ABSTRACT
The development of whole-brain models that can infer effective (directed) connection strengths from fMRI data represents a central challenge for computational neuroimaging. A recently introduced generative model of fMRI data, regression dynamic causal modeling (rDCM), moves towards this goal as it scales gracefully to very large networks. However, large-scale networks with thousands of connections are difficult to interpret; additionally, one typically lacks information (data points per free parameter) for precise estimation of all model parameters. This paper introduces sparsity constraints to the variational Bayesian framework of rDCM as a solution to these problems in the domain of task-based fMRI. This sparse rDCM approach enables highly efficient effective connectivity analyses in whole-brain networks and does not require a priori assumptions about the network's connectivity structure but prunes fully (all-to-all) connected networks as part of model inversion. Following the derivation of the variational Bayesian update equations for sparse rDCM, we use both simulated and empirical data to assess the face validity of the model. In particular, we show that it is feasible to infer effective connection strengths from fMRI data using a network with more than 100 regions and 10,000 connections. This demonstrates the feasibility of whole-brain inference on effective connectivity from fMRI data - in single subjects and with a run-time below 1â¯min when using parallelized code. We anticipate that sparse rDCM may find useful application in connectomics and clinical neuromodeling - for example, for phenotyping individual patients in terms of whole-brain network structure.
Subject(s)
Brain/physiology , Connectome/methods , Models, Neurological , Models, Theoretical , Nerve Net/physiology , Bayes Theorem , Humans , Magnetic Resonance Imaging/methodsABSTRACT
The development of large-scale network models that infer the effective (directed) connectivity among neuronal populations from neuroimaging data represents a key challenge for computational neuroscience. Dynamic causal models (DCMs) of neuroimaging and electrophysiological data are frequently used for inferring effective connectivity but are presently restricted to small graphs (typically up to 10 regions) in order to keep model inversion computationally feasible. Here, we present a novel variant of DCM for functional magnetic resonance imaging (fMRI) data that is suited to assess effective connectivity in large (whole-brain) networks. The approach rests on translating a linear DCM into the frequency domain and reformulating it as a special case of Bayesian linear regression. This paper derives regression DCM (rDCM) in detail and presents a variational Bayesian inversion method that enables extremely fast inference and accelerates model inversion by several orders of magnitude compared to classical DCM. Using both simulated and empirical data, we demonstrate the face validity of rDCM under different settings of signal-to-noise ratio (SNR) and repetition time (TR) of fMRI data. In particular, we assess the potential utility of rDCM as a tool for whole-brain connectomics by challenging it to infer effective connection strengths in a simulated whole-brain network comprising 66 regions and 300 free parameters. Our results indicate that rDCM represents a computationally highly efficient approach with promising potential for inferring whole-brain connectivity from individual fMRI data.
Subject(s)
Brain/physiology , Connectome/methods , Magnetic Resonance Imaging/methods , Models, Neurological , Adult , Bayes Theorem , Brain/diagnostic imaging , HumansABSTRACT
Over the past decade, computational approaches to neuroimaging have increasingly made use of hierarchical Bayesian models (HBMs), either for inferring on physiological mechanisms underlying fMRI data (e.g., dynamic causal modelling, DCM) or for deriving computational trajectories (from behavioural data) which serve as regressors in general linear models. However, an unresolved problem is that standard methods for inverting the hierarchical Bayesian model are either very slow, e.g. Markov Chain Monte Carlo Methods (MCMC), or are vulnerable to local minima in non-convex optimisation problems, such as variational Bayes (VB). This article considers Gaussian process optimisation (GPO) as an alternative approach for global optimisation of sufficiently smooth and efficiently evaluable objective functions. GPO avoids being trapped in local extrema and can be computationally much more efficient than MCMC. Here, we examine the benefits of GPO for inverting HBMs commonly used in neuroimaging, including DCM for fMRI and the Hierarchical Gaussian Filter (HGF). Importantly, to achieve computational efficiency despite high-dimensional optimisation problems, we introduce a novel combination of GPO and local gradient-based search methods. The utility of this GPO implementation for DCM and HGF is evaluated against MCMC and VB, using both synthetic data from simulations and empirical data. Our results demonstrate that GPO provides parameter estimates with equivalent or better accuracy than the other techniques, but at a fraction of the computational cost required for MCMC. We anticipate that GPO will prove useful for robust and efficient inversion of high-dimensional and nonlinear models of neuroimaging data.
Subject(s)
Bayes Theorem , Brain Mapping/methods , Brain/physiology , Magnetic Resonance Imaging/methods , Models, Neurological , Algorithms , Computer Simulation , Humans , Normal DistributionABSTRACT
Inferring on others' (potentially time-varying) intentions is a fundamental problem during many social transactions. To investigate the underlying mechanisms, we applied computational modeling to behavioral data from an economic game in which 16 pairs of volunteers (randomly assigned to "player" or "adviser" roles) interacted. The player performed a probabilistic reinforcement learning task, receiving information about a binary lottery from a visual pie chart. The adviser, who received more predictive information, issued an additional recommendation. Critically, the game was structured such that the adviser's incentives to provide helpful or misleading information varied in time. Using a meta-Bayesian modeling framework, we found that the players' behavior was best explained by the deployment of hierarchical learning: they inferred upon the volatility of the advisers' intentions in order to optimize their predictions about the validity of their advice. Beyond learning, volatility estimates also affected the trial-by-trial variability of decisions: participants were more likely to rely on their estimates of advice accuracy for making choices when they believed that the adviser's intentions were presently stable. Finally, our model of the players' inference predicted the players' interpersonal reactivity index (IRI) scores, explicit ratings of the advisers' helpfulness and the advisers' self-reports on their chosen strategy. Overall, our results suggest that humans (i) employ hierarchical generative models to infer on the changing intentions of others, (ii) use volatility estimates to inform decision-making in social interactions, and (iii) integrate estimates of advice accuracy with non-social sources of information. The Bayesian framework presented here can quantify individual differences in these mechanisms from simple behavioral readouts and may prove useful in future clinical studies of maladaptive social cognition.
Subject(s)
Bayes Theorem , Decision Making , Learning , Models, Psychological , Social Behavior , Adult , Games, Experimental , Humans , Intention , Male , Motivation , Young AdultABSTRACT
Pain is known to comprise sensory, cognitive, and affective aspects. Despite numerous previous fMRI studies, however, it remains open which spatial distribution of activity is sufficient to encode whether a stimulus is perceived as painful or not. In this study, we analyzed fMRI data from a perceptual decision-making task in which participants were exposed to near-threshold laser pulses. Using multivariate analyses on different spatial scales, we investigated the predictive capacity of fMRI data for decoding whether a stimulus had been perceived as painful. Our analysis yielded a rank order of brain regions: during pain anticipation, activity in the periaqueductal gray (PAG) and orbitofrontal cortex (OFC) afforded the most accurate trial-by-trial discrimination between painful and non-painful experiences; whereas during the actual stimulation, primary and secondary somatosensory cortex, anterior insula, dorsolateral and ventrolateral prefrontal cortex, and OFC were most discriminative. The most accurate prediction of pain perception from the stimulation period, however, was enabled by the combined activity in pain regions commonly referred to as the 'pain matrix'. Our results demonstrate that the neural representation of (near-threshold) pain is spatially distributed and can be best described at an intermediate spatial scale. In addition to its utility in establishing structure-function mappings, our approach affords trial-by-trial predictions and thus represents a step towards the goal of establishing an objective neuronal marker of pain perception.
Subject(s)
Brain Mapping/methods , Brain/physiology , Image Interpretation, Computer-Assisted/methods , Pain Perception/physiology , Adult , Female , Humans , Magnetic Resonance Imaging , Male , Young AdultABSTRACT
Decoding models, such as those underlying multivariate classification algorithms, have been increasingly used to infer cognitive or clinical brain states from measures of brain activity obtained by functional magnetic resonance imaging (fMRI). The practicality of current classifiers, however, is restricted by two major challenges. First, due to the high data dimensionality and low sample size, algorithms struggle to separate informative from uninformative features, resulting in poor generalization performance. Second, popular discriminative methods such as support vector machines (SVMs) rarely afford mechanistic interpretability. In this paper, we address these issues by proposing a novel generative-embedding approach that incorporates neurobiologically interpretable generative models into discriminative classifiers. Our approach extends previous work on trial-by-trial classification for electrophysiological recordings to subject-by-subject classification for fMRI and offers two key advantages over conventional methods: it may provide more accurate predictions by exploiting discriminative information encoded in 'hidden' physiological quantities such as synaptic connection strengths; and it affords mechanistic interpretability of clinical classifications. Here, we introduce generative embedding for fMRI using a combination of dynamic causal models (DCMs) and SVMs. We propose a general procedure of DCM-based generative embedding for subject-wise classification, provide a concrete implementation, and suggest good-practice guidelines for unbiased application of generative embedding in the context of fMRI. We illustrate the utility of our approach by a clinical example in which we classify moderately aphasic patients and healthy controls using a DCM of thalamo-temporal regions during speech processing. Generative embedding achieves a near-perfect balanced classification accuracy of 98% and significantly outperforms conventional activation-based and correlation-based methods. This example demonstrates how disease states can be detected with very high accuracy and, at the same time, be interpreted mechanistically in terms of abnormalities in connectivity. We envisage that future applications of generative embedding may provide crucial advances in dissecting spectrum disorders into physiologically more well-defined subgroups.
Subject(s)
Algorithms , Aphasia/physiopathology , Brain/physiopathology , Computational Biology/methods , Magnetic Resonance Imaging , Adult , Aged , Bayes Theorem , Brain/pathology , Databases, Factual , Humans , Male , Middle Aged , Models, Neurological , Nervous System Diseases/diagnosis , Nervous System Diseases/physiopathology , Pattern Recognition, Automated , Principal Component Analysis , Reproducibility of Results , Speech PerceptionABSTRACT
In this study, we make a general comparison of the accuracy and robustness of five multivariate calibration models: partial least squares (PLS) regression or projection to latent structures, polynomial partial least squares (Poly-PLS) regression, artificial neural networks (ANNs), and two novel techniques based on support vector machines (SVMs) for multivariate data analysis: support vector regression (SVR) and least-squares support vector machines (LS-SVMs). The comparison is based on fourteen (14) different datasets: seven sets of gasoline data (density, benzene content, and fractional composition/boiling points), two sets of ethanol gasoline fuel data (density and ethanol content), one set of diesel fuel data (total sulfur content), three sets of petroleum (crude oil) macromolecules data (weight percentages of asphaltenes, resins, and paraffins), and one set of petroleum resins data (resins content). Vibrational (near-infrared, NIR) spectroscopic data are used to predict the properties and quality coefficients of gasoline, biofuel/biodiesel, diesel fuel, and other samples of interest. The four systems presented here range greatly in composition, properties, strength of intermolecular interactions (e.g., van der Waals forces, H-bonds), colloid structure, and phase behavior. Due to the high diversity of chemical systems studied, general conclusions about SVM regression methods can be made. We try to answer the following question: to what extent can SVM-based techniques replace ANN-based approaches in real-world (industrial/scientific) applications? The results show that both SVR and LS-SVM methods are comparable to ANNs in accuracy. Due to the much higher robustness of the former, the SVM-based approaches are recommended for practical (industrial) application. This has been shown to be especially true for complicated, highly nonlinear objects.
Subject(s)
Artificial Intelligence , Spectroscopy, Near-Infrared/methods , Algorithms , Biofuels/analysis , Gasoline/analysis , Least-Squares Analysis , Neural Networks, Computer , Principal Component Analysis , SoftwareABSTRACT
A multilayer feed-forward artificial neural network (MLP-ANN) with a single, hidden layer that contains a finite number of neurons can be regarded as a universal non-linear approximator. Today, the ANN method and linear regression (MLR) model are widely used for quantum chemistry (QC) data analysis (e.g., thermochemistry) to improve their accuracy (e.g., Gaussian G2-G4, B3LYP/B3-LYP, X1, or W1 theoretical methods). In this study, an alternative approach based on support vector machines (SVMs) is used, the least squares support vector machine (LS-SVM) regression. It has been applied to ab initio (first principle) and density functional theory (DFT) quantum chemistry data. So, QC + SVM methodology is an alternative to QC + ANN one. The task of the study was to estimate the Møller-Plesset (MPn) or DFT (B3LYP, BLYP, BMK) energies calculated with large basis sets (e.g., 6-311G(3df,3pd)) using smaller ones (6-311G, 6-311G*, 6-311G**) plus molecular descriptors. A molecular set (BRM-208) containing a total of 208 organic molecules was constructed and used for the LS-SVM training, cross-validation, and testing. MP2, MP3, MP4(DQ), MP4(SDQ), and MP4/MP4(SDTQ) ab initio methods were tested. Hartree-Fock (HF/SCF) results were also reported for comparison. Furthermore, constitutional (CD: total number of atoms and mole fractions of different atoms) and quantum-chemical (QD: HOMO-LUMO gap, dipole moment, average polarizability, and quadrupole moment) molecular descriptors were used for the building of the LS-SVM calibration model. Prediction accuracies (MADs) of 1.62 ± 0.51 and 0.85 ± 0.24 kcal mol(-1) (1 kcal mol(-1) = 4.184 kJ mol(-1)) were reached for SVM-based approximations of ab initio and DFT energies, respectively. The LS-SVM model was more accurate than the MLR model. A comparison with the artificial neural network approach shows that the accuracy of the LS-SVM method is similar to the accuracy of ANN. The extrapolation and interpolation results show that LS-SVM is superior by almost an order of magnitude over the ANN method in terms of the stability, generality, and robustness of the final model. The LS-SVM model needs a much smaller numbers of samples (a much smaller sample set) to make accurate prediction results. Potential energy surface (PES) approximations for molecular dynamics (MD) studies are discussed as a promising application for the LS-SVM calibration approach.
Subject(s)
Quantum Theory , Algorithms , Least-Squares Analysis , Molecular Dynamics Simulation , Neural Networks, ComputerABSTRACT
Psychiatry faces fundamental challenges with regard to mechanistically guided differential diagnosis, as well as prediction of clinical trajectories and treatment response of individual patients. This has motivated the genesis of two closely intertwined fields: (i) Translational Neuromodeling (TN), which develops "computational assays" for inferring patient-specific disease processes from neuroimaging, electrophysiological, and behavioral data; and (ii) Computational Psychiatry (CP), with the goal of incorporating computational assays into clinical decision making in everyday practice. In order to serve as objective and reliable tools for clinical routine, computational assays require end-to-end pipelines from raw data (input) to clinically useful information (output). While these are yet to be established in clinical practice, individual components of this general end-to-end pipeline are being developed and made openly available for community use. In this paper, we present the Translational Algorithms for Psychiatry-Advancing Science (TAPAS) software package, an open-source collection of building blocks for computational assays in psychiatry. Collectively, the tools in TAPAS presently cover several important aspects of the desired end-to-end pipeline, including: (i) tailored experimental designs and optimization of measurement strategy prior to data acquisition, (ii) quality control during data acquisition, and (iii) artifact correction, statistical inference, and clinical application after data acquisition. Here, we review the different tools within TAPAS and illustrate how these may help provide a deeper understanding of neural and cognitive mechanisms of disease, with the ultimate goal of establishing automatized pipelines for predictions about individual patients. We hope that the openly available tools in TAPAS will contribute to the further development of TN/CP and facilitate the translation of advances in computational neuroscience into clinically relevant computational assays.
ABSTRACT
Artificial neural network (ANN) approach has been applied to estimate the density functional theory (DFT) energy with large basis set using lower-level energy values and molecular descriptors. A total of 208 different molecules were used for the ANN training, cross validation, and testing by applying BLYP, B3LYP, and BMK density functionals. Hartree-Fock results were reported for comparison. Furthermore, constitutional molecular descriptor (CD) and quantum-chemical molecular descriptor (QD) were used for building the calibration model. The neural network structure optimization, leading to four to five hidden neurons, was also carried out. The usage of several low-level energy values was found to greatly reduce the prediction error. An expected error, mean absolute deviation, for ANN approximation to DFT energies was 0.6+/-0.2 kcal mol(-1). In addition, the comparison of the different density functionals with the basis sets and the comparison of multiple linear regression results were also provided. The CDs were found to overcome limitation of the QD. Furthermore, the effective ANN model for DFT/6-311G(3df,3pd) and DFT/6-311G(2df,2pd) energy estimation was developed, and the benchmark results were provided.
ABSTRACT
In its full sense, perception rests on an agent's model of how its sensory input comes about and the inferences it draws based on this model. These inferences are necessarily uncertain. Here, we illustrate how the Hierarchical Gaussian Filter (HGF) offers a principled and generic way to deal with the several forms that uncertainty in perception takes. The HGF is a recent derivation of one-step update equations from Bayesian principles that rests on a hierarchical generative model of the environment and its (in)stability. It is computationally highly efficient, allows for online estimates of hidden states, and has found numerous applications to experimental data from human subjects. In this paper, we generalize previous descriptions of the HGF and its account of perceptual uncertainty. First, we explicitly formulate the extension of the HGF's hierarchy to any number of levels; second, we discuss how various forms of uncertainty are accommodated by the minimization of variational free energy as encoded in the update equations; third, we combine the HGF with decision models and demonstrate the inversion of this combination; finally, we report a simulation study that compared four optimization methods for inverting the HGF/decision model combination at different noise levels. These four methods (Nelder-Mead simplex algorithm, Gaussian process-based global optimization, variational Bayes and Markov chain Monte Carlo sampling) all performed well even under considerable noise, with variational Bayes offering the best combination of efficiency and informativeness of inference. Our results demonstrate that the HGF provides a principled, flexible, and efficient-but at the same time intuitive-framework for the resolution of perceptual uncertainty in behaving agents.
ABSTRACT
Near infrared (NIR) spectroscopy is a non-destructive (vibrational spectroscopy based) measurement technique for many multicomponent chemical systems, including products of petroleum (crude oil) refining and petrochemicals, food products (tea, fruits, e.g., apples, milk, wine, spirits, meat, bread, cheese, etc.), pharmaceuticals (drugs, tablets, bioreactor monitoring, etc.), and combustion products. In this paper we have compared the abilities of nine different multivariate classification methods: linear discriminant analysis (LDA), quadratic discriminant analysis (QDA), regularized discriminant analysis (RDA), soft independent modeling of class analogy (SIMCA), partial least squares (PLS) classification, K-nearest neighbor (KNN), support vector machines (SVM), probabilistic neural network (PNN), and multilayer perceptron (ANN-MLP) - for gasoline classification. Three sets of near infrared (NIR) spectra (450, 415, and 345 spectra) were used for classification of gasolines into 3, 6, and 3 classes, respectively, according to their source (refinery or process) and type. The 14,000-8000 cm(-1) NIR spectral region was chosen. In all cases NIR spectroscopy was found to be effective for gasoline classification purposes, when compared with nuclear magnetic resonance (NMR) spectroscopy or gas chromatography (GC). KNN, SVM, and PNN techniques for classification were found to be among the most effective ones. Artificial neural network (ANN-MLP) approach based on principal component analysis (PCA), which was believed to be efficient, has shown much worse results. We hope that the results obtained in this study will help both further chemometric (multivariate data analysis) investigations and investigations in the sphere of applied vibrational (infrared/IR, near-IR, and Raman) spectroscopy of sophisticated multicomponent systems.