Search PubMed⌕ Search

SEARCH · Search PubMed

Results for “sparse 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 235 records · Page 13Linked to original sources
PubMed2007

Lexical restructuring in the absence of literacy.

Vocabulary growth was suggested to prompt the implementation of increasingly finer-grained lexical representations of spoken words in children (e.g., [Metsala, J. L., & Walley, A. C. (1998). Spoken vocabulary growth and the segmental restructuring of lexical representations: precursors to phonemic awareness and early reading ability. In J. L. Metsala & L. C. Ehri (Eds.), Word recognition in beginning literacy (pp. 89-120). Hillsdale, NJ: Erlbaum.]). Although literacy was not explicitly mentioned in this lexical restructuring hypothesis, the process of learning to read and spell might also have a significant impact on the specification of lexical representations (e.g., [Carroll, J. M., & Snowling, M. J. (2001). The effects of global similarity between stimuli on children's judgments of rime and alliteration. Applied Psycholinguistics, 22, 327-342.]; [Goswami, U. (2000). Phonological representations, reading development and dyslexia: Towards a cross-linguistic theoretical framework. Dyslexia, 6, 133-151.]). This is what we checked in the present study. We manipulated word frequency and neighborhood density in a gating task (Experiment 1) and a word-identification-in-noise task (Experiment 2) presented to Portuguese literate and illiterate adults. Ex-illiterates were also tested in Experiment 2 in order to disentangle the effects of vocabulary size and literacy. There was an interaction between word frequency and neighborhood density, which was similar in the three groups. These did not differ even for the words that are supposed to undergo lexical restructuring the latest (low frequency words from sparse neighborhoods). Thus, segmental lexical representations seem to develop independently of literacy. While segmental restructuring is not affected by literacy, it constrains the development of phoneme awareness as shown by the fact that, in Experiment 3, neighborhood density modulated the phoneme deletion performance of both illiterates and ex-illiterates.

Adolescent↗
PubMed2003

Protein family classification using sparse markov transducers.

We present a method for classifying proteins into families based on short subsequences of amino acids using a new probabilistic model called sparse Markov transducers (SMT). We classify a protein by estimating probability distributions over subsequences of amino acids from the protein. Sparse Markov transducers, similar to probabilistic suffix trees, estimate a probability distribution conditioned on an input sequence. SMTs generalize probabilistic suffix trees by allowing for wild-cards in the conditioning sequences. Since substitutions of amino acids are common in protein families, incorporating wild-cards into the model significantly improves classification performance. We present two models for building protein family classifiers using SMTs. As protein databases become larger, data driven learning algorithms for probabilistic models such as SMTs will require vast amounts of memory. We therefore describe and use efficient data structures to improve the memory usage of SMTs. We evaluate SMTs by building protein family classifiers using the Pfam and SCOP databases and compare our results to previously published results and state-of-the-art protein homology detection methods. SMTs outperform previous probabilistic suffix tree methods and under certain conditions perform comparably to state-of-the-art protein homology methods.

Algorithms↗
PubMed2006

On the consistency of Bayesian variable selection for high dimensional binary regression and classification.

Modern data mining and bioinformatics have presented an important playground for statistical learning techniques, where the number of input variables is possibly much larger than the sample size of the training data. In supervised learning, logistic regression or probit regression can be used to model a binary output and form perceptron classification rules based on Bayesian inference. We use a prior to select a limited number of candidate variables to enter the model, applying a popular method with selection indicators. We show that this approach can induce posterior estimates of the regression functions that are consistently estimating the truth, if the true regression model is sparse in the sense that the aggregated size of the regression coefficients are bounded. The estimated regression functions therefore can also produce consistent classifiers that are asymptotically optimal for predicting future binary outputs. These provide theoretical justifications for some recent empirical successes in microarray data analysis.

Bayes Theorem↗
PubMed2025

PATTY corrects open-chromatin bias for improved bulk and single-cell CUT&Tag profiling.

Precise profiling of epigenomes is essential for better understanding chromatin biology and gene regulation. Cleavage Under Targets & Tagmentation (CUT&Tag) is an efficient epigenomic profiling technique that can be performed on a low number of cells and at the single-cell level. With its growing adoption, CUT&Tag datasets spanning diverse biological systems are rapidly accumulating in the field. CUT&Tag assays use the hyperactive transposase Tn5 for DNA tagmentation. Tn5's preference toward accessible chromatin alters CUT&Tag sequence read distributions in the genome and introduces open-chromatin bias that can confound downstream analysis, an issue more substantial in sparse single-cell data. We show that open-chromatin bias extensively exists in published CUT&Tag datasets, including those generated with recently optimized high-salt protocols. To address this challenge, we present PATTY (Propensity Analyzer for Tn5 Transposase Yielded bias), a comprehensive computational method that corrects open-chromatin bias in CUT&Tag data by leveraging accompanying ATAC-seq. By integrating transcriptomic and epigenomic data using machine learning and integrative modeling, we demonstrate that PATTY enables accurate and robust detection of occupancy sites for both active and repressive histone modifications, including H3K27ac, H3K27me3, and H3K9me3, with experimental validation. We further develop a single-cell CUT&Tag analysis framework built on PATTY and show improved cell clustering when using bias-corrected single-cell CUT&Tag data compared to using uncorrected data. Beyond CUT&Tag, PATTY sets a foundation for further development of bias correction methods for improving data analysis for all Tn5-based high-throughput assays.

Journal Article↗
PubMed2006

Gene selection in cancer classification using sparse logistic regression with Bayesian regularization.

MOTIVATION: Gene selection algorithms for cancer classification, based on the expression of a small number of biomarker genes, have been the subject of considerable research in recent years. Shevade and Keerthi propose a gene selection algorithm based on sparse logistic regression (SLogReg) incorporating a Laplace prior to promote sparsity in the model parameters, and provide a simple but efficient training procedure. The degree of sparsity obtained is determined by the value of a regularization parameter, which must be carefully tuned in order to optimize performance. This normally involves a model selection stage, based on a computationally intensive search for the minimizer of the cross-validation error. In this paper, we demonstrate that a simple Bayesian approach can be taken to eliminate this regularization parameter entirely, by integrating it out analytically using an uninformative Jeffrey's prior. The improved algorithm (BLogReg) is then typically two or three orders of magnitude faster than the original algorithm, as there is no longer a need for a model selection step. The BLogReg algorithm is also free from selection bias in performance estimation, a common pitfall in the application of machine learning algorithms in cancer classification. RESULTS: The SLogReg, BLogReg and Relevance Vector Machine (RVM) gene selection algorithms are evaluated over the well-studied colon cancer and leukaemia benchmark datasets. The leave-one-out estimates of the probability of test error and cross-entropy of the BLogReg and SLogReg algorithms are very similar, however the BlogReg algorithm is found to be considerably faster than the original SLogReg algorithm. Using nested cross-validation to avoid selection bias, performance estimation for SLogReg on the leukaemia dataset takes almost 48 h, whereas the corresponding result for BLogReg is obtained in only 1 min 24 s, making BLogReg by far the more practical algorithm. BLogReg also demonstrates better estimates of conditional probability than the RVM, which are of great importance in medical applications, with similar computational expense. AVAILABILITY: A MATLAB implementation of the sparse logistic regression algorithm with Bayesian regularization (BLogReg) is available from http://theoval.cmp.uea.ac.uk/~gcc/cbl/blogreg/

Algorithms↗
PubMedDate not supplied

State child mental health efforts to support youth in transition to adulthood.

The ability of state child mental health (MH) systems to facilitate the transition to adulthood of adolescents in their systems was studied by interviewing members of the Children, Youth and Families Division of the National Association of State Mental Health Program Directors (NASMHPD). Results demonstrated that transition services within the state child MH systems are sparse, nationally. Continuity of services as youth age into adulthood is hampered because of generally separate child and adult MH systems, each with separate policies defining who accesses those services, lack of clarity about procedures to access adult MH services, and lack of shared client planning between adult and child MH systems. These findings suggest that adolescents in state child MH systems have difficulty accessing services that will help them with the difficult task of learning to function as an adult. Public MH systems should examine their capacity to provide transition supports and make needed improvements.

Adolescent↗
PubMed2004

What can rural agencies do to address the additional costs of rural services? A typology of rural service innovation.

There is a national commitment to ensuring that, regardless of where patients live, they should be provided with an acceptable level of service in terms of quality, effectiveness and accessibility. Because of differences in the distributions of their populations, rural and urban areas present quite different challenges for the optimal design of health services and social care. However, this has not been fully acknowledged in the development of national policies to unify service standards. The problems of providing services in sparsely populated areas are not new. However, until the case for a rural premium in English health resource allocation is accepted, rural agencies must either tolerate lower levels of services (an option made difficult by the introduction of national service standards) or develop very different approaches to service delivery. To date, there has been little systematic knowledge about the extent of innovative rural practice, a paucity of evaluation of such initiatives and few opportunities to disseminate learning from one area to another. The present paper begins to address this deficit. Drawing upon a review of the formal literature and a comprehensive evaluation of projects developed within a rural Health Action Zone, it presents a typology of innovative responses at the health/social care interface. Examples of service innovations which fall into six broad categories are provided. These not only suggest possibilities for the transfer of good practice, but also the potential for future research.

Diffusion of Innovation↗
PubMed2005

Lexical frequency and neighborhood density effects on the recognition of native and Spanish-accented words by native English and Spanish listeners.

This study examined the effect of presumed mismatches between speech input and the phonological representations of English words by native speakers of English (NE) and Spanish (NS). The English test words, which were produced by a NE speaker and a NS speaker, varied orthogonally in lexical frequency and neighborhood density and were presented to NE listeners and to NS listeners who differed in English pronunciation proficiency. It was hypothesized that mismatches between phonological representations and speech input would impair word recognition, especially for items from dense lexical neighborhoods which are phonologically similar to many other words and require finer sound discrimination. Further, it was assumed that L2 phonological representations would change with L2 proficiency. The results showed the expected mismatch effect only for words from dense neighborhoods. For Spanish-accented stimuli, the NS groups recognized more words from dense neighborhoods than the NE group did. For native-produced stimuli, the low-proficiency NS group recognized fewer words than the other two groups. The-high proficiency NS participants' performance was as good as the NE group's for words from sparse neighborhoods, but not for words from dense neighborhoods. These results are discussed in relation to the development of phonological representations of L2 words. (200 words).

Adult↗
PubMed2006

Functional connectivity in fMRI: A modeling approach for estimation and for relating to local circuits.

Although progress has been made in relating neuronal events to changes in brain metabolism and blood flow, the interpretation of functional neuroimaging data in terms of the underlying brain circuits is still poorly understood. Computational modeling of connection patterns both among and within regions can be helpful in this interpretation. We present a neural network model of the ventral visual pathway and its relevant functional connections. This includes a new learning method that adjusts the magnitude of interregional connections in order to match experimental results of an arbitrary functional magnetic resonance imaging (fMRI) data set. We demonstrate that this method finds the appropriate connection strengths when trained on a model system with known, randomly chosen connection weights. We then use the method for examining fMRI results from a one-back matching task in human subjects, both healthy and those with schizophrenia. The results discovered by the learning method support previous findings of a disconnection between left temporal and frontal cortices in the group with schizophrenia and a concomitant increase of right-sided temporo-frontal connection strengths. We then demonstrate that the disconnection may be explained by reduced local recurrent circuitry in frontal cortex. This method extends currently available methods for estimating functional connectivity from human imaging data by including both local circuits and features of interregional connections, such as topography and sparseness, in addition to total connection strengths. Furthermore, our results suggest how fronto-temporal functional disconnection in schizophrenia can result from reduced local synaptic connections within frontal cortex rather than compromised interregional connections.

Brain↗
PubMed1999

Neuronal nitric oxide synthase expression in developing and adult human CNS.

Neuronal nitric oxide synthase (nNOS) is constitutively expressed by subpopulations of neurons in the CNS and is involved in neurotransmission, learning and memory, and neuronal injury. While the distribution of nNOS neurons has been characterized in the rodent CNS, the expression in human brain has not been well documented. We determined the expression of nNOS in second trimester human fetal and adult brain. In second trimester fetal brain, the nNOS neurons are concentrated in the developing cerebral cortex at the subplate zone and in layer VI, the striatum, and in certain brainstem nuclei. The nNOS neurons are sparsely distributed in the hippocampus, and virtually absent in the cerebellar cortex. The nNOS neurons in the subplate zone extend their processes radially, suggesting a developmental role, perhaps in guidance. The number and distribution of NADPH diaphorase-positive neurons corresponds to that of the nNOS neurons. While the distribution of nNOS neurons in the adult brain is similar to that found in fetal brain, the overall density is lower in the adult. The highest density of nNOS neurons is found in the striatum followed by the neocortex. A region-specific role for nNOS neurons in human brain and a potential developmental role for nNOS in the cerebral cortex are suggested by these data.

Adult↗
PubMed2002

The influence of phonological similarity neighborhoods on speech production.

The influence of phonological similarity neighborhoods on the speed and accuracy of speech production was investigated with speech-error elicitation and picture-naming tasks. The results from 2 speech-error elicitation techniques-the spoonerisms of laboratory induced predisposition technique (B. J. Baars, 1992; B. J. Baars & M. T. Motley, 1974; M. T. Motley & B. J. Baars, 1976) and tongue twisters-showed that more errors were elicited for words with few similar sounding words (i.e., a sparse neighborhood) than for words with many similar sounding words (i.e., a dense neighborhood). The results from 3 picture-naming tasks showed that words with sparse neighborhoods were also named more slowly than words with dense neighborhoods. These findings demonstrate that multiple word forms are activated simultaneously and influence the speed and accuracy of speech production. The implications of these findings for current models of speech production are discussed.

Attention↗
PubMed1994

Optimal Hebbian learning rules and the role of asymmetry.

We study the storage properties associated with generalized Hebbian learning rules which present four free parameters that allow for asymmetry. We also introduce two extra parameters in the post-synaptic potentials in order to further improve the critical capacity. Using signal-to-noise analysis, as well as computer simulations on an analog network, we discuss the performance of the rules for arbitrarily biased patterns and find that the critical storage capacity alpha c becomes maximal for a particular symmetric rule (alpha c diverges in the sparse coding limit). Departures from symmetry decrease alpha c but can increase the robustness of the model.

Artificial Intelligence↗
PubMedDate not supplied

Multiple views of the response of an ensemble of spectro-temporal features support concurrent classification of utterance, prosody, sex and speaker identity.

Models of auditory processing, particularly of speech, face many difficulties. These difficulties include variability among speakers, variability in speech rate and robustness to moderate distortions such as time compression. In contrast to the 'invariance of percept' (across different speakers, of different sexes, using different intonation, and so on) is the observation that we are sensitive to the identity, sex and intonation of the speaker. In previous work we have reported that a model based on ensembles of spectro-temporal feature detectors, derived from onset sensitive pre-processing of a limited class of stimuli, preserves significant information about the stimulus class. We have also shown that this is robust with respect to the exact choice of feature set, moderate time compression in the stimulus and speaker variation. Here we extend these results to show a) that by using a classifier based on a network of spiking neurons with spike-driven plasticity, the output of the ensemble constitutes an effective rate coding representation of complex sounds; and b) that the same set of spectro-temporal features concurrently preserve information about a range of qualitatively different classes into which the stimulus might fall. We show that it is possible for multiple views of the same pattern of responses to generate different percepts. This is consistent with suggestions that multiple parallel processes exist within the auditory 'what' pathway with attentional modulation enhancing the task-relevant classification type. We also show that the responses of the ensemble are sparse in the sense that a small number of features respond for each stimulus type. This has implications for the ensembles' ability to generalise, and to respond differentially to a wide variety of stimulus classes.

Acoustic Stimulation↗
PubMed2026

Beyond predictive performance: A systematic review and critical methodological appraisal of AI/ML and conventional modelling strategies in breast, colorectal, and pancreatic Cancer.

BACKGROUND: Predictive modelling for cancer risk, treatment-related complications, and survival is central to precision oncology. Conventional logistic regression (LR) and Cox proportional hazards (CoxPH) regression remain widely used but are limited when modelling nonlinear interactions, high-dimensional imaging features, and multimodal clinical-metabolic predictors. Artificial intelligence (AI) and machine learning (ML) methods offer expanded capability through automated feature extraction, ensemble learning, and flexible survival modelling, but the evidence on when AI/ML adds value over conventional models across cancer sites and predictive tasks remains fragmented. OBJECTIVE: To systematically evaluate the methodological performance, validation strategies, and translational limitations of AI/ML models compared with conventional statistical models in published predictive-modelling studies for breast, colorectal, or pancreatic cancer. METHODS: PubMed, Scopus, and Web of Science were searched for studies published between January 2019 and March 2025. Two reviewers independently conducted title-and-abstract screening, full-text eligibility assessment, and PROBAST risk-of-bias assessment. Sixty-five studies (n = 907,567 participants) were narratively synthesised by cancer site, predictive task, model family, comparator, validation strategy, predictor modality, and calibration or explainability reporting. RESULTS: The 65 studies comprised breast cancer (n = 35), colorectal cancer (n = 21), and pancreatic cancer (n = 9). AI/ML superiority over LR and CoxPH was task- and data-dependent. CNN- and U-Net-based models predominated in imaging and body-composition tasks, tree-based ensembles consistently outperformed LR for tabular perioperative complication prediction, and CoxPH remained competitive, and in the largest pancreatic risk study, superior to XGBoost (C-index 0.802 vs 0.723) in well-structured datasets. PROBAST analysis-domain risk was moderate in 54 of 65 studies (83%), driven by limited external validation, sparse calibration reporting (11/65), and few decision-curve analyses (7/65). CONCLUSION: AI/ML adds the most methodological value in imaging-derived feature extraction and nonlinear perioperative prediction, while conventional regression remains preferable in large, structured datasets with linear predictors. Clinical translation requires standardised body-composition definitions, external validation, calibration assessment, decision-curve analysis, and explainability, in line with TRIPOD+AI and CLAIM standards.

Humans↗
PubMed2005

Synaptic contacts between an identified type of ON cone bipolar cell and ganglion cells in the mouse retina.

We surveyed the potential contacts between an identified type of bipolar cell and retinal ganglion cells in the mouse. By crossing two existing mouse strains (line 357 and line GFP-M), we created a double transgenic strain in which GFP is expressed by all members of a single type of ON cone bipolar cell and a sparse, mixed population of retinal ganglion cells. The GFP-expressing bipolar cells appear to be those termed CB4a of Pignatelli & Strettoi [(2004) J. Comp. Neurol., 476, 254-266] and type 7 of Ghosh et al. [(2004) J. Comp. Neurol., 469, 70-82 and J. Comp. Neurol., 476, 202-203]. The labelled ganglion cells include examples of most or all types of ganglion cells present in the mouse. By studying the juxtaposition of their processes in three dimensions, we could learn which ganglion cell types are potential synaptic targets of the line 357 bipolar cell. Of 12 ganglion cell types observed, 10 types could be definitively ruled out as major synaptic targets of the line 357 bipolar cells. One type of monostratified ganglion cell and one bistratified cell tightly cofasciculate with axon terminals of the line 357 bipolar cells. Double labelling for kinesin II demonstrates colocalization of bipolar cell ribbons at the sites of contact between these two types of ganglion cell and the line 357 bipolar cells.

Animals↗
PubMed1995

The sparse fur mouse as a model for gene therapy in ornithine carbamoyltransferase deficiency.

The sparse fur (spf/Y) mouse was evaluated as a model for studying gene therapy in ornithine carbamoyltransferase deficiency (OCTD), the most common inborn error of urea synthesis. Previous studies have defined a number of biochemical characteristics of this animal model that are analogous to the human disease: OCTD in liver, elevated ammonium and glutamine, low citrulline and arginine in plasma, elevated urinary orotic acid excretion, neurochemical alterations and responsiveness to alternative pathway therapy. In this study, metabolic flux, survival, behavior and learning of these animals were examined in preparation for a trial of gene therapy. We found that, as has been previously reported, OCT activity in liver ranged from 10 to 20% of control. Yet, stable isotope studies using 15N ammonium chloride to follow ureagenesis in vivo showed 55% of normal urea synthetic capacity. This suggests that partial correction with gene therapy may be sufficient to normalize urea synthesis. Although it has been suggested that liver OCTD and its consequent metabolic effects normalize without treatment by adulthood in the spf/Y mouse, we did not find this to be the case. We documented that the spf/Y mouse had a markedly decreased lifespan (< 10% of normal) and remained runted throughout life. In terms of behavior, the spf/Y mice had evidence of decreased learning in a passive avoidance task that was not attributable to alterations in activity. These clearly definable metabolic and behavioral abnormalities suggest that the spf/Y mouse should prove a useful model for studying the efficacy of gene therapy in OCTD.

Amino Acid Metabolism, Inborn Errors↗
PubMed1991

Forebrain ischemia induces selective behavioral impairments associated with hippocampal injury in rats.

Two groups of rats were tested on a variety of motor and cognitive tasks after either 10 minutes of two-vessel occlusion forebrain ischemia (n = 8) or sham operative procedures (n = 6). Histological injury was absent in the sham-operated group. In the ischemic group, hippocampal injury was restricted to field CA1, while damage in the neocortex and caudoputamen was sparse. Motor tests performed on postoperative days 18 and 28 revealed no significant differences between the ischemic and sham-operated groups. Retention performance of a radial maze discrimination task was impaired, with a significant but transient increase in both working and reference memory errors. Passive avoidance acquisition and retention were not significantly affected, although conclusions concerning the utility of this task must be reserved because of variability in the behavior of the sham-operated rats. Morris maze spatial navigation (place learning) and open-field activity were insensitive to treatment group. These functional results are consistent with the observed histological injury and what is known about hippocampal injury and behavior, and they provide further guidance for the development of neurological assays appropriate for discriminating outcome from forebrain ischemia in rats.

Animals↗
PubMed2025

Integrative multi-omics profiling deciphers tumor microenvironment heterogeneity and immunotherapy vulnerabilities in lung neuroendocrine carcinomas.

INTRODUCTION: Lung neuroendocrine carcinomas (Lu-NECs) are rare, highly aggressive lung tumors with poor prognosis and limited therapeutic options. Understanding the tumor immune microenvironment (TIME) is crucial towards personalized therapeutic strategies. OBJECTIVES: This study aims to systematically characterize the heterogeneity and complexity of the TIME in Lu-NECs by integrating proteomic, transcriptomic, and genomic data. METHODS: We performed comprehensive immune-proteomic profiling of 76 Lu-NECs across diverse histopathological subtypes to elucidate intra-tumoral TIME heterogeneity at the proteomic level. Validation was conducted in multiple independent cohorts, including 112 Lu-NECs using immunohistochemistry, 147 Lu-NECs, and 17 small cell lung carcinoma samples using transcriptomics. We integrated proteomic, transcriptomic, genomic, and clinical data to assess molecular, immunological, and clinical features, as well as therapeutic vulnerabilities across different immune subtypes. RESULTS: We delineated the immuno-proteomic landscape of Lu-NECs and identified two major immuno-proteomic clusters with distinct immunological, molecular, and clinical characteristics. IPC1 was characterized by high immune cell infiltration, while IPC2 exhibited sparse immune cell presence. Genomic analysis revealed distinct mutational patterns, with IPC1 showing a higher incidence of APOBEC-associated mutation signatures and IPC2 being enriched for mutations associated with defective DNA mismatch repair and tobacco-related mutagens. Functional analyses indicated that IPC1 was related to immune and oncogenic signaling activity, whereas IPC2 was associated with cancer stemness and proliferation-related features. Furthermore, IPC1 and IPC2 demonstrated histological subtype-specific clinical benefits from postoperative chemotherapy. Finally, we developed a machine learning model (iPROM) to predict Lu-NECs immune classification and improve risk stratification, which was validated across multiple independent cohorts. CONCLUSIONS: This study advances the understanding of the tumor immune microenvironment in Lu-NECs through multi-omics characterization and highlights potential personalized therapeutic vulnerabilities tailored to the specific immune landscapes of Lu-NECs.

Humans↗