceptual praxis representations are stored in the left hemisphere, analysis of lesion sites did not reveal where in the left hemisphere they may be stored.

Aged↗
PubMed2005

Processes of change in brain and cognitive development.

We review recent advances in the understanding of the mechanisms of change that underlie cognitive development. We begin by describing error-driven, self-organizing and constructivist learning systems. These powerful mechanisms can be constrained by intrinsic factors, other brain systems and/or the physical and social environment of the developing child. The results of constrained learning are representations that themselves are transformed during development. One type of transformation involves the increasing specialization and localization of representations, resulting in a neurocognitive system with more dissociated streams of processing with complementary computational functions. In human development, integration between such streams of processing might occur through the mediation of language.

Adolescent↗
PubMed2001

Delayed upregulation of GABA(A) alpha1 receptor subunit mRNA in somatosensory cortex of mice following learning-dependent plasticity of cortical representations.

Experience-dependent modifications of cortical representational maps are accompanied by changes in several components of GABAergic inhibitory neurotransmission system. We examined with in situ hybridization to 35S-labeled oligoprobe changes of expression of GABA(A) receptor alpha1 subunit mRNA in the barrel cortex of mice after sensory conditioning training. One day and 5 days after the end of short lasting (3 daily sessions) training an increased expression of GABA(A) alpha1 mRNA was observed at the cortical site where the plastic changes were previously found. Learning associated activation of the cerebral cortex increases expression of GABA(A) receptor mRNA after a short post-training delays.

Animals↗
PubMed2002

What can the hippocampal representation of environmental geometry tell us about Hebbian learning?

The importance of the hippocampus in spatial representation is well established. It is suggested that the rodent hippocampal network should provide an optimal substrate for the study of unsupervised Hebbian learning. We focus on the firing characteristics of hippocampal place cells in morphologically different environments. A hard-wired quantitative geometric model of individual place fields is reviewed and presented as the framework in which to understand the additional effects of synaptic plasticity. Existent models employing Hebbian learning are also reviewed. New information is presented regarding the dynamics of place field plasticity over short and long time scales in experiments using barriers and differently shaped walled environments. It is argued that aspects of the temporal dynamics of stability and plasticity in the hippocampal place cell representation both indicate modifications to, and inform the nature of, the synaptic plasticity in place cell models. Our results identify a potential neural basis for long-term incidental learning of environments and provide strong constraints for the way the unsupervised learning in cell assemblies envisaged by Hebb might occur within the hippocampus.

Action Potentials↗
PubMedDate not supplied

Infants' developing appreciation of similarities between model objects and their real-world referents.

Previous research suggests that model competence does not emerge until relatively late in infancy (20-26 months). Development was systematically analyzed within 3 key areas--count noun learning, dual representation, and categorization-hypothesized to support the emergence of model competence in the second year. In an object-handling preferential looking task, 21- to 26-month-olds matched model objects to referents only when count noun knowledge was high. When dual representation demands were reduced through the use of pictures in place of model objects, 20-month-olds with low count noun vocabularies succeeded in relating symbols to referents. Finally, change in infants' construal of a model object as a member of a category was documented between 14 and 20 months of age.

Attitude↗
PubMed2000

Memory systems in the brain.

The operation of different brain systems involved in different types of memory is described. One is a system in the primate orbitofrontal cortex and amygdala involved in representing rewards and punishers, and in learning stimulus-reinforcer associations. This system is involved in emotion and motivation. A second system in the temporal cortical visual areas is involved in learning invariant representations of objects. A third system in the hippocampus is implicated in episodic memory and in spatial function. Fourth, brain systems in the frontal and temporal cortices involved in short term memory are described. The approach taken provides insight into the neuronal operations that take place in each of these brain systems, and has the aim of leading to quantitative biologically plausible neuronal network models of how each of these memory systems actually operates.

Animals↗
PubMed2006

Multi-dimensional profiling of medical students' cognitive models about learning.

INTRODUCTION: In current constructivist paradigms, learners' previous subject-matter knowledge, or cognitive models, provide the foundations for the construction of new knowledge. Learners' cognitive models about learning also mediate students' capacities to learn in their chosen topics of study. The diverse backgrounds of students entering medicine suggest that they might come to medical studies equipped with a wide variety of cognitive models about learning. Some current theories tend to reduce students' cognitions about learning to parsimonious representations, such as surface-deep approaches or mastery-performance goals. It is possible that such reduced representations underrepresent, or misrepresent, the complexity of students' cognitive models about learning. Good quality teaching needs to take account of learners' cognitive models, not just about subject matter, but also about learning. This study investigated the diversity and complexity of medical students' cognitive models about learning. METHODS: A total of 7 graduate entry, clinical-year medical students volunteered for in-depth interviews about learning. NUD*IST text analysis software and correspondence analysis were employed to identify dimensions and to profile students' responses. RESULTS: The correspondence analysis identified a significant 4-dimensional solution that illustrates the contributions of multiple variables to students' cognitive models about learning. Individual profiles highlight diversity between participants. DISCUSSION: This study provides evidence that students' cognitive models about learning are complex and highly differentiated. Representations of what students know about learning need to take account of such complexity in order to inform instructional practice more adequately.

Cognition↗
PubMed2026

Disentangling covariate effects on single-cell-resolved epigenomes with DeepDive.

Understanding the effects of individual biological factors from single-cell-resolved epigenomic data is hindered by multicollinearity, particularly in human cohorts. We introduce DeepDive, a deep-learning framework designed to systematically disentangle known and unknown sources of variation in single-nucleus ATAC-seq data. DeepDive accurately reconstructs chromatin accessibility, outperforms state-of-the-art methods with incomplete covariate information, and robustly recovers true biological signals from even highly entangled covariates, unlocking counterfactual, "what-if," analyses. Applying DeepDive to pancreatic islet cells, we perform counterfactual analyses to prioritize covariates associated with a type 2 diabetes-linked beta-cell subtype and nominate transcription regulators. DeepDive offers a powerful and unbiased tool for mechanistic discovery in complex human disease cohorts.

disentanglement↗
PubMed2026

A nonlinear multi-omics data integration and classification model based on pathway self-attention and graph convolutional networks.

The abundance of omics data has significantly advanced the development of multi-omics data integration techniques. Non-linear embedding approaches for data integration have gradually become the mainstream in multi-omics research, as these approaches can substantially improve cancer analysis by enhancing the quality of the embeddings. However, current multi-omics data integration methods are typically confined to omics measurements, neglecting domain-specific prior knowledge encompassing biological pathways. In this study, we proposed a multi-omics integrated classification model, PathTransGCN, based on pathway self-attention and graph convolutional networks (GCN). The model integrated biological pathway information into multi-omics data analysis with the aim of enhancing the accuracy of cancer classification. Multi-omics data for breast cancer (BRCA), non-small cell lung cancer (NSCLC), and low-grade glioma (LGG) were obtained from The Cancer Genome Atlas (TCGA) and UCSC Xena databases. These data included gene mutations, DNA methylation, copy number variations, and gene expression, and were used to assess the model's generalizability across different cancers. First, PathTransGCN employed a pathway self-attention module to learn latent representations of samples across different pathways, thereby obtaining multi-omics integration vectors. Concurrently, a patient similarity network (PSN) was constructed using the similarity network fusion (SNF) approach. Second, the integrated vectors and the PSN were jointly fed into a GCN for end-to-end training, enabling precise classification of cancer subtypes. Through multi-omics data analysis of the BRCA dataset, PathTransGCN outperformed several popular algorithms (such as MoGCN and DeePathNet) in the five-class classification of cancer subtypes, achieving an accuracy rate of 87.6% and an F1 score of 86.4%. Moreover, the model demonstrated robust generalization capabilities across both NSCLC and LGG datasets, while effectively identifying key disease-associated biomarkers at the pathway level. Experimental results demonstrate that PathTransGCN exhibits outstanding performance in integrating omics data and delivering interpretable classification outcomes, presenting significant potential for clinical applications.

Humans↗
PubMed1994

The use of pigmentation and shading information in recognising the sex and identities of faces.

An investigation of what can be learned about representational processes in face recognition from the independent and combined effects of inverting and negating facial images is reported. In experiment 1, independent effects of inversion and negation were observed in a task of identifying famous faces. In experiments 2 through 4 the question of whether effects of negation were still obtained when effects due to the reversal of pigmentation in negative images were eliminated was examined. By the use of images of the 3-D surfaces of faces measured by laser, and displays as smooth surfaces devoid of pigmentation, only effects of inversion were obtained reliably, suggesting that the effects observed in experiment 1 arose largely through the inversion of pigmentation values in normal images of faces. The results of experiment 5 suggested that the difference was not due to the different task demands of experiments 2-4 compared with those of experiment 1. When normally pigmented face images were used in a task making similar demands to that of experiment 4, independent effects of inversion and negation were again observed. When a task of sex classification was used in experiments 6 and 7, clear effects of negation as well as inversion were observed on latencies, though not accuracies, of responding. The results are interpreted in terms of the information content of pigmentation relative to shape from shading in different face-classification tasks. The results also reinforce other recent evidence demonstrating the importance of image intensity as well as spatial layout of face 'features'.

Adult↗
PubMed1996

A computational model of cholinergic disruption of septohippocampal activity in classical eyeblink conditioning.

A previous neurocomputational model of corticohippocampal interaction (Gluck & Myers, 1993) can provide a framework for examining the behavioral effects of septohippocampal modulation during classical conditioning. The model assumes that the hippocampal region is necessary for forming new stimulus representations during learning, but not for the formation of simple associations. This paper considers how septohippocampal interaction could affect this function. The septal nuclei provide several modulatory inputs to the hippocampus, including a cholinergic input which Hasselmo (1995) has suggested may function to regulate hippocampal dynamics on a continuum between two states: a storage state in which incoming information is encoded as an intermediate-term memory and a recall state when this information is reactivated. In this theory, anticholinergic drugs such as scopolamine should disrupt learning by selectively reducing the hippocampus's ability to store new information. An approximation of Hasselmo's idea can be implemented in the corticohippocampal model by a simple manipulation of hippocampal learning rate; this manipulation is formally equivalent to adjusting the amount of time the hippocampus spends in learning and recall states. With this manipulation, the model successfully accounts for the effects of scopolamine in retarding classical conditioning in humans (Solomon, Groccia-Ellison, Flynn, Mirak, Edwards, Dunehew, & Stanton, 1993) and animals (Solomon, Soloman, van der Schaaf, & Perry, 1983). The model further predicts that although cholinergic agonists (such as Tacrine) may improve learning in subjects with artificially depressed brain acetylcholine levels, there may be limited memory improvement in normal subjects from such cholinergic therapy. This is consistent with the general finding of a U-shaped dose response curve for cholinergic drugs in normal subjects: low to moderate doses may improve learning, but higher doses are ineffective or even degrade learning (e.g., Ennaceur & Meliani, 1992; Dumery, Derer, & Blozovski, 1988; etc.).

Animals↗
PubMed1987

Cross-modal recognition of familiar and unfamiliar objects by the monkey: the effects of ablation of polysensory neocortex or of the amygdaloid complex.

Nine rhesus monkeys were trained to a standard level of cross-modal recognition (CMR) in the directions vision to touch and touch to vision. Their level of performance remained essentially unchanged with unfamiliar objects. Five then received bilateral removals of frontal, temporal and parietal polysensory cortex in one stage or successively, and 4 underwent removal of the amygdaloid complex in one stage. All animals were retrained to criterion with familiar objects and then again tested with unfamiliar objects. Postoperatively, the monkeys with extensive neocortical removals were unimpaired or slightly impaired with familiar objects, and slightly impaired (in only one direction) with unfamiliar objects. The animals with amygdaloid ablations showed a different pattern of change: with familiar objects they were unimpaired (if removals were less extensive), or were severely but transiently impaired (if the removals of the amygdala were more extensive and/or other structures were involved); with unfamiliar objects they were unimpaired, tending to improve. The neocortical polysensory areas may be necessary for generating new visual representations during learning--a performance required only for the CMR of unfamiliar objects.

Afferent Pathways↗