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 1,747 records · Page 97Linked to original sources
PubMed2003

Bilateral peripheral vestibular lesions produce long-term changes in spatial learning in the rat.

In order to investigate whether bilateral peripheral vestibular lesions cause long-term impairment of spatial learning, rats were tested in a reference memory radial arm maze learning task at least 5 weeks following a bilateral labyrinthectomy (BL) or sham control lesion. All control rats reached criterion (i.e., 1 error or less, averaged across 7 trials for 3 consecutive days of training) but only 4 of the 8 BL rats had reached criterion by day 21 of the training sessions. The control rats reached criterion more quickly than the lesioned rats (Control, 7.0 +/- 0.63 days, Lesioned, 15.8 +/- 1.4 days, t10= 5.84, p < 0.0001). This difference resulted from the greater number of errors made by the BL animals. However, the latency to respond was comparable as a result of the increased locomotor activity of the BL group (i.e., 'hyperkinesis), and the overall rate of acquisition of the task, as indicated by analysis of the exponential decrease in errors over the entire training period, was not significantly different between the 2 groups. The results of this study demonstrate that BL in rats produces long-term changes in performance in a spatial reference memory task, which are not simply due to the inability to move but may relate to the way that the brain uses vestibular information to create spatial representations and determines behavioural strategies on the basis of these representations.

Animals↗
PubMed2006

Individual recognition using gait energy image.

In this paper, we propose a new spatio-temporal gait representation, called Gait Energy Image (GEI), to characterize human walking properties for individual recognition by gait. To address the problem of the lack of training templates, we also propose a novel approach for human recognition by combining statistical gait features from real and synthetic templates. We directly compute the real templates from training silhouette sequences, while we generate the synthetic templates from training sequences by simulating silhouette distortion. We use a statistical approach for learning effective features from real and synthetic templates. We compare the proposed GEI-based gait recognition approach with other gait recognition approaches on USF HumanID Database. Experimental results show that the proposed GEI is an effective and efficient gait representation for individual recognition, and the proposed approach achieves highly competitive performance with respect to the published gait recognition approaches.

Algorithms↗
PubMed2007

A boosting approach to flexible semiparametric mixed models.

In linear mixed models the influence of covariates is restricted to a strictly parametric form. With the rise of semi- and non-parametric regression also the mixed model has been expanded to allow for additive predictors. The common approach uses the representation of additive models as mixed models. An alternative approach that is proposed in the present paper is likelihood based boosting. Boosting originates in the machine learning community where it has been proposed as a technique to improve classification procedures by combining estimates with reweighted observations. Likelihood based boosting is a general method which may be seen as an extension of L2 boost. In additive mixed models the advantage of boosting techniques in the form of componentwise boosting is that it is suitable for high dimensional settings where many explanatory variables are present. It allows to fit additive models for many covariates with implicit selection of relevant variables and automatic selection of smoothing parameters. Moreover, boosting techniques may be used to incorporate the subject-specific variation of smooth influence functions by specifying 'random slopes' on smooth effects. This results in flexible semiparametric mixed models which are appropriate in cases where a simple random intercept is unable to capture the variation of effects across subjects.

Cohort Studies↗
PubMed2002

Spatial bias: effects of early reading direction on Korean subjects.

Spatial bias may occur in subjects performing a number of cognitive and visual-motor tasks. These include coordinate visuospatial computations (e.g. bisecting a line) and categorical representations of syntactic information (e.g. drawing a picture depicting the action of a sentence). Readers of European languages scan from left-to-right and this learned visual scanning may contribute to leftward spatial bias. In 30 subjects who first learned to read in a top-to-bottom, right-to-left direction (right-left vertical readers, RL), we tested spatial-syntactic bias by reading sentences to subjects, who drew pictures depicting the actions. We noted whether the subject of the sentence was located leftward or rightward of the object. We assessed visual-spatial bias by measuring subjects' accuracy at bisecting lines, and by measuring how closely their drawings on the house-tree-person test were centered on the page. On the spatial-syntactic task, the RL were not right- or left-biased (P=0.581). Korean controls (left-right horizontal readers, LR) also showed no significant spatial-syntactic bias. RL only tended to bisect lines leftward, but displaced house-tree-person drawings left of page center (P<0.001). LR erred leftward on line bisection, and had a smaller magnitude leftward bias on drawing tasks. We conclude that a leftward spatial-syntactic bias may not be innate and does not appear to be influenced by learned reading direction. In contrast, the leftward visual-spatial bias may occur in subjects whose cultural and reading background is neither western nor left-to-right.

Adult↗
PubMed2000

Oscillator-based memory for serial order.

A computational model of human memory for serial order is described (OSCillator-based Associative Recall [OSCAR]). In the model, successive list items become associated to successive states of a dynamic learning-context signal. Retrieval involves reinstatement of the learning context, successive states of which cue successive recalls. The model provides an integrated account of both item memory and order memory and allows the hierarchical representation of temporal order information. The model accounts for a wide range of serial order memory data, including differential item and order memory, transposition gradients, item similarity effects, the effects of item lag and separation in judgments of relative and absolute recency, probed serial recall data, distinctiveness effects, grouping effects at various temporal resolutions, longer term memory for serial order, list length effects, and the effects of vocabulary size on serial recall.

Association↗
PubMed2002

Overlap of internal models in motor cortex for mechanical loads during reaching.

A hallmark of the human motor system is its ability to adapt motor patterns for different environmental conditions, such as when a skilled ice-hockey player accurately shoots a puck with or without protective equipment. Each object (stick, shoulder pad, elbow pad) imparts a distinct load upon the limb, and a key problem in motor neuroscience is to understand how the brain controls movement for different mechanical contexts. We addressed this issue by training non-human primates to make reaching movements with and without viscous loads applied to the shoulder and/or elbow joints, and then examined neural representations in primary motor cortex (MI) for each load condition. Even though the shoulder and elbow loads are mechanically independent, we found that some neurons responded to both of these single-joint loads. Furthermore, changes in activity of individual neurons during multi-joint loads could be predicted from their response to subordinate single-joint loads. These findings suggest that neural representations of different mechanical contexts in MI are organized in a highly structured manner that may provide a neural basis for how complex motor behaviour is learned from simpler motor tasks.

Action Potentials↗
PubMed2002

Prediction of contact maps by GIOHMMs and recurrent neural networks using lateral propagation from all four cardinal corners.

MOTIVATION: Accurate prediction of protein contact maps is an important step in computational structural proteomics. Because contact maps provide a translation and rotation invariant topological representation of a protein, they can be used as a fundamental intermediary step in protein structure prediction. RESULTS: We develop a new set of flexible machine learning architectures for the prediction of contact maps, as well as other information processing and pattern recognition tasks. The architectures can be viewed as recurrent neural network implemantations of a class of Bayesian networks we call generalized input-output HMMs (GIOHMMs). For the specific case of contact maps, contextual information is propagated laterally through four hidden planes, one for each cardinal corner. We show that these architectures can be trained from examples and yield contact map predictors that outperform previously reported methods. While several extensions and improvements are in progress, the current version can accurately predict 60.5% of contacts at a distance cutoff of 8 A and 45% of distant contacts at 10 A, for proteins of length up to 300.

Algorithms↗
PubMed2004

Feature subset selection for support vector machines through discriminative function pruning analysis.

In many pattern classification applications, data are represented by high dimensional feature vectors, which induce high computational cost and reduce classification speed in the context of support vector machines (SVMs). To reduce the dimensionality of pattern representation, we develop a discriminative function pruning analysis (DFPA) feature subset selection method in the present study. The basic idea of the DFPA method is to learn the SVM discriminative function from training data using all input variables available first, and then to select feature subset through pruning analysis. In the present study, the pruning is implement using a forward selection procedure combined with a linear least square estimation algorithm, taking advantage of linear-in-the-parameter structure of the SVM discriminative function. The strength of the DFPA method is that it combines good characters of both filter and wrapper methods. Firstly, it retains the simplicity of the filter method avoiding training of a large number of SVM classifier. Secondly, it inherits the good performance of the wrapper method by taking the SVM classification algorithm into account.

Journal Article↗
PubMed2000

Tactile coactivation-induced changes in spatial discrimination performance.

We studied coactivation-based cortical plasticity at a psychophysical level in humans. For induction of plasticity, we used a protocol of simultaneous pairing of tactile stimulation to follow as closely as possible the idea of Hebbian learning. We reported previously that a few hours of tactile coactivation resulted in selective and reversible reorganization of receptive fields and cortical maps of the hindpaw representation of the somatosensory cortex of adult rats (Godde et al., 1996). In the present study, simultaneous spatial two-point discrimination was tested on the tip of the right index finger in human subjects as a marker of plastic changes. After 2 hr of coactivation we found a significant improvement in discrimination performance that was reversible within 8 hr. Reduction of the duration of the coactivation protocol revealed that 30 min was not sufficient to drive plastic changes. Repeated application of coactivation over 3 consecutive days resulted in a delayed recovery indicating stabilization of the improvement over time. Perceptual changes were highly selective because no transfer of improved performance to fingers that were not stimulated was found. The results demonstrate the potential role of sensory input statistics (i.e., their probability of occurrence and spatiotemporal relationships) in the induction of cortical plasticity without involving cognitive factors such as attention or reinforcement.

Adult↗
PubMed2002

Self-organizing continuous attractor networks and path integration: one-dimensional models of head direction cells.

Some neurons encode information about the orientation or position of an animal, and can maintain their response properties in the absence of visual input. Examples include head direction cells in rats and primates, place cells in rats and spatial view cells in primates. 'Continuous attractor' neural networks model these continuous physical spaces by using recurrent collateral connections between the neurons which reflect the distance between the neurons in the state space (e.g. head direction space) of the animal. These networks maintain a localized packet of neuronal activity representing the current state of the animal. We show how the synaptic connections in a one-dimensional continuous attractor network (of for example head direction cells) could be self-organized by associative learning. We also show how the activity packet could be moved from one location to another by idiothetic (self-motion) inputs, for example vestibular or proprioceptive, and how the synaptic connections could self-organize to implement this. The models described use 'trace' associative synaptic learning rules that utilize a form of temporal average of recent cell activity to associate the firing of rotation cells with the recent change in the representation of the head direction in the continuous attractor. We also show how a nonlinear neuronal activation function that could be implemented by NMDA receptors could contribute to the stability of the activity packet that represents the current state of the animal.

Animals↗
PubMed2006

Contribution of familiarity and recollection to associative recognition memory: insights from event-related potentials.

Within the dual-process perspective of recognition memory, it has been claimed that familiarity is sufficient to support recognition of single items, but recollection is necessary for associative recognition of item pairs. However, there are some reports suggesting that familiarity might support associative recognition judgments when the items form an easy to access bound representation. In contrast, recollection seems to be required for the recognition of bindings that might be flexibly rearranged in novel situations. We investigated whether both forms of binding are mediated by different mechanisms as reflected by a qualitatively different spatiotemporal eventrelated potential (ERP) pattern. In a recognition memory experiment, subjects gave old/new judgments to words learned by focusing either on interitem associations or on size relation of word triplets. Results revealed higher hit rates in the relational condition as compared to the associative condition. In addition, the proportion of triplets from which all three items were remembered was significantly larger in the relational condition suggesting that memory retrieval in this condition relies primarily on bound representations of word triplets. The ERP revealed a late parietal old/new effect for both conditions, with relational processing resulting in a greater effect. In contrast, an early frontal old/new effect was solely present in the associative condition. Taken together, these data provide evidence that familiarity might support associative recognition if the associated components are coherently encoded into a bound representation. Recollection might foster the recognition of relational bindings among items. This indicates that the contribution of familiarity and recollection to associative recognition depends on the kind of binding operations performed on the items rather than on the single versus multiple item distinction.

Adult↗
PubMed2006

Children's autobiographies of learning to write.

BACKGROUND: Learning conceptions may be studied as specific implicit theories based on theory of mind. Previous studies suggest that a developmental shift from a direct implicit theory of learning to an interpretative one occurs during childhood. Here we explore the development of children's autobiographies of learning to write by adopting this framework. AIMS: We aim to study children's autobiographical accounts of learning to write and whether these accounts change according to the mentioned developmental shift and sociocultural environment. SAMPLE: Sixty children attending public schools in Argentina that were equally distributed according to school level (kindergarten, first grade, fourth grade) and sociocultural environment (middle and marginal). METHODS: Children were interviewed individually at school and requested to write 'as you used to when you were just beginning to write'. This question was repeated for one-year-intervals until child's current age was reached; anticipation of writing next year was also requested. Children's responses were coded into descriptive categories. A multiple correspondence factorial analysis studied the relations among response categories, school grade and sociocultural environment. On the basis of these results, categories were ordered into a Guttman scale. RESULTS: A developmental shift from a focus on isolated products to the integration of procedural and representational changes was evident. School grade and sociocultural environment were statistically significant. CONCLUSIONS: Children's writing autobiographies show developmental differences that suit the shift from a direct to an interpretative theory of learning on the basis of hierarchical integration rather than conceptual replacement.

Autobiographies as Topic↗
PubMedDate not supplied

A molecular biological approach to synaptic plasticity and learning.

Until the more recent advances made in molecular biology, attempts to link synaptic plasticity and learning have focused on using LTP as a marker of learning-induced synaptic plasticity, where one has expected to observe the same magnitude of change in synaptic strength as that observed with artificial stimulation. To a large extent this approach has been frustrated by the fact that it is generally assumed that the representation of the memory traces is distributed throughout widespread networks of cells. By implication it is more likely that one would observe small distributed changes within a network; a formidable task to measure. In this review we describe how the advances in molecular biology give us both the tools to investigate the mechanisms of synaptic plasticity and to apply these to investigations of the underlying mechanisms in learning and the formation of memories that have until now remained out of our grasp.

Animals↗
PubMed2026

NeuroOmics-Net: An interpretable multimodal deep learning framework for Alzheimer's disease diagnosis and progression prediction using neuroimaging, EEG, and genomic data.

Accurate diagnosis and progression prediction of Alzheimer's disease (AD) remain challenging due to the heterogeneous nature of the disease, which involves structural brain degeneration, electrophysiological dysfunction, and molecular dysregulation. Most existing deep learning approaches rely on a single modality or limited multimodal combinations, thereby failing to capture the complex cross-domain interactions underlying AD progression. Furthermore, the scarcity of large-scale datasets containing synchronized neuroimaging, electrophysiological, and genomic measurements restricts the development of comprehensive multimodal diagnostic systems. To address these challenges, this study proposes NeuroOmics-Net, a multimodal deep learning framework for Alzheimer's disease analysis that integrates structural magnetic resonance imaging (sMRI), electroencephalography (EEG), and gene expression data. The proposed framework combines a Hierarchical Multi-View Encoder (HME) for modality-specific feature extraction, a Cross-Omics Attention Fusion (CAF) module for adaptive integration of complementary biomarkers, and a Disease Progression Graph Learning (DPGL) module for modeling progression-related relationships across biological domains. To facilitate cross-modal integration from independent cohorts, Regularized Canonical Correlation Analysis (RCCA) is employed to align heterogeneous feature representations within a shared latent space. Experiments were conducted using publicly available datasets from ADNI, PhysioNet, and GEO repositories comprising 1120 diagnosis-aligned samples. The proposed framework achieved 94.3% classification accuracy and an AUC of 0.975 for distinguishing normal controls (NC), mild cognitive impairment (MCI), and Alzheimer's disease subjects, while attaining 93.7% accuracy for predicting conversion from stable mild cognitive impairment (sMCI) to progressive mild cognitive impairment (pMCI). However, a fairness sensitivity analysis using stratified demographic reweighting revealed accuracy ranging from 90.8% (low-education, high-comorbidity proxy subgroup) to 96.1% (low-risk, high-reserve proxy subgroup), a demographic parity gap of 5.3 percentage points, indicating that overall accuracy reflects a performance ceiling in a relatively homogeneous research cohort rather than a realistic estimate for demographically diverse clinical populations. Comparative evaluations demonstrated consistent improvements over state-of-the-art unimodal and multimodal deep learning models. Interpretability analysis further identified clinically relevant biomarkers, including hippocampal and entorhinal atrophy, theta-alpha EEG alterations, and APOE-associated molecular pathways. Because sMRI, EEG, and gene expression data were sourced from separate, unpaired cohorts with no subjects possessing all three synchronized measurements, all reported cross-modal associations reflect population-level statistical correspondence across diagnosis-matched groups rather than within-subject physiological coupling; no claim of intra-individual causal cross-modal interaction is made. These findings demonstrate that NeuroOmics-Net provides an effective computer-aided framework for multimodal biomedical data processing and Alzheimer's disease analysis. By integrating neuroimaging, electrophysiological, and genomic information, the proposed approach enables accurate diagnosis, progression prediction, and biologically interpretable decision support for clinical and translational applications.

Humans↗
PubMed2004

Navigation in virtual environment by the macaque monkey.

To determine whether monkeys can navigate a virtual environment, and whether they can represent a virtual environment in their brain, we trained two macaque monkeys to perform a navigation task in a virtual building and conducted behavioral analyses. After learning the task from the first starting point, they were trained to perform the navigation task from a second starting point which is already familiar to them during the first training session. The second training was completed faster than the first one. Furthermore, we found that the time required for a free-way-finding test from a novel starting point was shortened after the second training. These results suggest that monkeys can navigate a virtual environment and can construct a flexible representation of a virtual environment in their brain.

Animals↗
PubMed1995

Feedback circuitry within a song-learning pathway.

The song system of birds consists of several neural pathways. One of these, the anterior forebrain pathway, is necessary for the acquisition but not for the production of learned song in zebra finches. It has been shown that the anterior forebrain pathway sequentially connects the following nuclei: the high vocal center, area X of lobus parolfactorius, the medial portion of the dorsolateral thalamic nucleus, the lateral magnocellular nucleus of anterior neostriatum (IMAN), and the robust nucleus of the archistriatum (RA). We now show in zebra finches (Taeniopygia guttata) that IMAN cells that project to RA also project to area X, forming a feedback loop within the anterior forebrain pathway. The axonal endings of the IMAN projection into area X form cohesive and distinct domains. Small injections of tracer in subregions of area X backfill a spatially restricted subset of cells in IMAN, that, in turn, send projections to RA that are arranged in horizontal layers, which may correspond to the functional representation of vocal tract muscles demonstrated by others. We infer from our data that there is a myotopic representation throughout the anterior forebrain pathway. In addition, we suggest that the parcellation of area X into smaller domains by the projection from IMAN highlights a functional architecture within X, which might correspond to units of motor control, to the representation of acoustic features of song, or both.

Animals↗
PubMed2006

Asymmetrical participation of the left and right hippocampus for representing environmental geometry in homing pigeons.

Control, right and left HF lesioned homing pigeons (Columba livia) were trained to locate a goal in one corner of a rectangular enclosure with a distinctive feature cue. Probe tests revealed that all groups were able to encode in parallel geometric (enclosure shape) and feature information, and in the absence of one of them, they could us the other to locate the goal. However, left HF lesioned pigeons learned the task at a faster rate, and when the geometric and feature information were set in conflict, they relied more on the feature cue compared to control and right HF lesioned pigeons. It was also found that pigeons, independent of group, trained to a goal adjacent to the feature cue learned the task in fewer sessions and relied more on feature information compared to pigeons trained to a goal opposite the feature cue. The latter group relied more on geometric information. The results support the hypothesis that the left HF plays a more important role in the representation of a goal location with respect to environmental shape/geometry. We further propose that the observed functional asymmetry can be explained by the lateralized properties of the pigeon tectofugal visual system.

Analysis of Variance↗