Computational modelling of spoken-word recognition processes

Although the field of bone regeneration has experienced great advancements in the last decades, integrating all the relevant, patient-specific information into a personalized diagnosis and optimal treatment remains a challenging task due to the large number of variables that affect bone regeneration. Computational models have the potential to cope with this complexity and to improve the fundamental understanding of the bone regeneration processes as well as to predict and optimize the patient-specific treatment strategies. However, the current use of computational models in daily orthopedic practice is very limited or inexistent. We have identified three key hurdles that limit the translation of computational models of bone regeneration from bench to bed side. First, there exists a clear mismatch between the scope of the existing and the clinically required models. Second, most computational models are confronted with limited quantitative information of insufficient quality thereby hampering the determination of patient-specific parameter values. Third, current computational models are only corroborated with animal models, whereas a thorough (retrospective and prospective) assessment of the computational model will be crucial to convince the health care providers of the capabilities thereof. These challenges must be addressed so that computational models of bone regeneration can reach their true potential, resulting in the advancement of individualized care and reduction of the associated health care costs.

Combined speech enhancement and auditory modelling for robust distributed speech recognition

Dynamical analysis of Parkinsonian state emulated by hybrid Izhikevich neuron models

A Bayesian framework for simultaneously modeling neural and behavioral data

Representational Similarity Analysis (RSA) has emerged as a popular method for relating representational spaces from human brain activity, behavioral data, and computational models. RSA is based on the comparison of representational (dis-)similarity matrices (RDMs or RSMs), which characterize the pairwise (dis-)similarities of all conditions across all features (e.g. fMRI voxels or units of a model). However, classical RSA treats each feature as equally important. This ‘equal weights’ assumption contrasts with the flexibility of multivariate decoding, which reweights individual features for predicting a target variable. As a consequence, classical RSA may lead researchers to underestimate the correspondence between a model and a brain region and, in case of model comparison, may lead them to select an inferior model. The aim of this work is twofold: First, we sought to broadly test feature-reweighted RSA (FR-RSA) applied to computational models and reveal the extent to which reweighting model features improves RSM correspondence and affects model selection. Previous work suggested that reweighting can improve model selection in RSA but it has remained unclear to what extent these results generalize across datasets and data modalities. To draw more general conclusions, we utilized a range of publicly available datasets and three popular deep neural networks (DNNs). Second, we propose voxel-reweighted RSA, a novel use case of FR-RSA that reweights fMRI voxels, mirroring the rationale of multivariate decoding of optimally combining voxel activity patterns. We found that reweighting individual model units markedly improved the fit between model RSMs and target RSMs derived from several fMRI and behavioral datasets and affected model selection, highlighting the importance of considering FR-RSA. For voxel-reweighted RSA, improvements in RSM correspondence were even more pronounced, demonstrating the utility of this novel approach. We additionally show that classical noise ceilings can be exceeded when FR-RSA is applied and propose an updated approach for their computation. Taken together, our results broadly validate the use of FR-RSA for improving the fit between computational models, brain, and behavioral data, allowing us to better adjudicate between competing computational models. Further, our results suggest that FR-RSA applied to brain measurement channels could become an important new method to assess the correspondence between representational spaces.

Evidence that listeners use durational cues to help resolve temporarily ambiguous speech input has accumulated over the past few years. In this paper, we investigate whether durational cues are also beneficial for word recognition in a computational model of spoken-word recognition. Two sets of simulations were carried out using the acoustic signal as input. The simulations showed that the computational model, like humans, takes benefit from durational cues during word recognition, and uses these to disambiguate the speech signal. These results thus provide support for the theory that durational cues play a role in spoken-word recognition.

Computational models have become more and more important in testing processing mechanisms assumed to underlie human spoken-word recognition. Models like TRACE (McClelland and Elman, 1986) and Shortlist (Norris, 1994) have given us much insight in the effects of, for instance, competition between words in the mental lexicon and the use of lexical information during word recognition. However, these models neglect the effects of coarticulation and variability over time by using mock speech instead of real speech input. We describe a new connectionist model for spoken-word recognition which differs on a number of points from other models, in that it takes real speech as input, is based on a new architecture for the representation of time, and can adapt its own weights. Simulations with the model accurately reproduce some important effects found in human word recognition. However, the representations of words in the model and the implementation of the frequency effect should be investigated more thoroughly.

Understanding how the human visual system recognizes objects is one of the key challenges in neuroscience. Inspired by a large body of physiological evidence, a general class of recognition models has emerged, which is based on a hierarchical organization of visual processing, with succeeding stages being sensitive to image features of increasing complexity. However, these models appear to be incompatible with some well-known psychophysical results. Prominent among these are experiments investigating recognition impairments caused by vertical inversion of images, especially those of faces. It has been reported that faces that differ 'featurally' are much easier to distinguish when inverted than those that differ 'configurally'; a finding that is difficult to reconcile with the physiological models. Here, we show that after controlling for subjects' expectations, there is no difference between 'featurally' and 'configurally' transformed faces in terms of inversion effect. This result reinforces the plausibility of simple hierarchical models of object representation and recognition in the cortex.

This chapter provides an alternative approach by developing a computational model of the cognitive processes underlying pilot performance while flying a descent in an automated cockpit. The computational model was built from a cognitive task analysis coupled with empirical performance data. Observations of the problems encountered by the model in flying the simulator suggested a number of interventions that might mitigate error in the cockpit. Two of these interventions were selected for empirical testing. First, model runs and eye track data both suggested that the pilots/model were often unaware of changes in automation mode that were driven by the software rather than the pilot. Second, when the model was interrupted, it often was unable to remember the goal that it was trying to achieve; thus, the model was unable to continue flying.

The recent assembly of the adult Drosophila melanogaster central brain connectome, containing more than 125,000 neurons and 50 million synaptic connections, provides a template for examining sensory processing throughout the brain1,2. Here we create a leaky integrate-and-fire computational model of the entire Drosophila brain, on the basis of neural connectivity and neurotransmitter identity3, to study circuit properties of feeding and grooming behaviours. We show that activation of sugar-sensing or water-sensing gustatory neurons in the computational model accurately predicts neurons that respond to tastes and are required for feeding initiation4. In addition, using the model to activate neurons in the feeding region of the Drosophila brain predicts those that elicit motor neuron firing5—a testable hypothesis that we validate by optogenetic activation and behavioural studies. Activating different classes of gustatory neurons in the model makes accurate predictions of how several taste modalities interact, providing circuit-level insight into aversive and appetitive taste processing. Additionally, we applied this model to mechanosensory circuits and found that computational activation of mechanosensory neurons predicts activation of a small set of neurons comprising the antennal grooming circuit, and accurately describes the circuit response upon activation of different mechanosensory subtypes6–10. Our results demonstrate that modelling brain circuits using only synapse-level connectivity and predicted neurotransmitter identity generates experimentally testable hypotheses and can describe complete sensorimotor transformations.

Purpose of the study.The aim is to create a technique targeted at the development of simulation skills in higher education environment, where students are competent in applying information technologies in economics. The relevance of the research lies in the fact that the existing methodological developments often focus on a specific software tool or methodology that cannot respond to all economic problems. A specialist in simulation modeling should possess integrative interdisciplinary knowledge from related scientific fields, for example, probability theory and mathematical statistics, higher mathematics, be familiar with other methods of solving economic problems: linear, nonlinear, dynamic programming, optimization; show proficiency in structural and functional analysis; be able to explore complex processes and systems comprehensively.Materials and methods.The following pedagogical approaches and teaching methods were implemented in this research: ● a systematic approach to solving complex problems based on the modelling economic objects as systems operating in a certain environment, ● activity approach to develop students’ professional competences in the process of creation, debugging and optimization of computer models of economic systems; ● problem teaching method in the framework of research and analysis of educational problems of the subject area; ● implemented interactive teaching methods; ● multimedia methods in the content of teaching materials of the discipline, including electronic manuals, educational videos, as well as multimedia presentations. The research also utilized information technologies, in which computers, communication equipment and software environments are: ● means to provide educational material to students for the transfer of knowledge; ● tools for designing, developing and conducting simulation experiments. In addition, we used the following special professional technologies, methods and tools in the process of teaching students: ● structural and functional modeling methodology; ● discrete-event approach to simulation methodology; ● special software for development and research of simulation models of economic processes and systems: Arena 15.0, AnyLogic 8.3.2.According to the requirements of the new educational standards, the student must master a sufficiently large amount of general cultural, professional and specialized competencies included in the curriculum. The application of the proposed approaches and methods allows to provide effective development of skills of simulation modeling of educational programs for bachelors of «Applied Informatics» and «Business Informatics».Results.The study created a method of teaching students the skills of simulation modeling. The research also established the model of formation of readiness of the IT specialist to the development of simulation models of economic processes and systems in higher school. We also identified important methodological conditions for the formation of professional competencies of students in the field of modeling, such as: ● application of a systematic approach to the analysis of domain problems, as well as for the synthesis of mathematical simulation models of business processes and economic systems; ● practical orientation of the content of training (selection and research in the educational process of the most characteristic, typical problems of the economy); ● integration of interdisciplinary knowledge, methods and approaches to solve complex problems.Conclusion.The method can equip students with skills of simulation modeling with various areas of practical application. First, this technique can be used by university students who are engaged in pursuing practical skills and basic system knowledge in the field of simulation. Secondly, teachers can use it, conducting courses: “Computer modeling”, “Mathematical and simulation modeling”, “Modeling of processes and systems” in the educational process of the University to improve the professional competence of students training simulation modeling. Third, the outcomes may be of interest to managers of educational programs in the areas of: “Applied Informatics”, “Business Informatics”, etc. to improve the structure and sequence of disciplines of competence-oriented curricula. Finally, the application of the proposed methodology in the educational process of the University will enhance professional expertise of young specialists and undoubtedly address the needs of potential employers.

Recent neurocognitive models commonly consider speech perception as a hierarchy of processes, each corresponding to specific temporal scales of collective oscillatory processes in the cortex: 30–80 Hz gamma oscillations in charge of phonetic analysis, 4–9 Hz theta oscillations in charge of syllabic segmentation, 1–2 Hz delta oscillations processing prosodic/syntactic units and the 15–20 Hz beta channel possibly involved in top-down predictions. Several recent neuro-computational models thus feature theta oscillations, driven by the speech acoustic envelope, to achieve syllabic parsing before lexical access. However, it is unlikely that such syllabic parsing, performed in a purely bottom-up manner from envelope variations, would be totally efficient in all situations, especially in adverse sensory conditions. We present a new probabilistic model of spoken word recognition, called COSMO-Onset, in which syllabic parsing relies on fusion between top-down, lexical prediction of onset events and bottom-up onset detection from the acoustic envelope. We report preliminary simulations, analyzing how the model performs syllabic parsing and phone, syllable and word recognition. We show that, while purely bottom-up onset detection is sufficient for word recognition in nominal conditions, top-down prediction of syllabic onset events allows overcoming challenging adverse conditions, such as when the acoustic envelope is degraded, leading either to spurious or missing onset events in the sensory signal. This provides a proposal for a possible computational functional role of top-down, predictive processes during speech recognition, consistent with recent models of neuronal oscillatory processes.

The neural mechanisms of cognitive dysfunctions in neurological diseases remain poorly understood. Here, we conjecture that this unsatisfying state-of-the-art is in part due to the non-specificity of the typical behavioral indicators for cognitive dysfunctions. Our study addresses the topic by advancing the assessment of cognitive dysfunctions through computational modeling. We investigate bradyphrenia in Parkinson’s disease (PD) as an exemplary case of cognitive dysfunctions in neurological diseases. Our computational model conceptualizes trial-by-trial behavioral data as resulting from parallel cognitive and sensorimotor reinforcement learning. We assessed PD patients ‘on’ and ‘off’ their dopaminergic medication and matched healthy control (HC) participants on a computerized version of the Wisconsin Card Sorting Test. PD patients showed increased retention of learned cognitive information and decreased retention of learned sensorimotor information from previous trials in comparison to HC participants. Systemic dopamine replacement therapy did not remedy these cognitive dysfunctions in PD patients but incurred non-desirable side effects such as decreasing cognitive learning from positive feedback. Our results reveal novel insights into facets of bradyphrenia that are indiscernible by observable behavioral indicators of cognitive dysfunctions. We discuss how computational modeling may contribute to the advancement of future research on brain–behavior relationships and neuropsychological assessment.

Introduction: In order to assess how well conventional and hybrid pitch detection techniques perform in speech processing applications, a comparative analysis of the two types of methods is conducted. Method: A proposed hybrid approach, Proposed PEF+CEP, is examined alongside five traditional algorithms, namely Normalized Correlation Function (NCF), Pitch Estimation Filter (PEF), Log- Harmonic Summation (LHS), Summation of Residual Harmonics (SRH) and Cepstrum Pitch Determination (CEP). The effectiveness is evaluated using performance metrics like accuracy, specificity, sensitivity, and Gross Pitch Error (GPE). Results and Discussion: Our findings show that the accuracy and specificity of the traditional methods are impressive; the accuracy and sensitivity of the suggested hybrid method surpass their performance, with an astounding 98.8% accuracy and 99.2% sensitivity. Conclusion: Furthermore, the Proposed PEF+CEP method is a promising solution for accurate and dependable pitch detection in speech processing applications because it strikes a strong balance between computational efficiency and robustness. These results open up new avenues for research in the field of speech processing and demonstrate the potential of hybrid approaches.

Computational modelling of the lungs is an active field of study that integrates computational advances with lung biophysics, biomechanics, physiology and medical imaging to promote individualized diagnosis, prognosis and therapy evaluation in lung diseases. The complex and hierarchical architecture of the lung offers a rich, but also challenging, research area demanding a cross-scale understanding of lung mechanics and advanced computational tools to effectively model lung biomechanics in both health and disease. Various approaches have been proposed to study different aspects of respiration, ranging from compartmental to discrete micromechanical and continuum representations of the lungs. This article reviews several developments in computational lung modelling and how they are integrated with preclinical and clinical data. We begin with a description of lung anatomy and how different tissue components across multiple length scales affect lung mechanics at the organ level. We then review common physiological and imaging data acquisition methods used to inform modelling efforts. Building on these reviews, we next present a selection of model-based paradigms that integrate data acquisitions with modelling to understand, simulate and predict lung dynamics in health and disease. Finally, we highlight possible future directions where computational modelling can improve our understanding of the structure–function relationship in the lung.