Spatial position constraint for unsupervised learning of speech representations.
Ontology highlight
ABSTRACT:
SUBMITTER: Humayun MA
PROVIDER: S-EPMC8323719 | biostudies-literature |
REPOSITORIES: biostudies-literature
Similar Datasets
Project description:Infants rapidly learn the sound categories of their native language, even though they do not receive explicit or focused training. Recent research suggests that this learning is due to infants' sensitivity to the distribution of speech sounds and that infant-directed speech contains the distributional information needed to form native-language vowel categories. An algorithm, based on Expectation-Maximization, is presented here for learning the categories from a sequence of vowel tokens without (i) receiving any category information with each vowel token, (ii) knowing in advance the number of categories to learn, or (iii) having access to the entire data ensemble. When exposed to vowel tokens drawn from either English or Japanese infant-directed speech, the algorithm successfully discovered the language-specific vowel categories (/I, i, epsilon, e/ for English, /I, i, e, e/ for Japanese). A nonparametric version of the algorithm, closely related to neural network models based on topographic representation and competitive Hebbian learning, also was able to discover the vowel categories, albeit somewhat less reliably. These results reinforce the proposal that native-language speech categories are acquired through distributional learning and that such learning may be instantiated in a biologically plausible manner.
| S-EPMC1934922 | biostudies-literature
Project description:The paper deals with the analysis of spatial distribution of Swiss population using fractal concepts and unsupervised learning algorithms. The research methodology is based on the development of a high dimensional feature space by calculating local growth curves, widely used in fractal dimension estimation and on the application of clustering algorithms in order to reveal the patterns of spatial population distribution. The notion "unsupervised" also means, that only some general criteria-density, dimensionality, homogeneity, are used to construct an input feature space, without adding any supervised/expert knowledge. The approach is very powerful and provides a comprehensive local information about density and homogeneity/fractality of spatially distributed point patterns.
| S-EPMC7877650 | biostudies-literature
Project description:Cellular microscopy images contain rich insights about biology. To extract this information, researchers use features, or measurements of the patterns of interest in the images. Here, we introduce a convolutional neural network (CNN) to automatically design features for fluorescence microscopy. We use a self-supervised method to learn feature representations of single cells in microscopy images without labelled training data. We train CNNs on a simple task that leverages the inherent structure of microscopy images and controls for variation in cell morphology and imaging: given one cell from an image, the CNN is asked to predict the fluorescence pattern in a second different cell from the same image. We show that our method learns high-quality features that describe protein expression patterns in single cells both yeast and human microscopy datasets. Moreover, we demonstrate that our features are useful for exploratory biological analysis, by capturing high-resolution cellular components in a proteome-wide cluster analysis of human proteins, and by quantifying multi-localized proteins and single-cell variability. We believe paired cell inpainting is a generalizable method to obtain feature representations of single cells in multichannel microscopy images.
| S-EPMC6743779 | biostudies-literature
Project description:Learning non-native phonetic categories in adulthood is an exceptionally challenging task, characterized by large inter-individual differences in learning speed and outcomes. The neurobiological mechanisms underlying the inter-individual differences in the learning efficacy are not fully understood. Here we examined the extent to which training-induced neural representations of non-native Mandarin tone categories in English listeners (n = 53) are increasingly similar to those of the native listeners (n = 33) who acquired these categories early in infancy. We particularly assessed whether the neural similarities in representational structure between non-native learners and native listeners are robust neuromarkers of inter-individual differences in learning success. Using inter-subject neural representational similarity (IS-NRS) analysis and predictive modeling on two functional magnetic resonance imaging (fMRI) datasets, we examined the neural representational mechanisms underlying speech category learning success. Learners' neural representations that were significantly similar to the native listeners emerged in brain regions mediating speech perception following training; the extent of the emerging neural similarities with native listeners significantly predicted the learning speed and outcome in learners. The predictive power of IS-NRS outperformed models with other neural representational measures. Furthermore, neural representations underlying successful learning are multidimensional but cost-efficient in nature. The degree of the emergent native-similar neural representations was closely related to the robust neural sensitivity to feedback in the frontostriatal network. These findings provide important insights on experience-dependent representational neuroplasticity underlying successful speech learning in adulthood and could be leveraged in designing individualized feedback-based training paradigms that maximize learning efficiency.
| S-EPMC8345815 | biostudies-literature
Project description:BackgroundThe rise of spatial transcriptomics technologies is leading to new insights about how gene regulation happens in a spatial context. Determining which genes are expressed in similar spatial patterns can reveal gene regulatory relationships across cell types in a tissue. However, many current analysis methods do not take full advantage of the spatial organization of the data, instead treating pixels as independent features. Here, we present CoSTA: a novel approach to learn spatial similarities between gene expression matrices via convolutional neural network (ConvNet) clustering.ResultsBy analyzing simulated and previously published spatial transcriptomics data, we demonstrate that CoSTA learns spatial relationships between genes in a way that emphasizes broader spatial patterns rather than pixel-level correlation. CoSTA provides a quantitative measure of expression pattern similarity between each pair of genes rather than only classifying genes into categories. We find that CoSTA identifies narrower, but biologically relevant, sets of significantly related genes as compared to other approaches.ConclusionsThe deep learning CoSTA approach provides a different angle to spatial transcriptomics analysis by focusing on the shape of expression patterns, using more information about the positions of neighboring pixels than would an overlap or pixel correlation approach. CoSTA can be applied to any spatial transcriptomics data represented in matrix form and may have future applications to datasets such as histology in which images of different genes are from similar but not identical biological sections.
| S-EPMC8351440 | biostudies-literature
Project description:An emerging view suggests that spatial position is an integral component of working memory (WM), such that non-spatial features are bound to locations regardless of whether space is relevant [1, 2]. For instance, past work has shown that stimulus position is spontaneously remembered when non-spatial features are stored. Item recognition is enhanced when memoranda appear at the same location where they were encoded [3-5], and accessing non-spatial information elicits shifts of spatial attention to the original position of the stimulus [6, 7]. However, these findings do not establish that a persistent, active representation of stimulus position is maintained in WM because similar effects have also been documented following storage in long-term memory [8, 9]. Here we show that the spatial position of the memorandum is actively coded by persistent neural activity during a non-spatial WM task. We used a spatial encoding model in conjunction with electroencephalogram (EEG) measurements of oscillatory alpha-band (8-12 Hz) activity to track active representations of spatial position. The position of the stimulus varied trial to trial but was wholly irrelevant to the tasks. We nevertheless observed active neural representations of the original stimulus position that persisted throughout the retention interval. Further experiments established that these spatial representations are dependent on the volitional storage of non-spatial features rather than being a lingering effect of sensory energy or initial encoding demands. These findings provide strong evidence that online spatial representations are spontaneously maintained in WM-regardless of task relevance-during the storage of non-spatial features.
| S-EPMC5661984 | biostudies-literature
Project description:Conceptual knowledge is fundamental to human cognition. Yet, the extent to which it is influenced by language is unclear. Studies of semantic processing show that similar neural patterns are evoked by the same concepts presented in different modalities (e.g., spoken words and pictures or text) [1-3]. This suggests that conceptual representations are "modality independent." However, an alternative possibility is that the similarity reflects retrieval of common spoken language representations. Indeed, in hearing spoken language users, text and spoken language are co-dependent [4, 5], and pictures are encoded via visual and verbal routes [6]. A parallel approach investigating semantic cognition shows that bilinguals activate similar patterns for the same words in their different languages [7, 8]. This suggests that conceptual representations are "language independent." However, this has only been tested in spoken language bilinguals. If different languages evoke different conceptual representations, this should be most apparent comparing languages that differ greatly in structure. Hearing people with signing deaf parents are bilingual in sign and speech: languages conveyed in different modalities. Here, we test the influence of modality and bilingualism on conceptual representation by comparing semantic representations elicited by spoken British English and British Sign Language in hearing early, sign-speech bilinguals. We show that representations of semantic categories are shared for sign and speech, but not for individual spoken words and signs. This provides evidence for partially shared representations for sign and speech and shows that language acts as a subtle filter through which we understand and interact with the world.
| S-EPMC6839399 | biostudies-literature
Project description:We consider the problem of performing unsupervised learning in the presence of outliers - that is, observations that do not come from the same distribution as the rest of the data. It is known that in this setting, standard approaches for unsupervised learning can yield unsatisfactory results. For instance, in the presence of severe outliers, K-means clustering will often assign each outlier to its own cluster, or alternatively may yield distorted clusters in order to accommodate the outliers. In this paper, we take a new approach to extending existing unsupervised learning techniques to accommodate outliers. Our approach is an extension of a recent proposal for outlier detection in the regression setting. We allow each observation to take on an "error" term, and we penalize the errors using a group lasso penalty in order to encourage most of the observations' errors to exactly equal zero. We show that this approach can be used in order to develop extensions of K-means clustering and principal components analysis that result in accurate outlier detection, as well as improved performance in the presence of outliers. These methods are illustrated in a simulation study and on two gene expression data sets, and connections with M-estimation are explored.
| S-EPMC3716393 | biostudies-literature
Project description:We address the problem, fundamental to linguistics, bioinformatics, and certain other disciplines, of using corpora of raw symbolic sequential data to infer underlying rules that govern their production. Given a corpus of strings (such as text, transcribed speech, chromosome or protein sequence data, sheet music, etc.), our unsupervised algorithm recursively distills from it hierarchically structured patterns. The adios (automatic distillation of structure) algorithm relies on a statistical method for pattern extraction and on structured generalization, two processes that have been implicated in language acquisition. It has been evaluated on artificial context-free grammars with thousands of rules, on natural languages as diverse as English and Chinese, and on protein data correlating sequence with function. This unsupervised algorithm is capable of learning complex syntax, generating grammatical novel sentences, and proving useful in other fields that call for structure discovery from raw data, such as bioinformatics.
| S-EPMC1187953 | biostudies-literature
Project description:Brain-Computer Interfaces (BCI) aim to bypass the peripheral nervous system to link the brain to external devices via successful modeling of decoding mechanisms. BCI based on electrocorticogram or ECoG represent a viable compromise between clinical practicality, spatial resolution, and signal quality when it comes to extracellular electrical potentials from local neuronal assemblies. Classic analysis of ECoG traces usually falls under the umbrella of Time-Frequency decompositions with adaptations from Fourier analysis and wavelets as its most prominent variants. However, analyzing such high-dimensional, multivariate time series demands for specialized signal processing and neurophysiological principles. We propose a generative model for single-channel ECoGs that is able to fully characterize reoccurring rhythm-specific neuromodulations as weighted activations of prototypical templates over time. The set of timings, weights and indexes comprise a temporal marked point process (TMPP) that accesses a set of bases from vector spaces of different dimensions-a dictionary. The shallow nature of the model admits the equivalence between latent variables and representations. In this way, learning the model parameters is a case of unsupervised representation learning. We exploit principles of Minimum Description Length (MDL) encoding to effectively yield a data-driven framework where prototypical neuromodulations (not restricted to a particular duration) can be estimated alongside the timings and features of the TMPP. We validate the proposed methodology on discrimination of movement-related tasks utilizing 32-electrode grids implanted in the frontal cortex of six epileptic subjects. We show that the learned representations from the high-gamma band (85-145 Hz) are not only interpretable, but also discriminant in a lower dimensional space. The results also underscore the practicality of our algorithm, i.e., 2 main hyperparameters that can be readily set via neurophysiology, and emphasize the need of principled and interpretable representation learning in order to model encoding mechanisms in the brain.
| S-EPMC6882771 | biostudies-literature