Search PubMed⌕ Search

SEARCH · Search PubMed

Results for “Ensemble 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 91 records · Page 5Linked to original sources
PubMed2005

A stochastic population approach to the problem of stable recruitment hierarchies in spiking neural networks.

Synchrony-driven recruitment learning addresses the question of how arbitrary concepts, represented by synchronously active ensembles, may be acquired within a randomly connected static graph of neuron-like elements. Recruitment learning in hierarchies is an inherently unstable process. This paper presents conditions on parameters for a feedforward network to ensure stable recruitment hierarchies. The parameter analysis is conducted by using a stochastic population approach to model a spiking neural network. The resulting network converges to activate a desired number of units at each stage of the hierarchy. The original recruitment method is modified first by increasing feedforward connection density for ensuring sufficient activation, then by incorporating temporally distributed feedforward delays for separating inputs temporally, and finally by limiting excess activation via lateral inhibition. The task of activating a desired number of units from a population is performed similarly to a temporal k-winners-take-all network.

Feedback, Physiological↗
PubMed2006

A two-stage approach for improved prediction of residue contact maps.

BACKGROUND: Protein topology representations such as residue contact maps are an important intermediate step towards ab initio prediction of protein structure. Although improvements have occurred over the last years, the problem of accurately predicting residue contact maps from primary sequences is still largely unsolved. Among the reasons for this are the unbalanced nature of the problem (with far fewer examples of contacts than non-contacts), the formidable challenge of capturing long-range interactions in the maps, the intrinsic difficulty of mapping one-dimensional input sequences into two-dimensional output maps. In order to alleviate these problems and achieve improved contact map predictions, in this paper we split the task into two stages: the prediction of a map's principal eigenvector (PE) from the primary sequence; the reconstruction of the contact map from the PE and primary sequence. Predicting the PE from the primary sequence consists in mapping a vector into a vector. This task is less complex than mapping vectors directly into two-dimensional matrices since the size of the problem is drastically reduced and so is the scale length of interactions that need to be learned. RESULTS: We develop architectures composed of ensembles of two-layered bidirectional recurrent neural networks to classify the components of the PE in 2, 3 and 4 classes from protein primary sequence, predicted secondary structure, and hydrophobicity interaction scales. Our predictor, tested on a non redundant set of 2171 proteins, achieves classification performances of up to 72.6%, 16% above a base-line statistical predictor. We design a system for the prediction of contact maps from the predicted PE. Our results show that predicting maps through the PE yields sizeable gains especially for long-range contacts which are particularly critical for accurate protein 3D reconstruction. The final predictor's accuracy on a non-redundant set of 327 targets is 35.4% and 19.8% for minimum contact separations of 12 and 24, respectively, when the top length/5 contacts are selected. On the 11 CASP6 Novel Fold targets we achieve similar accuracies (36.5% and 19.7%). This favourably compares with the best automated predictors at CASP6. CONCLUSION: Our final system for contact map prediction achieves state-of-the-art performances, and may provide valuable constraints for improved ab initio prediction of protein structures. A suite of predictors of structural features, including the PE, and PE-based contact maps, is available at http://distill.ucd.ie.

Algorithms↗
PubMed2006

DNA molecule classification using feature primitives.

BACKGROUND: We present a novel strategy for classification of DNA molecules using measurements from an alpha-Hemolysin channel detector. The proposed approach provides excellent classification performance for five different DNA hairpins that differ in only one base-pair. For multi-class DNA classification problems, practitioners usually adopt approaches that use decision trees consisting of binary classifiers. Finding the best tree topology requires exploring all possible tree topologies and is computationally prohibitive. We propose a computational framework based on feature primitives that eliminates the need of a decision tree of binary classifiers. In the first phase, we generate a pool of weak features from nanopore blockade current measurements by using HMM analysis, principal component analysis and various wavelet filters. In the next phase, feature selection is performed using AdaBoost. AdaBoost provides an ensemble of weak learners of various types learned from feature primitives. RESULTS AND CONCLUSION: We show that our technique, despite its inherent simplicity, provides a performance comparable to recent multi-class DNA molecule classification results. Unlike the approach presented by Winters-Hilt et al., where weaker data is dropped to obtain better classification, the proposed approach provides comparable classification accuracy without any need for rejection of weak data. A weakness of this approach, on the other hand, is the very "hands-on" tuning and feature selection that is required to obtain good generalization. Simply put, this method obtains a more informed set of features and provides better results for that reason. The strength of this approach appears to be in its ability to identify strong features, an area where further results are actively being sought.

Computational Biology↗
PubMed2005

Activity of striatal neurons reflects dynamic encoding and recoding of procedural memories.

Learning to perform a behavioural procedure as a well-ingrained habit requires extensive repetition of the behavioural sequence, and learning not to perform such behaviours is notoriously difficult. Yet regaining a habit can occur quickly, with even one or a few exposures to cues previously triggering the behaviour. To identify neural mechanisms that might underlie such learning dynamics, we made long-term recordings from multiple neurons in the sensorimotor striatum, a basal ganglia structure implicated in habit formation, in rats successively trained on a reward-based procedural task, given extinction training and then given reacquisition training. The spike activity of striatal output neurons, nodal points in cortico-basal ganglia circuits, changed markedly across multiple dimensions during each of these phases of learning. First, new patterns of task-related ensemble firing successively formed, reversed and then re-emerged. Second, task-irrelevant firing was suppressed, then rebounded, and then was suppressed again. These changing spike activity patterns were highly correlated with changes in behavioural performance. We propose that these changes in task representation in cortico-basal ganglia circuits represent neural equivalents of the explore-exploit behaviour characteristic of habit learning.

Acoustic Stimulation↗
PubMed2026

Investigating cross-organism prediction of prokaryotic essential proteins using unsupervised language model and ensemble strategy.

Cross-organism prediction of essential proteins is a critical task for drug discovery and microbial engineering, yet the generalizability of existing machine learning models across diverse species remains a significant challenge. In this study, we propose DeepPEP, a large language model-based framework designed to reliably transfer essential protein annotations between distantly related organisms. Utilizing 66 curated prokaryotic datasets, we systematically evaluated DeepPEP's cross-organism performance under various conditions. Initial pairwise predictions revealed a correlation between performance and evolutionary distance; however, further investigation demonstrated that integrating training data from multiple organisms yields superior predictive power. In a benchmark scenario designed to simulate real-world applications, DeepPEP outperformed the state-of-the-art tool Geptop 2.0, showcasing a robust ability to identify species-specific essential proteins. Finally, a case study on novel genomes confirmed the model's practical effectiveness. Our results suggest that DeepPEP is a powerful strategy for prokaryotic essential protein prediction, and the rigorous evaluation framework established in this study provides a new benchmark for the field.

Large Language Models↗
PubMed1975

"Stimulus generalization" between differentiated visual, auditory, and central stimuli.

Cats were trained to discriminate between two different repetition rates of flicker and of click. Both approach-approach and avoidance-avoidance discriminations were used. After substantial overtraining, transfer of frequency discrimination was initiated to stimulation of the reticular formation using bursts of electrical pulses at the same two repetition rates. Significant levels of discriminated performance were obtained in all cats very quickly, indicating good cross-modal transfer between the peripheral discriminanda and the central stimuli. The literature on stimulus generalization and cross-modal transfer is reviewed and the findings of this experiment are discussed in that context. Certain conditions are defined which, if satisfied, justify the interpretation that stimulus generalization or rapid cross-modal transfer indicate that facilitation of subsequent tasks in a training sequence can be attributed to mediation by a specific neuronal mechanism established by training on a previous task. The present experiment was designed in view of such criteria. The evidence of good cross-modal transfer is interpreted to mean that brain mechanisms storing memories about discriminations between visual or auditory stimuli with different repetition rates can be effectively activated by gross electrical stimuli at the same repetition rates. Conflict trials were then carried out in which flicker or click at either frequency was contradicted by concurrent RF stimuli at the other frequency. As the current level of RF stimuli was parametrically increased, it was found that the central stimuli achieved almost complete control over the behavioral outcome in most cases. Concurrent transfer of training, using a counterbalanced training sequence, was then carried out to stimulation of the visual cortex, lateral geniculate, medial geniculate, and the intralaminar nuclei of the thalamus. In each case, rapid transfer was displayed by at least one animal. Once performance to brain stimulation at a given repition rate was established, little change was observed when the fine structure of the stimulus was altered by changing parameters of the stimulus burst. These findings are interpreted as providing support for a statistical theory of memory, since they constitute evidence that previously learned discriminative behavior can readily be elicited by compelling large ensembles of neurons in various brain regions to discharge with particular temporal patterns. It is difficult to reconcile these results with theories which postulate that learning establishes new synaptic pathways in which discharge must occur for memories to be retrieved.

Animals↗
PubMedDate not supplied

[Physical models of neural networks].

The state of art in computer modelling of neural networks with associative memory is reviewed. The available experimental data are considered on learning and memory of small neural systems, on isolated synapses and on molecular level. Computer simulations demonstrate that realistic models of neural ensembles exhibit properties which can be interpreted as image recognition, categorization, learning, prototype forming, etc. A bilayer model of associative neural network is proposed. One layer corresponds to the short-term memory, the other one to the long-term memory. Patterns are stored in terms of the synaptic strength matrix. We have studied the relaxational dynamics of neurons firing and suppression within the short-term memory layer under the influence of the long-term memory layer. The interaction among the layers has found to create a number of novel stable states which are not the learning patterns. These synthetic patterns may consist of elements belonging to different non-intersecting learning patterns. Within the framework of a hypothesis of selective and definite coding of images in brain one can interpret the observed effect as the "ideaś generating" process.

Animals↗
PubMed2001

Enhanced auditory reversal learning by genetic activation of protein kinase C in small groups of rat hippocampal neurons.

The hippocampus has a central role in specific types of learning, but there is only limited evidence identifying the requisite molecular changes in ensembles of hippocampal neurons. To investigate the role of protein kinase C (PKC) pathways in hippocampal mediated learning, a constitutively active, catalytic domain of rat PKC betaII was delivered into hippocampal dentate granule neurons using a Herpes Simplex Virus (HSV-1) vector. This PKC causes a long-lasting, activation-dependent increase in neurotransmitter release from cultured cells. Activation of PKC pathways in a small percentage (< or =0.26%) of dentate granule neurons was sufficient to enhance rat auditory discrimination reversal learning. The affected neurons altered hippocampal physiology as revealed by elevated NMDA receptor densities in specific hippocampal areas. Thus, these results directly suggest that activation of PKC pathways in a specific hippocampal area alters rat auditory discrimination reversal learning. Because each rat may contain a unique pattern of affected neurons, there appears to be considerable flexibility and/or redundancy in the groups of neurons that can modify learning.

Animals↗
PubMed2026

Linking MRI radiomics to transcriptomics-based radiosensitivity in lower-grade glioma: A radiogenomic framework.

BACKGROUND: RSI is a transcriptomics-based biomarker associated with radiotherapy outcomes, but its clinical application is constrained by the requirement for tumor tissue and RNA sequencing. This study investigates whether MRI-derived radiomic features can reflect RSI-defined intrinsic radiosensitivity in lower-grade glioma.This addresses a critical gap arising from the limited availability of matched imaging and genomic data in routine clinical practice. METHODS: MRI-derived radiomic features were extracted from FLAIR images of lower-grade glioma patients obtained from TCIA and matched with transcriptomic data from TCGA. A total of 107 patients with both MRI and RNA sequencing data were included in the radiogenomic analysis. Radiomic features were ranked using a Borda-based ensemble feature selection strategy. Five supervised machine-learning classifiers were trained to predict RSI-based radiosensitivity classification, and model interpretability was assessed using SHAP within radiogenomic framework. RESULTS: Classification performance increased with feature number and stabilized at compact subset of 13 radiomic features. Logistic regression showed stable performance with an AUC of 0.82 (95&#xa0;% CI: 0.71-0.93). SHAP analysis indicated that heterogeneity-related texture features were dominant contributors to model predictions, with many associated with the RR phenotype, while others were linked to the RS phenotype. CONCLUSION: An MRI-based radiomic signature enables non-invasive prediction of RSI-defined radiosensitivity in lower-grade glioma. Rather than offering an immediately deployable clinical tool, this study establishes a proof-of-concept radiogenomic framework demonstrating that intrinsic radiosensitivity, traditionally assessed through invasive molecular assays, can be approximated using quantitative imaging features. These findings highlight the potential of imaging-based radiosensitivity assessment and provide a foundation for future radiogenomic investigations.

Lower-grade glioma↗
PubMed2026

Ensemble DNA methylation clock demonstrates Immune-metabolic aging signatures associated with mortality.

Aging is a multifactorial process that is best described in terms of the progressive acquisition of multiple layers of phenotypic changes, such as epigenetic modifications, inflammation, and metabolic dysregulation. DNA methylation clocks have been extensively used to construct epigenetic clocks based on the DNAm profiles that can be used to estimate biological age and predict age-associated outcomes. Nevertheless, the vast majority of clocks constructed so far have been based on linear models, which are unlikely to fully account for the heterogeneity and non-linearity of survival-related DNAm signatures. In this work, we constructed a heterogeneous stacked ensemble survival model based on DNAm data obtained from the Framingham Heart Study. We first identified 190 CpG loci using elastic net Cox regression and subsequently constructed a survival prediction model based on the fusion of five complementary survival models by means of a neural network meta-learner. The prediction power of the survival model was evaluated in an external validation cohort, where we observed strong performance for predicting all-cause mortality that significantly exceeded PhenoAge and was statistically comparable to GrimAge. These performance estimates were derived in cohorts of European ancestry and externally validated in postmenopausal women aged 50-79 years, and should therefore be interpreted as applicable only to demographically similar populations.

Humans↗
PubMed2006

Brain-computer interface technology as a tool to augment plasticity and outcomes for neurological rehabilitation.

Brain-computer interfaces (BCIs) are a rehabilitation tool for tetraplegic patients that aim to improve quality of life by augmenting communication, control of the environment, and self-care. The neurobiology of both rehabilitation and BCI control depends upon learning to modify the efficacy of spared neural ensembles that represent movement, sensation and cognition through progressive practice with feedback and reward. To serve patients, BCI systems must become safe, reliable, cosmetically acceptable, quickly mastered with minimal ongoing technical support, and highly accurate even in the face of mental distractions and the uncontrolled environment beyond a laboratory. BCI technologies may raise ethical concerns if their availability affects the decisions of patients who become locked-in with brain stem stroke or amyotrophic lateral sclerosis to be sustained with ventilator support. If BCI technology becomes flexible and affordable, volitional control of cortical signals could be employed for the rehabilitation of motor and cognitive impairments in hemiplegic or paraplegic patients by offering on-line feedback about cortical activity associated with mental practice, motor intention, and other neural recruitment strategies during progressive task-oriented practice. Clinical trials with measures of quality of life will be necessary to demonstrate the value of near-term and future BCI applications.

Brain↗
PubMed1999

Building neural representations of habits.

Memories for habits and skills ("implicit or procedural memory") and memories for facts ("explicit or episodic memory") are built up in different brain systems and are vulnerable to different neurodegenerative disorders in humans. So that the striatum-based mechanisms underlying habit formation could be studied, chronic recordings from ensembles of striatal neurons were made with multiple tetrodes as rats learned a T-maze procedural task. Large and widely distributed changes in the neuronal activity patterns occurred in the sensorimotor striatum during behavioral acquisition, culminating in task-related activity emphasizing the beginning and end of the automatized procedure. The new ensemble patterns remained stable during weeks of subsequent performance of the same task. These results suggest that the encoding of action in the sensorimotor striatum undergoes dynamic reorganization as habit learning proceeds.

Action Potentials↗
PubMed1994

An electrophysiological correlate of learning in motion perception.

We investigated learning in a motion-detection task using both psychophysical and neurophysiological methods in normal humans. A total of 20 naive observers had to discriminate between a small motion to the left versus to the right (jump displacement) or between a motion upward versus downward. Their performance improved significantly within less than 30 min in discriminating between directions in the psychophysical jump-displacement task. The improvement of performance with practice was very specific and did not transfer to the same stimulus rotated by 90 degrees. After training for the same task, multichannel evoked-potential recordings changed significantly in component latency and in the distribution of field potentials. This indicates that neuronal ensembles rather than single cells are involved in perceptual learning. Significant differences between the potential distributions occur for potentials at latencies of less than 100 ms over the occipital pole, suggesting an involvement of and plasticity in the primary visual cortex of human adults.

Adult↗
PubMed1987

Two-muscle coordination versus natural treadmill locomotion.

When a single-muscle learned behavior was superimposed upon natural human treadmill locomotion, in previous work, it operated as a self-contained behavioral unit. The new behavior altered some features, however, of ongoing stepping patterns. These findings prompted broader consideration of how individual muscle actions combine to form large, patterned ensembles. Accordingly, the present experiment constructed a larger, double-muscle learned behavior to see if it would compete with natural treadmill locomotion or combine with it harmoniously. A demanding requirement was made for in-phase bilateral EMG and contraction by rectus femoris (RF), in opposition to its natural out-of-phase interlimb pattern. EMG bursts were controlled, through computer-assisted operant conditioning, by a flash shortly after left heel strike. The new, double ensemble was conditioned rapidly, within 1-6 days, for all four adults. Harmonious stepping continued, for the most part, with little alteration in step cycle timings. Leg positioning was modified appreciably, however, pointing to complex neural mechanisms. The evidence argued that operant conditioning can construct fine-grained behaviors and also participate powerfully across the full range of single- and interlimb coordination.

Adult↗
PubMed1995

The assessment of LH surge for predicting ovulation time using clinical, hormonal, and ultrasonic indices in infertile women with an ensemble of neural networks.

An ensemble of independently trained neural networks (NN) is proposed for the assessment of luteinizing hormone (LH) surge for predicting ovulation time in infertile but ovulating women. The proposed ensemble involves a number of parallel NN modules. Each pair of the NNs learn specific data that are previously collected for monitoring timing function of LH estradiol (pg ml-1), and follicle diameter (mm) are used to train NN pairs to approximate the function of the LH values. A reasonable and accurate estimation places ovulation approximately 10-12 h after the LH peak. The double-valued (bi-phasic) regions of training data are separated into two single-valued (bi-phasic) regions of training data are separated into two single-valued parts (not exactly preovulatory, postovulatory division) that can be learned by each module of the NN pair. During testing, after the initial decision to have single-valued sides, the assessment is obtained by a linear opinion pool (consensus rule) using the decisions of NNs on the corresponding side without waiting. The network ensemble has various desirable properties: high assessment accuracy of a double-valued multisource data, minimized learning and recall times, and a parallel structure. The ovulation time can be predicted through the assessment of LH peak with a better precision and fewer number of tests.

Diagnosis, Computer-Assisted↗
PubMed1997

Marr's theory of the neocortex as a self-organizing neural network.

Marr's proposal for the functioning of the neocortex (Marr, 1970) is the least known of his various theories for specific neural circuitries. He suggested that the neocortex learns by self-organization to extract the structure from the patterns of activity incident upon it. He proposed a feedforward neural network in which the connections to the output cells (identified with the pyramidal cells of the neocortex) are modified by a mechanism of competitive learning. It was intended that each output cell comes to be selective for the input patterns from a different class and is able to respond to new patterns from the same class that have not been seen before. The learning rule that Marr proposed was underspecified, but a logical extension of the basic idea results in a synaptic learning rule in which the total amount of synaptic strength of the connections from each input ("presynaptic") cell is kept at a constant level. In contrast, conventional competitive learning involves rules of the "postsynaptic" type. The network learns by exploiting the structure that Marr assumed to exist within the ensemble of input patterns. For this case, analysis is possible that extends that carried out by Marr, which was restricted to the binary classification task. This analysis is presented here, together with results from computer simulations of different types of competitive learning mechanisms. The presynaptic mechanism is best known in the computational neuroscience literature. In neural network applications, it may be a more suitable mechanism of competitive learning than those normally considered.

Algorithms↗
PubMed2026

An Integrated Machine Learning and Genomic Framework for Precise Detection of Gastric Cancer.

This study presents a novel integrative approach for the analysis of high-dimensional gene expression data, leveraging the complementary strengths of unsupervised clustering and supervised classification. Using K-means clustering, the data set is stratified into three distinct clusters, revealing intrinsic biological patterns and relationships. The resulting cluster assignments are subsequently used as pseudolabels to train machine learning models, including support vector machines, random forest, and a stacking ensemble classifier. To validate and enhance the robustness of clustering, complementary methods, such as hierarchical clustering and density-based spatial clustering of applications with noise (DBSCAN), are used, with results visualized through principal component analysis-driven dimensionality reduction. The high predictive accuracy achieved by the classifiers underlines the separability and reliability of the identified clusters. Furthermore, feature importance analysis highlighted key genetic determinants within each cluster, offering actionable insights into potential biomarkers and critical genomic features. This framework bridges the gap between exploratory unsupervised learning and predictive supervised modeling, providing a scalable and interpretable method for analyzing complex genomic data sets. Its applicability extends to biomarker discovery, patient stratification, and other precision medicine applications, emphasizing its utility in advancing genomic research and clinical practice.

Humans↗
PubMed1995

Learning mechanisms in the temporal lobe visual cortex.

Neurophysiological experiments are described which show that neurons form ensemble encoded representations of stimuli such as faces which are relatively invariant with respect to size, contrast, spatial frequency, translation, and view. It is shown that new representations of objects can be formed with less than 5 s of visual experience with those objects. Mechanisms by which the brain could perform this invariant recognition, and learn the representations required for recognition, are described. A neural network simulation of these mechanisms for learning invariant representations is then described. The model uses a multistage feed-forward architecture, and is able to learn invariant representations of objects including faces by use of a Hebbian synaptic modification rule which incorporates a short memory trace (0.5 s) of preceding activity. This trace rule enables the network to learn the properties of objects which are spatio-temporally invariant over this time scale.

Animals↗