Advances in Brain Inspired Cognitive Systems

Humans use semantics during parsing; so should computers. In contrast to phrase structure-based parsers, COGPARSE seeks to determine which meaning-bearing components are present in a text, using world knowledge and lexical semantics for construction grammar form selection, syntactic overlap processing, disambiguation, and confidence calculation. In a brain-inspired way, COGPARSE aligns parsing with the structure of the lexicon, providing a linguistic representation, parsing algorithm, associated linguistic theory, and preliminary metrics for evaluating parse quality. Given sufficient information on nuanced word and construction semantics, COGPARSE can also assemble detailed full-semantics meaning representations of input texts. Beyond the ability to determine which parses are most likely to be intended and to use knowledge in disambiguation, full-semantics parsing enables nuanced meaning representation, learning, summarization, natural language user interfaces, and the taking of action based on natural language input.

In human cognition, the capacity to reason and make decisions is strictly dependent on our common sense knowledge about the world and our inner emotional states: we call this ability affective common sense reasoning. In previous works, graph mining and multi-dimensionality reduction techniques have been employed in attempt to emulate such a process and, hence, to semantically and affectively analyze natural language text. In this work, we exploit a novel cognitive model based on the combined use of principal component analysis and artificial neural networks to perform reasoning on a knowledge base obtained by merging a graph representation of common sense with a linguistic resource for the lexical representation of affect. Results show a noticeable improvement in emotion recognition from natural language text and pave the way for more bio-inspired approaches to the emulation of affective common sense reasoning.

A major thrust in building intelligent systems is to encourage users to capitalize their existing real world non-digital skills and seamlessly integrate the representations users have built in real world activities with their ongoing interactions with computing systems. In virtual environments, studies have demonstrated that users treat virtual environments as if they were real even when the environments were a crude approximation to real environments. These results revealed that the feeling of presence is a fluid dynamic psychological state that could vary depending on individual users’ characteristics. The current study aimed to investigate whether individual differences in working memory capacity impact the feelings of presence in virtual environments. Participants performed a vegetable cutting task on behalf of an avatar in a desktop virtual environment. Experiment 1 revealed that high working memory (HWM) individuals cut closer to the fingertip of the avatar than low working memory (LWM) individuals. Experiment 2 used eye-tracking measures and revealed that HWM participants mitigated the demonstrated risk by planning. Taken together, Experiments 1 and 2 provide evidence how individual differences in working memory capacity can impact the feelings of presence in virtual environments and demonstrated that our conscious experiences of what is real is malleable.

Clinical risk assessment of chronic illnesses in the cardiovascular domain is quite a challenging and complex task which entails the utilization of standardized clinical practice guidelines and documentation procedures to ensure clinical governance, efficient and consistent care for patients. In this paper, we present a cardiovascular decision support framework based on key ontology engineering principles and a Bayesian Network. The primary objective of this demarcation is to separate domain knowledge (clinical expert’s knowledge and clinical practice guidelines) from probabilistic information. Using ontologies is a cost effective and pragmatic solution to implementing a shift from simple patient interviewing systems to more intelligent systems in primary and secondary care. The key components of the proposed cardiovascular decision support framework have been developed using an ontology driven approach. We have also utilized a Bayesian Network (BN) approach for modelling clinical uncertainty in the Electronic Healthcare Records (EHRs). The cardiovascular decision support framework has been validated using a sample of real patients’ data acquired from the Raigmore Hospital’s RACPC (Rapid Access Chest Pain Clinic). A variable elimination algorithm has been used to implement the BN Inference and clinical validation of the “Coronary Angiography” treatment has been carried out using Electronic Healthcare Records.

The adoption of Electronic Healthcare Records (EHRs) holds the key for the success of next generation intelligent healthcare systems to improve the quality of healthcare and patient safety by facilitating the exchange of critical patient’s episodic information among different stakeholders. The primary and secondary care healthcare systems store the episodic information for future reuse and for auditing purposes. The conventional healthcare information management systems for primary and secondary care are expected to be able to communicate and exchange complex medical knowledge (often expressed in numerous languages in different parts of the world) in an efficient and unequivocal way. For the purpose of this research, we present a novel technique to transform conventional patients’ data into OWL-based Electronic Healthcare Records (EHRs) which addresses the issues of interoperability, flexibility, and scalability through the utilization of ontology inspired framework. Using ontologies is a cost effective and pragmatic solution to implementing a shift from simple patient interviewing systems to more intelligent systems in the primary and secondary care. The Patient Semantic Profile specifically developed for generating EHRs has been validated using a sample of real patients’ data acquired from the Raigmore Hospital’s RACPC (Rapid Access Chest Pain Clinic).

Hubel and Wiesel’s hypothesis on the emergence of orientation selectivity of simple cells meets some difficulties. It requires the receptive fields of GC and LGN to be highly similar in size and sub-structure while arranged in perfect order. The strict regularities make the model uneconomical in both evolution and neural computation. Varying from the classical model, we propose a new model based on an algebraic method, which estimates orientation by solving constraint satisfaction problems (CSP). The algebraic model needs not to obey the constraints of Hubel and Wiesel’s hypothesis and it is easily implemented as neural network. We also prove that both precision and efficiency of the model are practicable in mathematics. This study is significant in the aspect of explaining the neural mechanism of orientation detection, as well as of finding the circuit structure and computational route in neural network.

In this paper we focus on the effect of different interface designs on the performance and cognitive workload of sensor operators (SO) during a target detection task in a simulated environment. Functional near-infrared (fNIR) spectroscopy is used to investigate whether there is a relationship between target detection performance across three SO interfaces and brain activation data obtained from the subjects’ prefrontal cortices that are associated with relevant higher-order cognitive functions such as attention, response selection and decision making. The preliminary findings of the study suggest that brain regions in the vicinity of medial frontal gyrus of the right hemisphere respond differentially to the cognitive workload induced by different interfaces.

Many BMI (brain machine interface) researches on the control of a prosthetic upper-limb/hand have been conducted. However, the BMI researches on the control of a walking-assistive device were few. Otherwise, brain activation was usually measured in a synchronous control mode. This reduced the naturality of brain activation. To realize asynchronous BMI control of a walking-assistive device, this paper studied cerebral activation pattern in both the spontaneous (asynchronous) and evoked (synchronous) movement states. Stepping and squatting stances movements were performed. Cerebral activation was simultaneously measured using NIRS (near-infrared spectroscopy) technology. Analysis of variation revealed that cerebral activation patterns in the two motion modes had a significant difference in both the imaginary/ preparation periods and movement periods. Particularly, the spontaneous movement achieved a more distinct difference than the evoked movement. It is confirmed that cerebral activation in the preparation periods of spontaneous movement is preferable for identifying motion intention of lower limbs.

A novel soft switching control approach is presented in this paper for autonomous vehicles by using a new functional model for Basal Ganglia (BG). In the proposed approach, a family of fundamental controllers is treated as each of a set of basic controllers are thought of as an ‘action’ which may be selected by the BG in a soft switching regime for real-time control of autonomous vehicle systems. Three controllers, i.e., conventional Proportional-Integral-Derivative (PID) controller, a PID structure-based pole-zero placement controller, and a pole only placement controller are used in this paper to support the proposed soft switching control strategy. To demonstrate the effectiveness of the proposed soft switching approach for nonlinear autonomous vehicle control (AVC), the throttle, brake and steering subsystems are focused on in this paper because they are three key subsystems in the whole AVC system. Simulation results are provided to illustrate the performance and effectiveness of the proposed soft switching control approach by applying it to the abovementioned subsystems.

This paper presents an intelligent multiple-controller framework for the integrated control of throttle, brake and steering subsystems of realistic validated nonlinear autonomous vehicles. In the developed multiple-controller framework, a fuzzy logic-based switching and tuning supervisor operates at the highest level of the system and makes a switching decision on the basis of the required performance measure, between an arbitrary number of adaptive controllers: in the current case, between a conventional Proportional-Integral-Derivative (PID) controller and a PID structure-based pole-zero placement controller. The fuzzy supervisor is also able to adaptively tune the parameters of the multiple controllers. Sample simulation results using a realistic autonomous vehicle model demonstrate the ability of the intelligent controller to both simultaneously track the desired throttle, braking force, and steering changes, whilst penalising excessive control actions - with significant potential implications for both fuel and emission economy. We conclude by demonstrating how this work has laid the foundation for ongoing neuro-biologically motivated algorithmic development of a more cognitively inspired multiple-controller framework.

The ontogenetic process that forms the structure of biological brains is believed to be ruled primarily by optimizing principles of resource allocation and constraint satisfaction for resources such as metabolic energy, wiring length, cranial volume, etc. These processes lead to networks that have interesting macroscopic structures, such as small-world and scale-free organization. However, open questions remain about the importance of these structures in cognitive performance, and how information processing constraints might provide requirements that dictate the types of macro structures observed. Constraints on the physical and metabolic needs of biological brains must be balanced with information processing constraints. It is therefore plausible that observed structures of biological brains are the result of both physical and information processing needs. In this paper we show that small-world structure can evolve under combined physical and functional constraints for a simulated evolution of a neuronal controller for an embodied agent in a navigational task.

In this paper, we propose a novel template updating modeling algorithm inspired by human brain memory model. Three memory spaces are defined according to the human brain three-stage memory theory. The three memory spaces are used to store the current estimated template and the historical templates. To simulate the memorization process of human brain, such as information updating or exchanging, some behaviors and rules are also defined. The proposed memory-based template updating mechanism can remember or forget what the target appearance has ever been, which helps the tracker adapt to the variation of an object’s appearance more quickly. Experimental results show that the proposed algorithm can handle sudden appearance changes and occlusions robustly when tracking moving objects under complex background by particle filter.

A bio-inspired silicon Mixed Signal integrated circuit is designed in this paper to emulate the brain development in Superior Colliculus of barn owl. For the juvenile barn owl, it can adapt localization mismatch to prism wearing. Visual and auditory maps alignment in Superior Colliculus is adjusted. Visual and auditory input information can recover their registration after several weeks’ training. A mathematical model has been built previously to emulate this process. Based on the model, we designed a VLSI circuit in 0.35

μm

CMOS process which has been fabricated. In this paper we present the chip test results of a silicon superior colliculus and show a novel method for adaptive spiking neural information integration when disparity is caused by the environment.

In order to further improve the convergence rate of membrane computing, the membrane computing optimization method based on the catalytic factor (BCMC) is proposed from the inspiration of biological catalyzing enzymes. This algorithm is based on the standard membrane computing, and the catalytic factor is used to control the number of communication objects between membranes, so that the number of communication objects between membrane changes with the change of membrane environment. That is to say, if the average fitness value is relatively larger than the individual fitness value of the membrane, then reduce the number of communication objects of the membrane, conversely, increase the number. In order to test the feasibility and correctness of the algorithm, the simulation test functions are used to simulate, through comparing with the calculated results by using the SGA method, we can see the convergence rate of the membrane computing optimization method based on the catalytic factor is faster and the results are more accurate.

This study focuses on the positional effects on hemodynamic changes monitored by the functional near infrared (fNIR) spectroscopy. The motivation behind this exploratory study is to provide a standard approach for a number of bedside, and postural applications where the body-head position can influence the fNIR signal readings. By administering two consecutive experimental protocols, we investigated effects of the potential body-head positions that may be the cases during sleep and anesthesia recordings. Furthermore dynamic tilting was used to address positional effects from lying to standing up. Positions of supine and tilted are significantly different for HbO

2

and Hb (

p

<

.05

). The natural positions, i.e., sitting, prone, supine, and sideways showed differentiations in the fNIR measures. The deoxygenated hemoglobin values seem to be the least effected component of fNIR recordings across all different positions.

The efficiency and safety of many complex human-machine systems are closely related to the cognitive workload and situational awareness of their human operators. In this study, we utilized functional near infrared (fNIR) spectroscopy to monitor anterior prefrontal cortex activation of experienced operators during a standard working memory and attention task, the n-back. Results indicated that task efficiency can be estimated using operator’s fNIR and behavioral measures together. Moreover, fNIR measures had more predictive power than behavioral measures for estimating operator’s future task performance in higher difficulty conditions.

The goal of the present study is to test whether there is “Arousal” or “Activation” Dysfunction in the frontal region of children with Attention-Deficit/Hyperactivity Disorder (AD/HD). The sample consists of 62 children (31 with AD/HD and 31 non-AD/HD children as controls) who were drawn from an elementary school. Patterns of cortical activity were measured using EEG under three conditions: Eyes-Closed (EC), Eyes-Opened (EO) resting and Mental Arithmetic Task (MAT) conditions, and compared according to AD/HD diagnostic status. Significant main effects for all frequency bands across conditions were found. The AD/HD group showed less elevation of beta relative power than controls suggesting deficiency of cortical activation in the AD/HD group. AD/HD group showed significantly elevated alpha power in eyes-opened resting state. Theta/beta ratio was less reduced for AD/HD than for controls when going from EC to EO to MAT state. The implications of the results were discussed.

In this work a new video database of Italian Sign Language (Lingua Italiana dei Segni - LIS) is proposed. Several other attempts have been made in the literature, but they are typically oriented to international languages (like the American Sign Language - ASL). As in speech, also this kind of language presents different peculiarities strictly depending on the geographical location where it is used. The authors have firstly observed that a specific database for LIS is missing and this shoved them to develop the one here presented. It has been conceived to be used in Automatic Sign Recognition and Synthesis (often referred as Automatic Translation into Sign Languages) applications, which represent an important technological opportunity to augment the social inclusion of people with severe hearing impairments. The Database, namely

A3LIS-147

, is free and available for download.

This study investigates phase relationships between electrocorticogram (ECoG) signals through Hilbert-Huang Transform (HHT), combined with Empirical Mode Decomposition (EMD). We perform spatial and temporal filtering of the raw signals, followed by tuning the EMD parameters. It can be seen that carefully tuning of EMD filter, it is possible to capture distinct features of non-stationary data. This makes EMD, combined with HHT a valuable tool of complex brain signal analysis and modeling.

This study used functional near-infrared spectroscopy (fNIRs) to test the hypothesis that non-treatment seeking alcohol-dependent participants (NTSA) would show greater response in dorsolateral prefrontal cortex (DLPFC) to alcohol cues than recovering alcoholics (RA; sober 90-180 days) or social drinkers. Opposite predictions were made for responses to natural reward cues. NTSA (n=4), RA (n=6), and social drinkers (n=4) were exposed to alcohol and natural reward cues while being monitored with fNIRs. Results confirmed enhanced responses to alcohol cues among NTSA vs. RA in right middle frontal gyrus. The opposite effect (RA>NTSA) was found in response to natural reward cues. Neural responses to alcohol and natural reward cues were negatively correlated in right DLPFC. Real-time craving ratings were positively correlated with greater neural response to alcohol cues. Differential responses to drug and natural reward cues suggest that a psychological mechanism related to treatment status may modulate drug cue responses in DLPFC.

This paper proposes a two-stage filtering method for pressure time series in oil pipelines. First, adopt moving mean filter to smooth the signal contaminated by noise. Second, utilize the result as the input of the discrete or stationary wavelet filter to process the signal without losing its singularities including useful information. By testing this method on real data from oil pipelines, the results demonstrate an excellent performance on filtering the pressure time series and retaining data characteristics.

Studies in event perception have shown that people impose boundaries onto the constant flux of perceptual information and perceive the world to be composed of a series of discrete events. A significant question arises to whether humans impose boundaries onto events that unfold along multiple tracks and perceive them to be one psychological entity (i.e., temporal chunking). The traditional method of event segmentation has difficulties with investigating simultaneous events. The current study investigated whether and how talking about events reveals the psychological event boundaries imposed by perceivers. The current study manipulated the temporal parameters of stimulus events, controlled the causality of events, and thus translated the linguistic differences into measurable properties of events. Participants viewed films of simultaneous events, and performed linguistic acceptability judgments. The results showed there is a correspondence between how people talk about the event sequence and the order in which events occur following event segmentations.

A novel hyperchaotification control method for a class of 3D four-wing chaotic systems is presented in this paper. A simple state feedback is introduced into this kind of chaotic systems as a new variable’s evolving law. By choosing appropriate feedback gain and feedback variable, the generated system can be guaranteed to be dissipative and have two positive Lyapunov exponents. Therefore, the hyperchaos is generated. Simulation study is carried out to verify the effectiveness of the proposed hyperchaotification approach.

We have developed an intelligent agent to engage with users in virtual drama improvisation previously. The intelligent agent was able to perform sentence-level affect detection from user inputs with strong emotional indicators. However, we noticed that many inputs with weak or no affect indicators also contain emotional implication but were regarded as neutral expressions by the previous interpretation. In this paper, we employ latent semantic analysis to perform topic theme detection and identify target audiences for such inputs. We also discuss how such semantic interpretation of the dialog context is used to interpret affect and recognize metaphorical phenomena. Our work contributes to the conference themes on emotion and affect and semantic-based dialogue processing.

This paper proposes a framework for representing the subjective dimension of experience within artificial systems, in particular information systems that emulate behaviour of natural agents. As opposed to the mainstream approach in knowledge engineering it is proposed that knowledge is not equal to experience, in the sense that experience is a broader term which encapsulates both knowledge and subjective, affective component of experience and as such can be represented in formal systems, which has not been so far properly addressed by knowledge representation theories. We show also how our work could enhance the mainstream approach to modelling rational agency with BDI framework.

Self-control is considered one of the strongest predictor of crime. Low self-control has emerged as a consistent and strong predictor of criminal behaviors. Theory and emerging evidence suggest that failing to regulate emotion may result in one’s emotional state being in disorder. Emotional balance may, therefore, be related to self-control or even delinquency. The present study examined emotional balance as a predictor of self-control within prisoners and non-prisoners samples, by using the modified Affect Balance Scale (ABS) and the Self-Control Scale (SCS). Five forms of self-control were assessed: Impulse control (IC), Healthy habit (HH), Resist temptation (RT), Focus on work (FW) and Restrained amusement (RA). However, emotional balance emerged as a significant predictor of only impulse control, after controlling for age, marital status, income and education. Notably, the high emotional balance level was most predictive of increased impulse control capability. Finally, these results would be helpful to preventive interventions of delinquency, criminal or social deviance behaviors.

In Sensorimotor Contingency Theory (SMCT) differences between the perceptual qualities of sensory modalities are explained by the different structure of dependencies between a human’s actions and the ensuing changes in sensory stimulation. It distinguishes modality-related Sensory-Motor Contingencies (SMCs), that describe the structure of changes for individual sensory modalities, and object-related SMCs, that capture the multisensory patterns caused by actions directed towards objects. These properties suggest a division of time scales in that modality-related SMCs describe the immediate effect of actions on characteristics of the sensory signal, and object-related SMCs account for sequences of actions and sensory observations. We present a computational model of SMCs that implements this distinction and allows to analyze the properties of the different SMC types. The emergence of perceptual capabilities is demonstrated in a locomotive robot controlled by this model that develops an action-based understanding for the size of its confinement without using any distance sensors.

The birefringent properties of a new type of Total Internal Reflection Photonic Crystal Fiber(TIR-PCF) filled symmetrically with magnetic fluid in the holes are studies by using the full-vector finite element method.To improve numerical precision, the perfectly matched layer is used as an absorbing boundary condition in computing. Theoretical calculations show that it may exhibit high birefringence in a properly designed PCF (whose birefringence can be as high as 0.05), and the birefringence can be tuned by magnetic fields or the structure parameter of the PCF. The birefringence effect is ten times higher than the general fiber and there is a fairly good linearity. This scheme provides theoretical foundation to use magnetic field to control light in PCF and also offers a potential method for making high-birefringent polarization fiber.

Sparse representation has been validated to be a common phenomenon in many sensory neural systems, but its underlying neural mechanism still remains unclear. This paper proposes a neurally plausible model towards solving this problem. We find that lateral inhibition is the fundamental neural mechanism for sparse representation in the visual cortex, by which cortical neurons not only compete with each other so that the input signal can be represented sparsely but also cooperate with each other to make the representation more accurate. We integrate this result into the matching pursuit framework, a quite suitable solution for neural implementation, to illustrate how an input signal is sparsely represented in V1. Our simulation results show that lateral inhibition can evidently decrease the average squared error in the representation and then the input signal can be sparsely represented very well by the proposed algorithm.

Global asymptotic stability problem for a general class of higher order recurrent neural networks (HRNN) with multiple delays has been studied based on delay-matrix decomposition method and linear matrix inequality (LMI) technique. The proposed stability criterion is suitable for a general class of multiple delayed higher order recurrent Neural Networks. Especially, for this system, we have also established corresponding LMI-based stability criteria which are simple in expression form and easy to check to deal with the different multiple delays. Compared with the existing results, our results are new and can be regarded as an alternative of M-matrix based stability results in the literature.

Circulating Fluidized Bed Boiler (CFBB) involves a kind of combustion boiler that can clean and desulfurize the coal efficiently in the combustion process. It is highly adapted to all kinds of high quality coals and low grade coals. CFB boiler has more superior performance, features and a wide range of applications than other boilers, accordingly, problems on automatic control underlined in the combustion process blocked its wide application. This is a typical thermal object, hard to control,and due to the special combustion type of CFBB that makes it a great inertia, multivariable, strong coupling, nonlinear, time-varying object. This paper designs an ART2-BP-BP Hybrid Neural Network of fusion cluster control system and completes data fusion from the data level, the feature level to the decision level. Results of simulation show that the control system in this paper is feasible and effective, in particular, the control system still has more satisfactory control effects in the case of a variety of sensor failures.

A new single neuron adaptive Proportional-Integral-Derivative (PID) controller based on Levenberg-Marquardt (LM) algorithm is presented in this paper. This new controller overcomes some drawbacks of the conventional single neuron adaptive PID controllers. There are two kinds of problems in traditional algorithms. Firstly, gradient descent algorithm is a one-order optimization method. Secondly, Newton iterative method costs much computing resource. For the improved controller, LM algorithm is applied, which combines steepest gradient descent and Gauss-Newton method. As a consequence, the convergence speed is increased and the control performance is greatly improved. The simulation results show that the control effect of this novel controller has strong robustness and good self-adaptation.

Based on the echo interference in audio transmission of Network Camera (IPC), this paper presents Variable Step Length Best Combination acoustic echo cancellation (AEC) algorithm .It adjusts the convergence speed of variance by changing the step of the adaptive filter , in accordance with the best ratio combination of NLMS algorithm and RLS algorithm, to achieve faster convergence speed and good steady state characteristics of a short delay. Simulation and measurement under the network environment show that, the residual echo signal is more than 30dB lower than the far-end signal, delays 25ms below the steady-state characteristics.

Stochastic resonance has been revealed as an important cue for understanding the performance of biological systems in detecting external weak signal. Here we show that the noise originated from signal phase may also induce stochastic resonance in an excitable neuronal system. We find that the neuronal system is better at detecting a subthreshold signal when the signal phase is not constant but time-varying like noise. Further, we find that an intermediate intensity of phase noise may help the neuronal system to obtain a optimal detection, forming an resonance-like phenomenon. Finally, a brief theory is formulated to explain the mechanism behind the resonance behavior.

Emotion recognition has aroused great concern recently. Physiological signals show its objective in the field of emotion recognition. This paper introduces emotion recognition system and physiological signals processing. The recognition system can be divided into four sections: signal preprocessing, biological feature extraction, feature matching to and feature classification. For each part, we studied existed methods and discussed their performance and characteristics. Lastly, the trend of emotion recognition for physiological signals was given.

In [1] we proposed two methods to identify the reference electrode signal under the key assumption that the reference signal is independent from EEG sources. This assumption is shown to be possibly true for intracranial EEG with a scalp reference. In this paper, we theoretically prove that the obtained reference signal by using the second method in [1] or the equivalent MPDR approach [1] outperforms the widely used average reference (AR) if the real reference is independent from EEG sources. The simulation results confirm the advantages over AR.

Understanding cognition has fascinated many neuroscientists and made them put their efforts in deciphering the brain’s information processing capabilities for cognition. Rodents perceive the environment through whisking during which tactile information is processed at the barrel columns of the somatosensory cortex (S1). The intra– and trans–columnar microcircuits in the barrel cortex segregate and integrate information during activation of this pathway. Local Field Potentials (LFPs) recorded from these barrel columns provide information about the microcircuits and the shape of the LFPs provide the fingerprint of the underlying neuronal network. Through a contour based sorting method, we could sort neuronal evoked LFPs recorded using high–resolution Electrolyte–Oxide–Semiconductor Field Effect Transistor (EOSFET) based neuronal probes. We also report that the latencies and amplitudes of the individual LFPs’ shapes vary among the different clusters generated by the method. The shape specific information of the single LFPs thus can be used in commenting on the underlying neuronal network generating those signals.

By the analysis of Odd symmetry error Function blind equalization Algorithm based Decision Feedback Equalizer (OFA-DFE), Variable Momentum Factor Momentum Odd symmetry error Function blind equalization Algorithm based Decision Feedback Equalizer(VMFMOFA-DFE)is proposed. The proposed algorithm uses error function with characteristics of odd symmetry to reduce mean square error, in order to further improve the performance of the algorithm using variable factors to control the momentum term and introducing variable momentum factor to decision feedback equalizer to adjust farword equalizer of decision feedback. Simulation tests with underwater acoustic channel indicate that the proposed algorithm has not only faster convergence rate but also less mean square error.

The implementation methods of Artificial Neural Network (ANN) can be classified into two types: hardware implementation methods and software implementation methods. The former can build truly distributed, parallel, high-speed ANN, but it is complicated and expensive. The latter can build ANNs on classical computer using software technology. It’s necessary in some cases, although the performance of ANNs built by this type of methods is limited because they are not parallel computing. In this paper, we propose a distributed implementation method based on multi-agent theory. Neurons in ANN are distributed, and the ANN can be easily to extend.

Predicted Mean Vote (PMV) is the most widely-used index for evaluating the thermal comfort in buildings. But, this index is calculated through complicated iterations so that it is not suitable for real-time applications. To avoid complicated iterative calculation, this paper presents a prediction model for this index. The proposed model utilizes type-2 fuzzy neural network to approximate the input-output characteristic of the PMV model. To tune the parameters of this type-2 fuzzy neural prediction model, a hybrid algorithm which is a combination of the least square estimate (LSE) method and the back-propagation (BP) algorithm is provided. Finally, simulations are given to verify the effectiveness of the proposed prediction model.

IBM Research and five leading universities are partnering to create computing systems that are expected to simulate and emulate the brain’s abilities. Although this project has achieved some successes, it meets great difficulties in the further research. The main difficulty is that it is almost impossible to analyze the dynamic character of neural networks in detail, when more than ten thousands neurons of complex nonlinear neural models are piled up. So it is nature to present such question: in order to simplify the design of

brain-like computers

, can we use simple neuron models to design

brain-like computers

or can we find a simplest neuron model which can simulate most neuron models with arbitrary precision? In this paper, we proved that almost all neural models found by neural scientists nowadays can be simulated by Hopfield neural networks. So it is possible to use simple neuron model to design

Brain-like computers.

In an effort to exploring the wealth of individual and social signals conveyed by human facial cues, collecting a high quality facial database is a resource-intensive yet important task, not to mention the manual labeling the captured emotional facial expressions, which can be error prone and expensive. To date, most facial expression analysis has been based on the databases which are, however, often plagued for limited scale, lack of flexibility and static, etc. Furthermore, many existing facial expression databases are even lack of categorization and detailed organization, not to mention the functional analysis. A comprehensive survey then become necessary for current analysis and future design. This paper surveys the current representative facial expression databases, we have analyzed and categorized facial expression databases according to the functional and non-functional attributes. Our survey provides a basis for comparison of existing databases. In doing so, we assist the readers to gain insights into the technology, strategies, and practices that are currently followed in this field.

Pesticides are used for controlling pests, but at the same time they have impacts on the environment as well as the product itself. Although cotton covers 2.5% of the world’s cultivated land yet uses 16% of the world’s insecticides, more than any other single major crop [1]. Pakistan is the world’s fourth largest cotton producer and a major pesticide consumer. Numerous state run organizations have been monitoring the cotton crop for decades through pest-scouting, agriculture surveys and meteorological data-gatherings. This non-digitized, dirty and non-standardized data is of little use for strategic analysis and decision support. An advanced intelligent Agriculture Decision Support System (ADSS) is employed in an attempt to harness the semantic power of that data, by closely connecting visualization and data mining to each other in order to better realize the cognitive aspects of data mining. In this paper, we discuss the critical issue of handling data anomalies of pest scouting data for the six year period: 2001-2006. Using the ADSS it was found that the pesticides were not sprayed based on the pests crossing the critical population threshold, but were instead based on centuries old traditional agricultural significance of the weekday (Monday), thus resulting in non optimized pesticide usage, that can potentially reduce yield.

Facial expression recognition is one of the hot spots in recent years, it applies in the emotional analysis, pattern recognition and interpersonal interaction. This paper introduces the recent advances and applications in facial expression recognition from the face detection, feature extraction, classification, and the ethnic expression recognition. The methods of feature extraction are divided to several different characteristic categories. Researches of classifications are based on space or time and space. What’s more, according to the facial expression recognition history and achievements, the development of ethnic facial expression recognition and the trend of facial expression recognition are given.

Considering the important reference value of weather radar data in the process of weather modification and based on the analysis of historical weather modification data of Yunnan province, china, and the study of various non-precipitation removal algorithms, this paper proposed a new model of intelligent identification and elimination of non-precipitation radar echoes in the environment of low-latitude plateaus, which has been proven to be effective in practical use.

This paper describes an efficient approach towards road sign detection and recognition. The proposed system is divided into three sections namely; Colour Segmentation of the road traffic signs using the HSV colour space considering varying lighting conditions, Shape Classification using the Contourlet Transform considering occlusion and rotation of the candidate signs and the Recognition of the road traffic signs using features of a Local Energy based Shape Histogram (LESH). We have provided three experimental results and a detailed analysis to justify that the algorithm described in this paper is robust enough to detect and recognize road signs under varying weather, occlusion, rotation and scaling conditions using video stream.