Different mechanisms involved in adaptation to stable and unstable dynamics.

Recently, we demonstrated that humans can learn to make accurate movements in an unstable environment by controlling magnitude, shape, and orientation of the endpoint impedance. Although previous studies of human motor learning suggest that the brain acquires an inverse dynamics model of the novel environment, it is not known whether this control mechanism is operative in unstable environments. We compared learning of multijoint arm movements in a "velocity-dependent force field" (VF), which interacted with the arm in a stable manner, and learning in a "divergent force field" (DF), where the interaction was unstable. The characteristics of error evolution were markedly different in the 2 fields. The direction of trajectory error in the DF alternated to the left and right during the early stage of learning; that is, signed error was inconsistent from movement to movement and could not have guided learning of an inverse dynamics model. This contrasted sharply with trajectory error in the VF, which was initially biased and decayed in a manner that was consistent with rapid feedback error learning. EMG recorded before and after learning in the DF and VF are also consistent with different learning and control mechanisms for adapting to stable and unstable dynamics, that is, inverse dynamics model formation and impedance control. We also investigated adaptation to a rotated DF to examine the interplay between inverse dynamics model formation and impedance control. Our results suggest that an inverse dynamics model can function in parallel with an impedance controller to compensate for consistent perturbing force in unstable environments.

Glucocorticoid receptor and protein/RNA synthesis-dependent mechanisms underlie the control of synaptic plasticity by stress

Learning and memory are exquisitely sensitive to behavioral stress, but the underlying mechanisms are still poorly understood. Because activity-dependent persistent changes in synaptic strength are believed to mediate memory processes in brain areas such as the hippocampus we have examined the means by which stress affects synaptic plasticity in the CA1 region of the hippocampus of anesthetized rats, Inescapable behavioral stress (placement on an elevated platform for 30 min) switched the direction of plasticity, favoring low frequency stimulation-induced decreases in synaptic transmission (long-term depression, LTD), and opposing the induction of long-term potentiation by high frequency stimulation, We have discovered that glucocorticoid receptor activation mediates these effects of stress on LTD and longterm potentiation in a protein synthesis-dependent manner because they were prevented by the glucocorticoid receptor antagonist RU 38486 and the protein synthesis inhibitor emetine. Consistent with this, the ability of exogenously applied corticosterone in non-stressed rats to mimic the effects of stress on synaptic plasticity was also blocked by these agents, The enablement of low frequency stimulation-induced LTD by both stress and exogenous corticosterone was also blocked by the transcription inhibitor actinomycin D, Thus, naturally occurring synaptic plasticity is liable to be reversed in stressful situations via glucocorticoid receptor activation and mechanisms dependent on the synthesis of new protein and RNA, This indicates that the modulation of hippocampus-mediated learning by acute inescapable stress requires glucocorticoid receptor-dependent initiation of transcription and translation.

Stress enables synaptic depression in CA1 synapses by acute and chronic morphine: Possible mechanisms for corticosterone on opiate addiction

The hippocampus, being sensitive to stress and glucocorticoids, plays significant roles in certain types of learning and memory. Therefore, the hippocampus is probably involved in the increasing drug use, drug seeking, and relapse caused by stress. We have studied the effect of stress with morphine on synaptic plasticity in the CA1 region of the hippocampus in vivo and on a delayed-escape paradigm of the Morris water maze. Our results reveal that acute stress enables long-term depression (LTD) induction by low-frequency stimulation (LFS) but acute morphine causes synaptic potentiation. Remarkably, exposure to an acute stressor reverses the effect of morphine from synaptic potentiation ( similar to 20%) to synaptic depression ( similar to 40%), precluding further LTD induction by LFS. The synaptic depression caused by stress with morphine is blocked either by the glucocorticoid receptor antagonist RU38486 or by the NMDA-receptor antagonist D-APV. Chronic morphine attenuates the ability of acute morphine to cause synaptic potentiation, and stress to enable LTD induction, but not the ability of stress in tandem with morphine to cause synaptic depression. Furthermore, corticosterone with morphine during the initial phase of drug use promotes later delayed-escape behavior, as indicated by the morphine-reinforced longer latencies to escape, leading to persistent morphine-seeking after withdrawal. These results suggest that hippocampal synaptic plasticity may play a significant role in the effects of stress or glucocorticoids on opiate addiction.

Prenatal stress modifies hippocampal synaptic plasticity and spatial learning in young rat offspring

Clinical studies demonstrate that prenatal stress causes cognitive deficits and increases vulnerability to affective disorders in children and adolescents. The underlying mechanisms are not yet fully understood. Here, we reported that prenatal stress (10

Principles of sensorimotor learning.

The exploits of Martina Navratilova and Roger Federer represent the pinnacle of motor learning. However, when considering the range and complexity of the processes that are involved in motor learning, even the mere mortals among us exhibit abilities that are impressive. We exercise these abilities when taking up new activities - whether it is snowboarding or ballroom dancing - but also engage in substantial motor learning on a daily basis as we adapt to changes in our environment, manipulate new objects and refine existing skills. Here we review recent research in human motor learning with an emphasis on the computational mechanisms that are involved.

Computational mechanisms of sensorimotor control.

In order to generate skilled and efficient actions, the motor system must find solutions to several problems inherent in sensorimotor control, including nonlinearity, nonstationarity, delays, redundancy, uncertainty, and noise. We review these problems and five computational mechanisms that the brain may use to limit their deleterious effects: optimal feedback control, impedance control, predictive control, Bayesian decision theory, and sensorimotor learning. Together, these computational mechanisms allow skilled and fluent sensorimotor behavior.

Principles of sensorimotor learning

The exploits of Martina Navratilova and Roger Federer represent the pinnacle of motor learning. However, when considering the range and complexity of the processes that are involved in motor learning, even the mere mortals among us exhibit abilities that are impressive. We exercise these abilities when taking up new activities-whether it is snowboarding or ballroom dancing-but also engage in substantial motor learning on a daily basis as we adapt to changes in our environment, manipulate new objects and refine existing skills. Here we review recent research in human motor learning with an emphasis on the computational mechanisms that are involved. © 2011 Macmillan Publishers Limited. All rights reserved.

The in vivo synaptic plasticity mechanism of EGb 761-induced enhancement of spatial learning and memory in aged rats

1 It has not been uniform to date that the Ginkgo biloba extracts enhance cognitive function in aged animals, and the mechanisms of action remain difficult to elucidate. In this study, the Morris water maze task and electrophysiological methods were used

Statistical inference for single- and multi-band probabilistic amplitude demodulation.

Amplitude demodulation is an ill-posed problem and so it is natural to treat it from a Bayesian viewpoint, inferring the most likely carrier and envelope under probabilistic constraints. One such treatment is Probabilistic Amplitude Demodulation (PAD), which, whilst computationally more intensive than traditional approaches, offers several advantages. Here we provide methods for estimating the uncertainty in the PAD-derived envelopes and carriers, and for learning free-parameters like the time-scale of the envelope. We show how the probabilistic approach can naturally handle noisy and missing data. Finally, we indicate how to extend the model to signals which contain multiple modulators and carriers.

Learning based automatic face annotation for arbitrary poses and expressions from frontal images only

Statistical approaches for building non-rigid deformable models, such as the Active Appearance Model (AAM), have enjoyed great popularity in recent years, but typically require tedious manual annotation of training images. In this paper, a learning based approach for the automatic annotation of visually deformable objects from a single annotated frontal image is presented and demonstrated on the example of automatically annotating face images that can be used for building AAMs for fitting and tracking. This approach employs the idea of initially learning the correspondences between landmarks in a frontal image and a set of training images with a face in arbitrary poses. Using this learner, virtual images of unseen faces at any arbitrary pose for which the learner was trained can be reconstructed by predicting the new landmark locations and warping the texture from the frontal image. View-based AAMs are then built from the virtual images and used for automatically annotating unseen images, including images of different facial expressions, at any random pose within the maximum range spanned by the virtually reconstructed images. The approach is experimentally validated by automatically annotating face images from three different databases. © 2009 IEEE.

Active learning of model evidence using Bayesian quadrature

Numerical integration is a key component of many problems in scientific computing, statistical modelling, and machine learning. Bayesian Quadrature is a modelbased method for numerical integration which, relative to standard Monte Carlo methods, offers increased sample efficiency and a more robust estimate of the uncertainty in the estimated integral. We propose a novel Bayesian Quadrature approach for numerical integration when the integrand is non-negative, such as the case of computing the marginal likelihood, predictive distribution, or normalising constant of a probabilistic model. Our approach approximately marginalises the quadrature model's hyperparameters in closed form, and introduces an active learning scheme to optimally select function evaluations, as opposed to using Monte Carlo samples. We demonstrate our method on both a number of synthetic benchmarks and a real scientific problem from astronomy.

Bayesian Learning of Noisy Markov Decision Processes

We consider the inverse reinforcement learning problem, that is, the problem of learning from, and then predicting or mimicking a controller based on state/action data. We propose a statistical model for such data, derived from the structure of a Markov decision process. Adopting a Bayesian approach to inference, we show how latent variables of the model can be estimated, and how predictions about actions can be made, in a unified framework. A new Markov chain Monte Carlo (MCMC) sampler is devised for simulation from the posterior distribution. This step includes a parameter expansion step, which is shown to be essential for good convergence properties of the MCMC sampler. As an illustration, the method is applied to learning a human controller.

Spatiotemporal variations of internal P-loading and the related mechanisms in the large shallow Lake Chaohu

Spatiotemporal variations of P species and adsorption behavior in water column, interstitial water, and sediments were investigated in the large shallow eutrophic Lake Chaohu. Orthophosphate (Ortho-P) and total phosphorus (TP) concentrations were significantly higher in the western part than in the eastern part of the lake, due to different nutrient inputs from the surrounding rivers. Moreover, particulate phosphorus (PP) concentration was in a similar spatial pattern to Ortho-P and TIP concentrations, and also showed significantly positive correlation with the biomass of Microcystis, indicating more uptake and store of phosphorus by Microcystis than by other algae. Increase of pH and intensive utilization of P by phytoplankton were the main factors promoting P (especially Fe-P) release from the sediment to interstitial water during the cyanobacterial blooms in Lake Chaohu. Spatial dynamics in TP concentration, P species and adsorption behavior of the sediment, coupled with the statistical analyses, suggested that the spatial heterogeneity of P contents in the sediment was influenced by various factors, e.g. human activities, soil geochemistry and mineral composition. In spite of similar TP contents in the sediments, increase in proportion of Fe-P concentration in the sediment may result in a high risk of P release.

基于精细结构描述及数值试验的断续节理岩体力学参数智能选取

Evaluating the mechanical properties of rock masses is the base of rock engineering design and construction. It has great influence on the safety and cost of rock project. The recognition is inevitable consequence of new engineering activities in rock, including high-rise building, super bridge, complex underground installations, hydraulic project and etc. During the constructions, lots of engineering accidents happened, which bring great damage to people. According to the investigation, many failures are due to choosing improper mechanical properties. ‘Can’t give the proper properties’ becomes one of big problems for theoretic analysis and numerical simulation. Selecting the properties reasonably and effectively is very significant for the planning, design and construction of rock engineering works. A multiple method based on site investigation, theoretic analysis, model test, numerical test and back analysis by artificial neural network is conducted to determine and optimize the mechanical properties for engineering design. The following outcomes are obtained: (1) Mapping of the rock mass structure Detailed geological investigation is the soul of the fine structure description. Based on statistical window，geological sketch and digital photography，a new method for rock mass fine structure in-situ mapping is developed. It has already been taken into practice and received good comments in Baihetan Hydropower Station. (2) Theoretic analysis of rock mass containing intermittent joints The shear strength mechanisms of joint and rock bridge are analyzed respectively. And the multiple modes of failure on different stress condition are summarized and supplied. Then, through introducing deformation compatibility equation in normal direction, the direct shear strength formulation and compression shear strength formulation for coplanar intermittent joints, as well as compression shear strength formulation for ladderlike intermittent joints are deducted respectively. In order to apply the deducted formulation conveniently in the real projects, a relationship between these formulations and Mohr-Coulomb hypothesis is built up. (3) Model test of rock mass containing intermittent joints Model tests are adopted to study the mechanical mechanism of joints to rock masses. The failure modes of rock mass containing intermittent joints are summarized from the model test. Six typical failure modes are found in the test, and brittle failures are the main failure mode. The evolvement processes of shear stress, shear displacement, normal stress and normal displacement are monitored by using rigid servo test machine. And the deformation and failure character during the loading process is analyzed. According to the model test, the failure modes quite depend on the joint distribution, connectivity and stress states. According to the contrastive analysis of complete stress strain curve, different failure developing stages are found in the intact rock, across jointed rock mass and intermittent jointed rock mass. There are four typical stages in the stress strain curve of intact rock, namely shear contraction stage, linear elastic stage, failure stage and residual strength stage. There are three typical stages in the across jointed rock mass, namely linear elastic stage, transition zone and sliding failure stage. Correspondingly, five typical stages are found in the intermittent jointed rock mass, namely linear elastic stage, sliding of joint, steady growth of post-crack, joint coalescence failure, and residual strength. According to strength analysis, the failure envelopes of intact rock and across jointed rock mass are the upper bound and lower bound separately. The strength of intermittent jointed rock mass can be evaluated by reducing the bandwidth of the failure envelope with geo-mechanics analysis. (4) Numerical test of rock mass Two sets of methods, i.e. the distinct element method (DEC) based on in-situ geology mapping and the realistic failure process analysis (RFPA) based on high-definition digital imaging, are developed and introduced. The operation process and analysis results are demonstrated detailedly from the research on parameters of rock mass based on numerical test in the Jinping First Stage Hydropower Station and Baihetan Hydropower Station. By comparison，the advantages and disadvantages are discussed. Then the applicable fields are figured out respectively. (5) Intelligent evaluation based on artificial neural network (ANN) The characters of both ANN and parameter evaluation of rock mass are discussed and summarized. According to the investigations, ANN has a bright application future in the field of parameter evaluation of rock mass. Intelligent evaluation of mechanical parameters in the Jinping First Stage Hydropower Station is taken as an example to demonstrate the analysis process. The problems in five aspects, i. e. sample selection, network design, initial value selection, learning rate and expected error, are discussed detailedly.