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 739 records · Page 41Linked to original sources
PubMed2003

How synapses in the auditory system wax and wane: theoretical perspectives.

Spike-timing-dependent synaptic plasticity has recently provided an account of both the acuity of sound localization and the development of temporal-feature maps in the avian auditory system. The dynamics of the resulting learning equation, which describes the evolution of the synaptic weights, is governed by an unstable fixed point. We outline the derivation of the learning equation for both the Poisson neuron model and the leaky integrate-and-fire neuron with conductance synapses. The asymptotic solutions of the learning equation can be described by a spectral representation based on a biorthogonal expansion.

Action Potentials↗
PubMed1996

Forward Models for Physiological Motor Control.

Based on theoretical and computational studies it has been suggested that the central nervous system (CNS) internally simulates the behaviour of the motor system in planning, control and learning. Such an internal "forward" model is a representation of the motor system that uses the current state of the motor system and motor command to predict the next state. We will outline the uses of such internal models for solving several fundamental computational problems in motor control and then review the evidence for their existence and use by the CNS. Finally we speculate how the location of an internal model within the CNS may be identified. Copyright 1996 Elsevier Science Ltd.

Journal Article↗
PubMed2003

Computational models of the hippocampal region: linking incremental learning and episodic memory.

The hippocampal region, a group of brain structures important for learning and memory, has been the focus of a large number of computational models. These tend to fall into two groups: (1) models of the role of the hippocampal region in incremental learning, which focus on the development of new representations that are sensitive to stimulus regularities and environmental context; (2) models that focus on the role of the hippocampal region in the rapid storage and retrieval of episodic memories. Rather than being in conflict, it is becoming apparent that both approaches are partially correct and might reflect the different functions of substructures of the hippocampal region. Future computational models will help to elaborate how these different substructures interact.

Journal Article↗
PubMed1997

Representational change in young children's understanding of familiar verb meaning.

The ability to generalize verbs to new examples of previously labelled events demonstrates an implicit understanding that verbs are representative symbols of categories of events. The present study examined when and how very young children generalize familiar verbs to novel events by using the preferential looking paradigm. Overall, 24 children aged 1;8 and 25 children aged 2;2 demonstrated their understanding of the verbs kick and pick-up by looking significantly longer at the target events on control trials. Additionally, children aged 1;8 with the largest expressive vocabulary generalized the same verbs to actions with different agents, but not to actions differing in outcome or manner of action. In contrast, children aged 2;2 consistently extended familiar action verbs to other actions differing in agent or manner, regardless of the size of their expressive vocabulary. These findings were not due to the saliency of any of the actions used and are interpreted in terms of representational change consistent with the acquisition of lexical learning principles.

Child Language↗
PubMed2000

Updating geographical knowledge: principles of coherence and inertia.

In 2 experiments, the authors investigated how representations of global geography are updated when people learn new location information about individual cities. Participants estimated the latitude of cities in North America (Experiment 1) and in the Old and New Worlds (Experiment 2). After making their first estimates, participants were given information about the latitudes of 2 cities and asked to make a second set of estimates. Both the first and second estimates revealed evidence for psychologically distinct geographical subregions that were coordinated, in an ordinal sense, across the Atlantic Ocean. Further, the second estimates were affected by the nature of the physical adjacency between regions (e.g., the southern U.S. and Mexico) and by accurate location information about distant, but coordinated, subregions (e.g., the southern U.S. and Mediterranean Europe). The data provide support for a framework for making geographical estimates in which people strike a balance between 2 principles: the need to keep their knowledge base coherent, and the inertial tendency to resist changing the knowledge base unless it is necessary to maintain coherence.

Cognition↗
PubMed2001

Syllable onsets II: three-element clusters in phonological treatment.

This study extends the application of the Sonority Sequencing Principle, as reported in J. A. Gierut (1999), to acquisition of word-initial 3-element clusters by children with functional phonological delays (ages in years;months: 3;4 to 6;3). The representational structure of 3-element clusters is complex and unusual because it consists of an s-adjunct plus a branching onset, which respectively violate and conform to the Sonority Sequencing Principle. Given the representational asymmetry, it is unclear how children might learn these clusters in treatment or whether such treatment may even be effective. Results of a single-subject staggered multiple-baseline experiment demonstrated that children learned the treated 3-element cluster in treatment but showed no further generalization to similar types of (asymmetric) onsets. Treatment of 3-element clusters did, however, result in widespread generalization to untreated singletons, including affricates. Moreover, there was differential generalization to untreated 2-element clusters, with individual differences being traced to the composition of children's singleton inventories. Theoretically, the results suggest a segmental-syllabic interface that holds predictive potential for determining the effectiveness and effects of clinical treatment as based on the notion of linguistic complexity.

Child↗
PubMed1996

Vowel perception in children with and without language impairment.

Twenty-four children with language impairment (LI) and 22 children without language impairment (LN) participated in a study of discrimination, identification, and serial ordering of the highly dissimilar vowels/a/ versus /i/, and the highly similar vowels /epsilon/ versus /ae/. The vowel pairs were presented to the subjects in long- and short-duration sets. Both groups had greater difficulty in identifying /epsilon/ versus /ae/ than /a/ versus /i/. Neither group had greater difficulty with the short- than the long-duration vowel sets. The LI children were less efficient than the LN in identifying /a/ versus /i/, but could identify them accurately. They were significantly less accurate than the LN in identifying /epsilon/ versus /ae/. The majority of the children who could identify the /a/ and /i/ vowels were able to order them serially as well, although this second task appeared to be more difficult than identification. Fewer LI than LN children were able to proceed to the serial ordering task with /epsilon/ and /ae/. The children who could not identify the vowels within a set were almost always able to discriminate them. It was concluded that LI children have an auditory perceptual learning deficit and consequently a less robust central representation for steady state vowels than LN.

Child↗
PubMed2004

Modeling and analysis of heterogeneous regulation in biological networks.

In this study, we propose a novel model for the representation of biological networks and provide algorithms for learning model parameters from experimental data. Our approach is to build an initial model based on extant biological knowledge and refine it to increase the consistency between model predictions and experimental data. Our model encompasses networks which contain heterogeneous biological entities (mRNA, proteins, metabolites) and aims to capture diverse regulatory circuitry on several levels (metabolism, transcription, translation, post-translation and feedback loops, among them). Algorithmically, the study raises two basic questions: how to use the model for predictions and inference of hidden variables states, and how to extend and rectify model components. We show that these problems are hard in the biologically relevant case where the network contains cycles. We provide a prediction methodology in the presence of cycles and a polynomial time, constant factor approximation for learning the regulation of a single entity. A key feature of our approach is the ability to utilize both high-throughput experimental data, which measure many model entities in a single experiment, as well as specific experimental measurements of few entities or even a single one. In particular, we use together gene expression, growth phenotypes, and proteomics data. We tested our strategy on the lysine biosynthesis pathway in yeast. We constructed a model of more than 150 variables based on an extensive literature survey and evaluated it with diverse experimental data. We used our learning algorithms to propose novel regulatory hypotheses in several cases where the literature-based model was inconsistent with the experiments. We showed that our approach has better accuracy than extant methods of learning regulation.

Algorithms↗
PubMed2026

A transcription factor regulatory atlas for activity inference and perturbation prediction.

Inferring transcription factor (TF) activity from transcriptomes and predicting transcriptome-wide responses to TF perturbations remain challenging, in part because available TF-mRNA resources often face a trade-off between precision and coverage and typically lack signed regulatory information. Here, we present TFActProfiler, a TF-mRNA resource and computational framework that learns signed, quantitative TF-mRNA regulatory coefficients by integrating heterogeneous prior evidence (ChIP-based, motif-based, and curated TF-mRNA annotations) with large-scale bulk and single-cell RNA-seq atlases. TFActProfiler contains 2 606 176 signed TF-mRNA interactions and improves TF activity inference in TF knockdown benchmarks relative to widely used regulon resources while retaining broad TF and target coverage. In addition, because the same learned regulatory coefficients can be used to model downstream transcriptional effects, TFActProfiler enables prediction of transcriptome-wide gene expression responses to TF knockdown without training on task-matched perturbation data. When perturbation datasets are available, TFActProfiler can be further refined to achieve performance comparable to state-of-the-art machine-learning baselines. By providing a direction-aware representation of TF-mRNA regulation for both activity inference and perturbation-response modeling, TFActProfiler supports systematic dissection of gene regulatory programs across diverse cellular contexts.

Transcription Factors↗
PubMed1996

Local structural motifs of protein backbones are classified by self-organizing neural networks.

Important and relevant information is expected to be encoded in local structural elements of proteins. An unsupervised learning algorithm (Kohonen algorithm) was applied to the representation and unbiased classification of local backbone structures contained in a set of proteins. Training yielded a two-dimensional Kohonen feature map with 100 different structural motifs including certain helical and strand structures. All motifs were represented in a phi-psi-plot and some of them as a three-dimensional model. The course of structural motifs along the backbone of four selected proteins (cytochrome b5, cytochrome b562, lysozyme, gamma crystallin) was investigated in detail. Trajectories and histograms visualizing the abundance of characteristic motifs allowed for the distinction between different types of protein overall folds. It is demonstrated how the histograms may be used to construct a structural similarity matrix for proteins. The Kohonen algorithm provides a simple procedure for classification of local protein structures independent of any a priori knowledge of leading structural motifs. Training of the Kohonen network leads to the generation of "consensus structures' serving for the task of classification.

Algorithms↗
PubMed2001

Within-session and between-session reproducibility of cerebral sensorimotor activation: a test--retest effect evidenced with functional magnetic resonance imaging.

The aim of the current study was to assess the reproducibility of functional magnetic resonance imaging (fMRI) brain activation signals in a sensorimotor task in healthy subjects. Because random or systematic changes are likely to happen when movements are repeated over time, the authors searched for time-dependent changes in the fMRI signal intensity and the extent of activation within and between sessions. Reproducibility was studied on a sensorimotor task called "the active task" that includes a motor output and a sensory feedback, and also on a sensory stimulation called "the passive task" that assessed the sensory input alone. The active task consisted of flexion and extension of the right hand. The subjects had performed it several times before fMRI scanning so that it was well learned. The passive task consisted of a calibrated passive flexion and extension of the right wrist. Tasks were 1 Hz-paced. The control state was rest. Subjects naïve to the MRI environment and non--MRI-naïve subjects were studied. Twelve MRI-naïve subjects underwent 3 fMRI sessions separated by 5 hours and 49 days, respectively. During MRI scanning, they performed the active task. Six MRI-naïve subjects underwent 2 fMRI sessions with the passive task 1 month apart. Three non--MRI-naïve subjects performed twice an active 2-Hz self-paced task. The data were analyzed with SPM96 software. For within-session comparison, for active or passive tasks, good reproducibility of fMRI signal activation was found within a session (intra-and interrun reproducibility) whether it was the first, second, or third session. Therefore, no within-session habituation was found with a passive or a well-learned active task. For between-session comparison, for MRI-naïve or non--MRI-naïve subjects, and with the active or the passive task, activation was increased in the contralateral premotor cortex and in ispsilateral anterior cerebellar cortex but was decreased in the primary sensorimotor cortex, parietal cortex, and posterior supplementary motor area at the second session. The lower cortical signal was characterized by reduced activated areas with no change in maximum peak intensity in most cases. Changes were partially reversed at the third session. Part of the test-retest effect may come from habituation of the MRI experiment context. Less attention and stress at the second and third sessions may be components of the inhibition of cortical activity. Because the changes became reversed, the authors suggest that, beyond the habituation process, a learning process occurred that had nothing to do with procedural learning, because the tasks were well learned or passive. A long-term memory representation of the sensorimotor task, not only with its characteristics (for example, amplitude, frequency) but also with its context (fMRI), can become integrated into the motor system along the sessions. Furthermore, the pattern observed in the fMRI signal changes might evoke a consolidation process.

Adult↗
PubMed2005

Motor cortex plasticity induced by extensive training revealed by transcranial magnetic stimulation in human.

This study examines the effect of high-level skilled behaviour on motor cortex representations of upper extremity muscles of ten sportswomen. We used transcranial magnetic stimulation to map proximal medial deltoid and distal extensor carpi radialis muscle representations on both hemispheres during low-level voluntary contraction. We compared cortical representation areas between two groups of subjects and between hemispheres within subjects. The first group comprised five elite volleyball attackers and the second group five runners. Four stimuli were delivered on multiple scalp sites (1.5 cm apart) to induce motor-evoked potentials recorded by surface EMG. Maps were described in terms of excitable scalp positions and of motor-evoked potentials. We observed differences in map areas between the two groups. Volleyball players had larger cortical representations of the proximal medial deltoid muscle than runners. Furthermore, the volleyball players had larger map areas for dominant muscles compared with non-dominant muscles. There was no difference, however, in map area for either muscle between the dominant and non-dominant arm in the runner group. Our results show that heavy training in a specific skill induces an expansion of proximal muscle representation in the contralateral primary motor cortex. This enlarged map area for proximal muscle is accompanied by an increase in the overlapping of proximal and distal muscle representations. This could reflect the fact that motor learning of co-ordinated movement involves a common control of both muscles. This reorganization supports the hypothesis of a cortical plasticity driven by activity.

Adult↗
PubMed2000

Knowledge discovery and data mining in toxicology.

Knowledge discovery and data mining tools are gaining increasing importance for the analysis of toxicological databases. This paper gives a survey of algorithms, capable to derive interpretable models from toxicological data, and presents the most important application areas. The majority of techniques in this area were derived from symbolic machine learning, one commercial product was developed especially for toxicological applications. The main application area is presently the detection of structure-activity relationships, very few authors have used these techniques to solve problems in epidemiological and clinical toxicology. Although the discussed algorithms are very flexible and powerful, further research is required to adopt the algorithms to the specific learning problems in this area, to develop improved representations of chemical and biological data and to enhance the interpretability of the derived models for toxicological experts.

Algorithms↗
PubMed2025

Exercise Therapy in Down Syndrome: A Systematic Review and Meta-Analysis Focused on Muscle Strength, Redox Balance, and Inflammatory Profile.

OBJECTIVE: This study systematically reviewed and meta-analyzed randomized and quasi-randomized controlled trials investigating the impact of exercise therapy on muscle strength, redox balance, and inflammatory profile in individuals with Down syndrome. DESIGN: Systematic review and meta-analysis. DATA SOURCES: Cochrane Central Register of Controlled Trials, MEDLINE, CINAHL, SPORTDiscus, EMBASE, and PEDro. ELIGIBILITY CRITERIA FOR SELECTING STUDIES: Randomized and quasi-randomized controlled trials exploring exercise therapy effects on muscle strength and redox balance in individuals with Down syndrome. Although no initial restrictions on age, gender, or health condition were applied during the search process, all included studies focused on adult participants (>18 yr old). No language restrictions were applied, and the search covered the period from 1970 to 2021. RESULTS: We assessed the abstract of 1964 studies. Of the 46 studies meeting the inclusion criteria for the period 2004-2021, 32 focused on muscle strength, and 14 examined redox balance and inflammation. A total of 1611 participants with a mean age of 27 yr were included. This review confirmed that different exercise modalities are prone to improve muscle strength (random effect (95% confidence interval): 0.66, 0.54 to 0.78), redox balance and inflammatory profile (random effect (95% confidence interval): -1.04, -1.31 to -0.76) in this population. The multimodel inference suggested that the frequency of training (times per week) might play a significant role in the main effect. Unsupervised machine learning algorithms displayed a pattern-based graphic representation to assess heterogeneity. CONCLUSIONS: Exercise training demonstrated a positive impact on muscle strength in adults with Down syndrome. The review provides valuable insights into the effects of exercise therapy on individuals with Down syndrome, emphasizing the need for tailored training prescriptions.

Humans↗
PubMed2001

The work environment justice fund: five years of funding work environment justice in Massachusetts.

This article provides an assessment of the impact of five years of funding on work environment justice in Massachusetts. It first outlines the economic, historical, and political context in the 1990s in Massachusetts. Next, it describes the distribution of the Work Environment Justice Fund grants in detail by amounts presented, types of organizations and grants funded, and geographic representation of grantees. A discussion of the lessons learned from the work funded by the WEJF introduces a summary of major accomplishments of the grantees. The report concludes with recommendations for continuation and expansion of funding for grassroots work environment justice in Massachusetts.

Journal Article↗
PubMed1996

[Clinical medical anthropology and immigrant's mental health in France].

Clinical anthropology offers a great advantage for understanding and managing patient/caretaker relationships in intercultural situations. Instead of falling into the trap of marginalizing and above all needless culturalization, health care workers must learn to integrate the cultural aspects of the representation of mental health and illness as opposed to using them as a guiding light. In this way, since the caretaker or therapist does not have to master anthropologic factors, he/she is not obliged to unknowingly hide his/her own nosographic explanatory model which does not necessarily take cultural aspects into account. Clinical anthropology allows the general practitioner and specialist as well as the psychologist and psychiatrist of all theoretical orientations to manage patient relationships and care more efficiently with regard to diagnosis, therapeutic decision-making, analysis, and psychotherapy. The question of whether the patient and caretaker are of the same origin is asked differently: the question of the universality of psychopathology is asked with greater clarity and less risk of error. Our health care system, which is based on common law benefits as do consulting immigrants since their request for services are answered more efficiently and directly. The only problem is that the conceptual and clinical horizon health care workers must be broadened. This goal cannot be achieved by magic and will require training and education.

Acculturation↗
PubMed2002

On the emergence of rules in neural networks.

A simple associationist neural network learns to factor abstract rules (i.e., grammars) from sequences of arbitrary input symbols by inventing abstract representations that accommodate unseen symbol sets as well as unseen but similar grammars. The neural network is shown to have the ability to transfer grammatical knowledge to both new symbol vocabularies and new grammars. Analysis of the state-space shows that the network learns generalized abstract structures of the input and is not simply memorizing the input strings. These representations are context sensitive, hierarchical, and based on the state variable of the finite-state machines that the neural network has learned. Generalization to new symbol sets or grammars arises from the spatial nature of the internal representations used by the network, allowing new symbol sets to be encoded close to symbol sets that have already been learned in the hidden unit space of the network. The results are counter to the arguments that learning algorithms based on weight adaptation after each exemplar presentation (such as the long term potentiation found in the mammalian nervous system) cannot in principle extract symbolic knowledge from positive examples as prescribed by prevailing human linguistic theory and evolutionary psychology.

Algorithms↗
PubMed1996

Neural mechanisms for visual memory and their role in attention.

Recent studies show that neuronal mechanisms for learning and memory both dynamically modulate and permanently alter the representations of visual stimuli in the adult monkey cortex. Three commonly observed neuronal effects in memory-demanding tasks are repetition suppression, enhancement, and delay activity. In repetition suppression, repeated experience with the same visual stimulus leads to both short- and long-term suppression of neuronal responses in subpopulations of visual neurons. Enhancement works in an opposite fashion, in that neuronal responses are enhanced for objects with learned behavioral relevance. Delay activity is found in tasks in which animals are required to actively hold specific information "on-line" for short periods. Repetition suppression appears to be an intrinsic property of visual cortical areas such as inferior temporal cortex and is thought to be important for perceptual learning and priming. By contrast, enhancement and delay activity may depend on feedback to temporal cortex from prefrontal cortex and are thought to be important for working memory. All of these mnemonic effects on neuronal responses bias the competitive interactions that take place between stimulus representations in the cortex when there is more than one stimulus in the visual field. As a result, memory will often determine the winner of these competitions and, thus, will determine which stimulus is attended.

Animals↗