Perceptual learning, roving and the unsupervised bias

Perceptual learning improves perception through training. Perceptual learning improves with most stimulus types but fails when . certain stimulus types are mixed during training (roving). This result is surprising because classical supervised and unsupervised neural network models can cope easily with roving conditions. What makes humans so inferior compared to these models? As experimental and conceptual work has shown, human perceptual learning is neither supervised nor unsupervised but reward-based learning. Reward-based learning suffers from the so-called unsupervised bias, i.e., to prevent synaptic " drift" , the . average reward has to be exactly estimated. However, this is impossible when two or more stimulus types with different rewards are presented during training (and the reward is estimated by a running average). For this reason, we propose no learning occurs in roving conditions. However, roving hinders perceptual learning only for combinations of similar stimulus types but not for dissimilar ones. In this latter case, we propose that a critic can estimate the reward for each stimulus type separately. One implication of our analysis is that the critic cannot be located in the visual system. © 2011 Elsevier Ltd.

Comparison of nonlinear and linear PCA on surface wind, surface height, and SST in the South China Sea

We compared nonlinear principal component analysis (NLPCA) with linear principal component analysis (LPCA) with the data of sea surface wind anomalies (SWA), surface height anomalies (SSHA), and sea surface temperature anomalies (SSTA), taken in the South China Sea (SCS) between 1993 and 2003. The SCS monthly data for SWA, SSHA and SSTA (i.e., the anomalies with climatological seasonal cycle removed) were pre-filtered by LPCA, with only three leading modes retained. The first three modes of SWA, SSHA, and SSTA of LPCA explained 86%, 71%, and 94% of the total variance in the original data, respectively. Thus, the three associated time coefficient functions (TCFs) were used as the input data for NLPCA network. The NLPCA was made based on feed-forward neural network models. Compared with classical linear PCA, the first NLPCA mode could explain more variance than linear PCA for the above data. The nonlinearity of SWA and SSHA were stronger in most areas of the SCS. The first mode of the NLPCA on the SWA and SSHA accounted for 67.26% of the variance versus 54.7%, and 60.24% versus 50.43%, respectively for the first LPCA mode. Conversely, the nonlinear SSTA, localized in the northern SCS and southern continental shelf region, resulted in little improvement in the explanation of the variance for the first NLPCA.

Vector Associative Maps: Unsupervised Real-time Error-based Learning and Control of Movement Trajectories

This article describes neural network models for adaptive control of arm movement trajectories during visually guided reaching and, more generally, a framework for unsupervised real-time error-based learning. The models clarify how a child, or untrained robot, can learn to reach for objects that it sees. Piaget has provided basic insights with his concept of a circular reaction: As an infant makes internally generated movements of its hand, the eyes automatically follow this motion. A transformation is learned between the visual representation of hand position and the motor representation of hand position. Learning of this transformation eventually enables the child to accurately reach for visually detected targets. Grossberg and Kuperstein have shown how the eye movement system can use visual error signals to correct movement parameters via cerebellar learning. Here it is shown how endogenously generated arm movements lead to adaptive tuning of arm control parameters. These movements also activate the target position representations that are used to learn the visuo-motor transformation that controls visually guided reaching. The AVITE model presented here is an adaptive neural circuit based on the Vector Integration to Endpoint (VITE) model for arm and speech trajectory generation of Bullock and Grossberg. In the VITE model, a Target Position Command (TPC) represents the location of the desired target. The Present Position Command (PPC) encodes the present hand-arm configuration. The Difference Vector (DV) population continuously.computes the difference between the PPC and the TPC. A speed-controlling GO signal multiplies DV output. The PPC integrates the (DV)·(GO) product and generates an outflow command to the arm. Integration at the PPC continues at a rate dependent on GO signal size until the DV reaches zero, at which time the PPC equals the TPC. The AVITE model explains how self-consistent TPC and PPC coordinates are autonomously generated and learned. Learning of AVITE parameters is regulated by activation of a self-regulating Endogenous Random Generator (ERG) of training vectors. Each vector is integrated at the PPC, giving rise to a movement command. The generation of each vector induces a complementary postural phase during which ERG output stops and learning occurs. Then a new vector is generated and the cycle is repeated. This cyclic, biphasic behavior is controlled by a specialized gated dipole circuit. ERG output autonomously stops in such a way that, across trials, a broad sample of workspace target positions is generated. When the ERG shuts off, a modulator gate opens, copying the PPC into the TPC. Learning of a transformation from TPC to PPC occurs using the DV as an error signal that is zeroed due to learning. This learning scheme is called a Vector Associative Map, or VAM. The VAM model is a general-purpose device for autonomous real-time error-based learning and performance of associative maps. The DV stage serves the dual function of reading out new TPCs during performance and reading in new adaptive weights during learning, without a disruption of real-time operation. YAMs thus provide an on-line unsupervised alternative to the off-line properties of supervised error-correction learning algorithms. YAMs and VAM cascades for learning motor-to-motor and spatial-to-motor maps are described. YAM models and Adaptive Resonance Theory (ART) models exhibit complementary matching, learning, and performance properties that together provide a foundation for designing a total sensory-cognitive and cognitive-motor autonomous system.

Store Working Memory Networks for Storage and Recall of Arbitrary Temporal Sequences

Neural network models of working memory, called Sustained Temporal Order REcurrent (STORE) models, are described. They encode the invariant temporal order of sequential events in short term memory (STM) in a way that mimics cognitive data about working memory, including primacy, recency, and bowed order and error gradients. As new items are presented, the pattern of previously stored items is invariant in the sense that, relative activations remain constant through time. This invariant temporal order code enables all possible groupings of sequential events to be stably learned and remembered in real time, even as new events perturb the system. Such a competence is needed to design self-organizing temporal recognition and planning systems in which any subsequence of events may need to be categorized in order to to control and predict future behavior or external events. STORE models show how arbitrary event sequences may be invariantly stored, including repeated events. A preprocessor interacts with the working memory to represent event repeats in spatially separate locations. It is shown why at least two processing levels are needed to invariantly store events presented with variable durations and interstimulus intervals. It is also shown how network parameters control the type and shape of primacy, recency, or bowed temporal order gradients that will be stored.

Temperature modelling of an homogeneous medium using genetically selected RBF (LIC)

Temperature modelling of human tissue subjected to ultrasound for therapeutic use is essencial for an accurate instrumental assessment and calibration. In this paper punctual temperature modeling of a homogeneous medium, radiated by therapeutic ultrasound, is presented.

New methods for PID autotuning

In this paper a recent approach for PID autotuning, involving neural networks, is ferther developed. To make this approach adaptive, optimal PID values must be known on-line. In this paper neural network models of tuning criteria, together with the use of genetic algorithms, are proposed to solve this problem.

‘Modeling and Analysis of Competing Risks Data’

there has been much research on analyzing various forms of competing risks data. Nevertheless, there are several occasions in survival studies, where the existing models and methodologies are inadequate for the analysis competing risks data. ldentifiabilty problem and various types of and censoring induce more complications in the analysis of competing risks data than in classical survival analysis. Parametric models are not adequate for the analysis of competing risks data since the assumptions about the underlying lifetime distributions may not hold well. Motivated by this, in the present study. we develop some new inference procedures, which are completely distribution free for the analysis of competing risks data.

Predicting stock index volatility: can market volume help?

This paper explores a number of statistical models for predicting the daily stock return volatility of an aggregate of all stocks traded on the NYSE. An application of linear and non-linear Granger causality tests highlights evidence of bidirectional causality, although the relationship is stronger from volatility to volume than the other way around. The out-of-sample forecasting performance of various linear, GARCH, EGARCH, GJR and neural network models of volatility are evaluated and compared. The models are also augmented by the addition of a measure of lagged volume to form more general ex-ante forecasting models. The results indicate that augmenting models of volatility with measures of lagged volume leads only to very modest improvements, if any, in forecasting performance.

Construction of optimal prediction intervals for load forecasting problems

Short-term load forecasting is fundamental for the reliable and efficient operation of power systems. Despite its importance, accurate prediction of loads is problematic and far remote. Often uncertainties significantly degrade performance of load forecasting models. Besides, there is no index available indicating reliability of predicted values. The objective of this study is to construct prediction intervals for future loads instead of forecasting their exact values. The delta technique is applied for constructing prediction intervals for outcomes of neural network models. Some statistical measures are developed for quantitative and comprehensive evaluation of prediction intervals. According to these measures, a new cost function is designed for shortening length of prediction intervals without compromising their coverage probability. Simulated annealing is used for minimization of this cost function and adjustment of neural network parameters. Demonstrated results clearly show that the proposed methods for constructing prediction interval outperforms the traditional delta technique. Besides, it yields prediction intervals that are practically more reliable and useful than exact point predictions.

A framework for mapping vegetation over broad spatial extents : a technique to aid land management across jurisdictional boundaries

Mismatches in boundaries between natural ecosystems and land governance units often complicate an ecosystem approach to management and conservation. For example, information used to guide management, such as vegetation maps, may not be available or consistent across entire ecosystems. This study was undertaken within a single biogeographic region (the Murray Mallee) spanning three Australian states. Existing vegetation maps could not be used as vegetation classifications differed between states. Our aim was to describe and map ‘tree mallee’ vegetation consistently across a 104 000km2 area of this region. Hierarchical cluster analyses, incorporating floristic data from 713 sites, were employed to identify distinct vegetation types. Neural network classification models were used to map these vegetation types across the region, with additional data from 634 validation sites providing a measure of map accuracy. Four distinct vegetation types were recognised: Triodia Mallee, Heathy Mallee, Chenopod Mallee and Shrubby Mallee. Neural network models predicted the occurrence of three of them with 79% accuracy. Validation results identified that map accuracy was 67% (kappa = 0.42) when using independent data. The framework employed provides a simple approach to describing and mapping vegetation consistently across broad spatial extents. Specific outcomes include: (1) a system of vegetation classification suitable for use across this biogeographic region; (2) a consistent vegetationmapto inform land-use planning and biodiversity management at local and regional scales; and (3) a quantification of map accuracy using independent data. This approach is applicable to other regions facing similar challenges associated with integrating vegetation data across jurisdictional boundaries.

Structural classification of proteins through amino acid sequence using interval type-2 fuzzy logic system

This paper introduces a new multi-output interval type-2 fuzzy logic system (MOIT2FLS) that is automatically constructed from unsupervised data clustering method and trained using heuristic genetic algorithm for a protein secondary structure classification. Three structure classes are distinguished including helix, strand (sheet) and coil which correspond to three outputs of the MOIT2FLS. Quantitative properties of amino acids are used to characterize the twenty amino acids rather than the widely used computationally expensive binary encoding scheme. Amino acid sequences are parsed into learnable patterns using a local moving window strategy. Three clustering tasks are performed using the adaptive vector quantization method to derive an equal number of initial rules for each type of secondary structure. Genetic algorithm is applied to optimally adjust parameters of the MOIT2FLS with the purpose of maximizing the Q3 measure. Comprehensive experimental results demonstrate the strong superiority of the proposed approach over the traditional methods including Chou-Fasman method, Garnier-Osguthorpe-Robson method, and artificial neural network models.

Performance analysis of three advanced controllers for polymerization batch reactor: an experimental investigation

The performances of three advanced non-linear controllers are analyzed for the optimal set point tracking of styrene free radical polymerization (FRP) in batch reactors. The three controllers are the artificial neural network-based MPC (NN-MPC), the artificial fuzzy logic controller (FLC) as well as the generic model controller (GMC). A recently developed hybrid model (Hosen et al., 2011a. Asia-Pac. J. Chem. Eng. 6(2), 274) is utilized in the control study to design and tune the proposed controllers. The optimal minimum temperature profiles are determined using the Hamiltonian maximum principle. Different types of disturbances are introduced and applied to examine the stability of controller performance. The experimental studies revealed that the performance of the NN-MPC is superior to that of FLC and GMC. © 2013 The Institution of Chemical Engineers.

An optimized mean variance estimation method for uncertainty quantification of wind power forecasts

A statistical optimized technique for rapid development of reliable prediction intervals (PIs) is presented in this study. The mean-variance estimation (MVE) technique is employed here for quantification of uncertainties related with wind power predictions. In this method, two separate neural network models are used for estimation of wind power generation and its variance. A novel PI-based training algorithm is also presented to enhance the performance of the MVE method and improve the quality of PIs. For an in-depth analysis, comprehensive experiments are conducted with seasonal datasets taken from three geographically dispersed wind farms in Australia. Five confidence levels of PIs are between 50% and 90%. Obtained results show while both traditional and optimized PIs are hypothetically valid, the optimized PIs are much more informative than the traditional MVE PIs. The informativeness of these PIs paves the way for their application in trouble-free operation and smooth integration of wind farms into energy systems. © 2014 Elsevier Ltd. All rights reserved.

Multi-output interval type-2 fuzzy logic system for protein secondary structure prediction

A new multi-output interval type-2 fuzzy logic system (MOIT2FLS) is introduced for protein secondary structure prediction in this paper. Three outputs of the MOIT2FLS correspond to three structure classes including helix, strand (sheet) and coil. Quantitative properties of amino acids are employed to characterize twenty amino acids rather than the widely used computationally expensive binary encoding scheme. Three clustering tasks are performed using the adaptive vector quantization method to construct an equal number of initial rules for each type of secondary structure. Genetic algorithm is applied to optimally adjust parameters of the MOIT2FLS. The genetic fitness function is designed based on the Q3 measure. Experimental results demonstrate the dominance of the proposed approach against the traditional methods that are Chou-Fasman method, Garnier-Osguthorpe-Robson method, and artificial neural network models.

Sistema não-Intrusivo para Estimação da Direção do olhar utilizando redes neurais artificiais

The fundamental senses of the human body are: vision, hearing, touch, taste and smell. These senses are the functions that provide our relationship with the environment. The vision serves as a sensory receptor responsible for obtaining information from the outside world that will be sent to the brain. The gaze reflects its attention, intention and interest. Therefore, the estimation of gaze direction, using computer tools, provides a promising alternative to improve the capacity of human-computer interaction, mainly with respect to those people who suffer from motor deficiencies. Thus, the objective of this work is to present a non-intrusive system that basically uses a personal computer and a low cost webcam, combined with the use of digital image processing techniques, Wavelets transforms and pattern recognition, such as artificial neural network models, resulting in a complete system that performs since the image acquisition (including face detection and eye tracking) to the estimation of gaze direction. The obtained results show the feasibility of the proposed system, as well as several feature advantages.