This paper introduces a novel approach that integrates graph theory into self-supervised representation learning. Traditional methods focus on intra-instance variations generated by applying augmentations. However, they often overlook important inter-instance relationships. While our method retains the intra-instance property, it further captures inter-instance relationships by constructing$k$-nearest neighbor (KNN) graphs for both teacher and student streams during pretraining. In these graphs, nodes represent samples along with their latent representations. Edges encode the similarity between instances. Following pretraining, a representation refinement phase is performed. In this phase, Graph Neural Networks (GNNs) propagate messages not only among immediate neighbors but also across multiple hops, thereby enabling broader contextual integration. Experimental results on CIFAR-10, ImageNet-100, and ImageNet-1K demonstrate accuracy improvements of 7.3%, 3.2%, and 1.0%, respectively, over state-of-the-art methods. These results highlight the effectiveness of the proposed graph-based mechanism. The code is publicly available athttps://github.com/alijavidani/SSL-GraphNNCLR.

In this paper, a fast, intelligent model is proposed for the order determination of linear dynamical systems by using convolutional neural networks. This model estimates the dynamic order of the system with considerably lower excitation order of stimulation signal and without any prior knowledge in comparison to former works. To this end, only step response of the system is taken to estimate the dynamic order for both stable and unstable linear systems. Unlike the conventional methods, in this deep-based approach, the order determination is performed quickly, automatically, at a low cost, and without any iteration. In addition, it is demonstrated that the proposed approach has low sensitivity against delay and noise. Such an intelligent model can satisfy the demands for a fast identifier in online and plug-and-play controllers.

There is an increasing demand to develop fast and reliable models to identify the class of control systems in developing online and plug-and-play controllers. In this paper, to perform automatic, reliable, and fast classification of linear processes, it is proposed to use Convolutional Neural Networks (CNNs). A process can be: unstable or stable, integrally or self-regulated, non-minimum phase or minimum phase, first-order or second-order, oscillatory damping, or over-damping. We consider six different classes of linear processes accordingly. The CNN is designed and trained to predict the class of linear processes by taking only their step responses. The results show clearly that the CNN has high generalization and accuracy in determining the behavior class of the process even in the presence of noise, delay, and high order dynamics.

This paper suggests evolving linear models as a powerful alternative for online forecasting of synchronous time series. First, using a priori knowledge of an expert, all known possible features are clustered in some categories so that each category includes homogeneous features. Then, a novel evolving correlation-based forward subset selection technique is used to determine relevant features from each category. Next, based on the selected features, an estimation of the output of the system is modeled through an evolving adaptive linear regression model. In evolving systems, selected features and their associated weights could vary over time based on new incoming data samples which contain new information. Finally, a soft combination of output estimations of all categories, in a new sense of Takagi-Sugeno fuzzy system, gives the prediction of the output of the system at each sampling time. The approach offers a certain new view at the enhancement of evolving forecasting models. The proposed approach embodies recursive learning and one-step-ahead incremental algorithms that progressively modify the model to ensure continuous learning, and self-organization of the model structure and its parameters. Two real-world problems, forecasting electricity load of the Electric Reliability Council of Texas region and stock price forecasting of technology sector of Standard & Poor's 500 index, are provided to validate the developed method. Pros and cons of the proposed approach are comprehensively discussed and shown through simulation results and comparisons with other state-of-the-art techniques.

We present a simple yet effective learning technique that significantly improves mAP of YOLO object detectors without compromising their speed. During network training, we carefully feed in localization information. We excite certain activations in order to help the network learn to better localize (Figure 2). In the later stages of training, we gradually reduce our assisted excitation to zero. We reached a new state-of-the-art in the speed-accuracy trade-off (Figure 1). Our technique improves the mAP of YOLOv2 by 3.8% and mAP of YOLOv3 by 2.2% on MSCOCO dataset. This technique is inspired from curriculum learning. It is simple and effective and it is applicable to most single-stage object detectors.

Social robots are becoming a companion in everyday life. To be well accepted by humans, they should efficiently understand meanings of their partners' motions and body language and respond accordingly. Learning concepts by imitation brings them this ability in a user-friendly way. This paper presents a fast and robust model for incremental learning of concepts by imitation (ILoCI). In ILoCI, observed multimodal spatiotemporal demonstrations are incrementally abstracted and generalized based on their perceptual and functional similarities during the imitation. Perceptually similar demonstrations are abstracted by a dynamic model of the mirror neuron system. The functional similarities of demonstrations are also learned through a limited number of interactions with the teacher. Incremental relearning of acquired concepts together through memory rehearsal enables the learner to gradually extract and utilize the common structural relations among demonstrations to expedite the learning process especially at the initial stages. Performance of ILoCI is assessed using a standard benchmark dataset and a human-robot interaction task in which a humanoid robot learns to abstract teacher's hand motions during imitation. Its performance is also evaluated on occluded observations that are probable in real environments. The results show efficiency of ILoCI in concept acquisition, recognition, prediction, and generation in addition to its robustness to occlusions and high variability in observations.

Complexity of computations, particularly due to large number of focal elements (FEs), in Dempster-Shafer theory (DST) motivates the development of approximation algorithms. Existing approximation methods include efficient algorithm for special hypothesis space, Monte Carlo based techniques, and simplification procedures. In this paper, the quality of the simplification-based approximation algorithms is evaluated using a new information-based comparison methodology. To this end, three structured testbeds are introduced. Each testbed is designed with an eye on a particular form of uncertainty associated with a body of evidence (BoE) in DST, i.e., conflict and non-specificity. Three proposed testbeds along with the classic testbed are utilized to evaluate the accuracy and complexity of existing algorithms. In light of the proposed evaluation methodology, a new approximation method is presented as well. The proposed algorithm has the ability to choose the most informative FEs without being forced to select the FEs with the largest mass function. Comparison of overall qualitative performance of approximation algorithms provides accuracy versus computational time tradeoff to choose an appropriate approximation method in different applications. Moreover, experiments with testbeds indicate that our proposed algorithm enhances the accuracy and computational tractability simultaneously.

Recently, online input selection has gained an increasing attention in evolving fuzzy models. In this paper, we proposed a new evolving fuzzy system referred to as evolving heterogeneous fuzzy inference system (eHFIS), which can simultaneously perform local input selection and system identification in an evolving and integrative manner. The introduced eHFIS is structured by some fuzzy rules with different effective input variables. This was achieved through inclusion of some parameters (local input selectors) in the structure of the Takagi-Sugeno system. An online learning algorithm is proposed to identify the eHFIS, where 1) the premise parameter learning and rule evolution take place with the usage of an incremental and evolving clustering for partitioning the data space; 2) a local input selection strategy based on switching to a neighboring model is adopted, and then, all fuzzy rules with the same input structure form a new category; and 3) for each category, the parameters of linear models, in consequent parts, are updated by weighted recursive fuzzily weighted least-squares estimator. The performance of the proposed eHFIS is evaluated and compared through several simulations on hand made as well as real-life datasets.

In this paper, by using of the idea that occurring face data may be generated by sampling a probability distribution that has support on or near a sub-manifold of ambient space, we propose the nonlinear method named MKDA based on neighborhood discriminant projection method, for feature extraction and face recognition in which geometric relations are preserved according to prior class-label information and complex nonlinear variations of face images are represented by nonlinear kernel mapping. Experiments on ORL, UMIST, FERET, YALE and CMU-PIE face databases are performed to test and evaluate the proposed algorithm by using some different methods. Experiments indicate the promising performance of the proposed method.

In general, imitation is imprecisely used to address different levels of social learning from high-level knowledge transfer to low-level regeneration of motor commands. However, true imitation is based on abstraction and conceptualization. This article presents a model for conceptual imitation through interaction with the teacher to abstract spatio-temporal demonstrations based on their functional meaning. Abstraction, concept acquisition, and self-organization of proto-symbols are performed through an incremental and gradual learning algorithm. In this algorithm, Hidden Markov Models (HMMs) are used to abstract perceptually similar demonstrations. However, abstract (relational) concepts emerge as a collection of HMMs irregularly scattered in the perceptual space but showing the same functionality. Performance of the proposed algorithm is evaluated in two experimental scenarios. The first one is a human-robot interaction task of imitating signs produced by hand movements. The second one is a simulated interactive task of imitating whole body motion patterns of a humanoid model. Experimental results show efficiency of our model for concept extraction, proto-symbol emergence, motion pattern recognition, prediction, and generation.

Representing an image as a set of its key and interesting lines facilitates the image understanding and classification. In this paper, we propose a method to extract the significant and interesting lines of the scene, which probably are useful in image classification. The proposed method is inspired from the Visual Attention, which is a perceptual mechanism in human and other primates that direct their perceptions to the limited regions of the scene. The attended regions are usually valuable in performing the task. Since the approach of using the lines to classify the images is particularly useful for omnidirectional images, we specialize our method to deal with these kinds of images. In the experiments, we demonstrate how our proposed methods improve the image classification performance with processing only small parts of the input images.

In this letter, we propose a learning system, active decision fusion learning (ADFL), for active fusion of decisions. Each decision maker, referred to as a local decision maker, provides its suggestion in the form of a probability distribution over all possible decisions. The goal of the system is to learn the active sequential selection of the local decision makers in order to consult with and thus learn the final decision based on the consultations. These two learning tasks are formulated as learning a single sequential decision-making problem in the form of a Markov decision process (MDP), and a continuous reinforcement learning method is employed to solve it. The states of this MDP are decisions of the attended local decision makers, and the actions are either attending to a local decision maker or declaring final decisions. The learning system is punished for each consultation and wrong final decision and rewarded for correct final decisions. This results in minimizing the consultation and decision-making costs through learning a sequential consultation policy where the most informative local decision makers are consulted and the least informative, misleading, and redundant ones are left unattended. An important property of this policy is that it acts locally. This means that the system handles any nonuniformity in the local decision maker's expertise over the state space. This property has been exploited in the design of local experts. ADFL is tested on a set of classification tasks, where it outperforms two well-known classification methods, Adaboost and bagging, as well as three benchmark fusion algorithms: OWA, Borda count, and majority voting. In addition, the effect of local experts design strategy on the performance of ADFL is studied, and some guidelines for the design of local experts are provided. Moreover, evaluating ADFL in some special cases proves that it is able to derive the maximum benefit from the informative local decision makers and to minimize attending to redundant ones.

A recursive extension of Gath-Geva clustering algorithm is proposed in this paper which is used as a basis for online tuning and development of neuro-fuzzy models. In comparison with other online modeling approaches which use spherical clusters for defining validity region of neurons, the proposed evolving neuro-fuzzy model (ENFM) has the ability to take advantage of elliptical clusters. This extension increases the ability of local linear neurons of ENFM to capture system behavior in more sophisticated regions which leads to decrease in number of neurons as well as increase in the modeling ability. The proposed model is capable of adapting to changes in system behavior by adding new neurons or merging similar existing fuzzy rules. Efficiency of evolving neuro-fuzzy model is investigated in prediction of Mackey-Glass and smoothed sunspot number time series. Results of these simulations show better performance of the proposed model as compared with other online modeling approaches.

With growing attention to ensemble learning, in recent years various ensemble methods for face recognition have been proposed that show promising results. Among diverse ensemble construction approaches, random subspace method has received considerable attention in face recognition. Although random feature selection in random subspace method improves accuracy in general, it is not free of serious difficulties and drawbacks. In this paper we present a learning scheme to overcome some of the drawbacks of random feature selection in the random subspace method. The proposed learning method derives a feature discrimination map based on a measure of accuracy and uses it in a probabilistic recall mode to construct an ensemble of subspaces. Experiments on different face databases revealed that the proposed method gives superior performance over the well-known benchmarks and state of the art ensemble methods.

This paper introduces a new method for learning top-down and task-driven visual attention control along with physical actions in interactive environments. Our method is based on the Reinforcement Learning of Visual Classes(RLVC) algorithm and adapts it for learning spatial visual selection in order to reduce computational complexity. Proposed algorithm also addresses aliasings due to not knowing previous actions and perceptions. Continuing learning shows our method is robust to perturbations in perceptual information. Our method also allows object recognition when class labels are used instead of physical actions. We have tried to gain maximum generalization while performing local processing. Experiments over visual navigation and object recognition tasks show that our method is more efficient in terms of computational complexity and is biologically more plausible.

Designing an intelligent situated agent is a difficult task because the designer must see the problem from the agent's viewpoint, considering all its sensors, actuators, and computation systems. In this paper, we introduce a bio-inspired hybridization of reinforcement learning, cooperative co-evolution, and a cultural-inspired memetic algorithm for the automatic development of behavior-based agents. Reinforcement learning is responsible for the individual-level adaptation. Cooperative co-evolution performs at the population level and provides basic decision-making modules for the reinforcement-learning procedure. The culture-based memetic algorithm, which is a new computational interpretation of the meme metaphor, increases the lifetime performance of agents by sharing learning experiences between all agents in the society. In this paper, the design problem is decomposed into two different parts: 1) developing a repertoire of behavior modules and 2) organizing them in the agent's architecture. Our proposed cooperative co-evolutionary approach solves the first problem by evolving behavior modules in their separate genetic pools. We address the problem of relating the fitness of the agent to the fitness of behavior modules by proposing two fitness sharing mechanisms, namelyuniformandvalue-basedfitness sharing mechanisms. The organization of behavior modules in the architecture is determined by our structure learning method. A mathematical formulation is provided that shows how to decompose the value of the structure into simpler components. These values are estimated during learning and are used to find the organization of behavior modules during the agent's lifetime. To accelerate the learning process, we introduce a culture-based method based on our new interpretation of the meme metaphor. Our proposed memetic algorithm is a mechanism for sharing learned structures among agents in the society. Lifetime performance of the agent, which is quite important for real-world applications, increases considerably when the memetic algorithm is in action. Finally, we apply our methods to two benchmark problems: an abstract problem and a decentralized multirobot object-lifting task, and we achieve human-competitive architecture designs.

This comment tries to describe a theoretical mistake made in the aforementioned paper [G. Feng,IEEE Trans. Fuzzy Syst., vol. 11, no. 5, pp. 605-612, Oct. 2003] to formulate the inverse of matrices used to construct the piecewise-quadratic Lyapunov functions. Derivation of these inverse matrices is the most critical step toward transforming the design constraints into linear matrix inequalities (LMIs). Therefore, the erroneous formulation essentially affects the validity of the approach and final results. Unfortunately, it seems that there is no simple correction for this problem. However, some close alternative approaches are suggested by the same authors in their other works.

In this paper, we propose a new approach to identify a neuro-fuzzy model. In our approach, data space is partitioned indirectly through a fuzzy clustering method. The clusters are not created directly through spatial features of data points. A gradient vector is defined as major feature of clustering in data space. This feature is estimated for each incoming data points. Creating and updating fuzzy membership functions, adding new clusters and removing redundant clusters are performed through it. Correspond with cluster parameters, fuzzy rules are defined and a neuro-fuzzy model is identified recursively. Prediction of monthly sunspots number is considered to demonstrate the capability of the proposed neuro-fuzzy model.

Similar to humans and primates, artificial creatures like robots are limited in terms of allocation of their resources to huge sensory and perceptual information. Serial processing mechanisms used in the design of such creatures demands engineering attentional control mechanisms. In this paper, we present a new algorithm for learning top-down sequential visual attention control for agents acting in interactive environments. Our method is based on the key idea, that attention can be learned best in concert with visual representations through automatic construction and discretization of the visual state space. The tree representing the top-down attention is incrementally refined whenever aliasing occurs by selecting the most appropriate saccadic direction. The proposed approach is evaluated on action-based object recognition and urban navigation tasks, where obtained results support applicability and usefulness of developed saccade movement method for robotics.

Most of the central pattern generator (CPG) models are based on defining explicit dynamical systems and finding the appropriate parameters. In this paper, we propose a novel CPG model that is based on altering a nonlinear oscillator to obtain desired limit cycle behavior. This CPG model benefits from an explicit basin of attraction and also fast convergence behavior. The presented CPG model is used in an imitation model that tries to learn the proper periodical behavior by looking at a mentor. First, a mentor performs the desired periodical behavior. Then, a hand-eye coordination process, inspired from infant babbling, is initiated to extract proper motor actions from what is observed. The extracted motor actions are finally embedded into the CPG model for smooth reproduction. This imitation model is implemented on a robotic marionette behavior learning task. The outcome of the final performance of the robotic marionette is behaviorally understandable smooth actions.

In this paper, we propose a fast algorithm for gesture recognition based on the saliency maps of visual attention. A tuned saliency-based model of visual attention is used to find potential hand regions in video frames. To obtain the overall movement of the hand, saliency maps of the differences of consecutive video frames are overlaid. An improved characteristic loci feature extraction method is introduced and used to code obtained hand movement. Finally, the extracted feature vector is used for training SVMs to classify the gestures. The proposed method along a hand-eye coordination model is used to play a robotic marionette and an approval/rejection phase is used to interactively correct the robotic marionette's behavior.

This paper presents the implementation of an intelligent pass strategy for soccer robots using fuzzy logic. The proposed strategy calculates pass confidence and detects the best destination robot and best position for passing. The most important factors of a proper pass are detected and are deployed to construct the inputs of fuzzy system. The rule base of the fuzzy system is extracted by utilizing the expert knowledge. The performance of fuzzy inference system is improved by optimally determining the fuzzy membership functions using an evolutionary algorithm. In order to evaluate the designed system a visual simulation environment is prepared which enables the user to arrange players and to implement the designed strategy to detect the best destination player and send the ball.

This paper presents a constructive Takagi-Sugeno fuzzy modeling method for a general class of nonlinear systems. This method is particularly suitable for stability analysis based on piecewise quadratic Lyapunov functions. The modeling error is appropriately inserted into the model and an algorithm is proposed to automatically determine the model parameters to keep the modeling error smaller than a desired upper bound. Based on the constructed fuzzy model, exponential stability analysis is performed and the stability constraints are transformed into linear matrix inequalities. Modeling error is also included in the stability analysis to validate the results for the original nonlinear system. The way to utilize the modeling method and stability analysis to systematically find a Lyapunov function for a nonlinear system is demonstrated via an example and the potential capability of the method in estimating the domain of attraction is discussed.

Radial Basis Function Networks (RBFNs) are widely used in curve-fitting problems and nonlinear dynamical systems modelling. Using the gradient of the function during the training phase leads to a smooth approximation of both the function itself, and its derivatives. The knowledge about gradient of the function in some identification and control tasks is desired, particularly when the stability and robustness of the system are studied. In this paper, a new clustering based algorithm for learning an Input-Output map along with its derivatives using RBFN is proposed. The input-output clustering (IOC) algorithm for the training of an RBFN is modified to improve the performance of the network in approximating a nonlinear single-input single-output map along with its derivatives utilizing a set of input-output data and the first derivative of the function in each data point. The advantage of the proposed algorithm, in comparison with orthogonal least square (OLS), is demonstrated with an example in the field of data interpolation.

It is generally agreed that faces are not recognized only by utilizing some holistic search among all learned faces, but also through a feature analysis that aimed to specify more important features of each specific face. This paper addresses a novel decision strategy that efficiently uses both holistic and facial component (left eye, right eye, nose and mouth) feature analysis to recognize faces. The proposed algorithm uses the whole face features in the first step of recognition task. If the decision machine fails to assign a class (with high confidence) then the individual facial components are processed and the resulting information are combined with those obtained from the whole face to assign the output. Simulation studies justify the superior performance of the proposed method as compared to that of Eigenface method. Experimental results also show that the proposed system is robust against small errors in facial component extractor.

In this paper, first an enhanced NeuroFuzzy method for modeling nonlinear system is presented. In this method we use EM algorithm for identification of local models, which gain us model mismatch covariance. The achieved model can be stated in state space model as a linear time-varying system. As the noise and model mismatch covariance is known, Kalman filter can be easily used for centralized estimation fusion. The simulations show that using centralized estimation fusion will enhance the estimation accuracy to a great deal.

Mutual information has been used in many clustering algorithms for measuring general dependencies between random data variables, but its difficulties in computing for small size datasets has limited its efficiency for clustering in many applications. A novel clustering method is proposed which estimates mutual information based on information potential computed pair-wise between data points and without any prior assumptions about cluster density function. The proposed algorithm increases the mutual information in each step in an agglomerative hierarchy scheme. We have shown experimentally that maximizing mutual information between data points and their class labels will lead to an efficient clustering. Experiments done on a variety of artificial and real datasets show the superiority of this algorithm, besides its low computational complexity, in comparison to other information based clustering methods and also some ordinary clustering algorithms.

In this paper a novel algorithm for structure identification of Takagi-Suegno (TS) fuzzy models based on split and merge clustering is purposed. In this algorithm, by using a sequential split procedure on data space, initial Gaussian functions as constructor blocks are created. By merging these initial blocks, new composite validity functions for locally linear models with a high degree of flexibility are estimated. The proposed algorithm results in TS-type locally linear fuzzy models with an abstract structure as well as high generalization. Desirable performance of this algorithm is illustrated at case study section.

The cyclic solar activity has significant effects on earth, climate, satellites and space missions. Several methods have been introduced for the prediction of sunspot number, which is a common measure of solar activity. In this study a co-evolutionary algorithm is presented for inferring the topology and parameters of a multilayered neural network with the minimum of experimentation to the sunspot number time series which will be used as a predictor in predicting such phenomena. The algorithm synthesizes an explicit model directly from the observed data produced by intelligently generated tests. The algorithm is composed of two co-evolving populations; one population evolves candidate neural networks. The second population evolves informative tests that either extract new information from the hidden system or elicit desirable behavior from it. The fitness of candidate neural networks is their ability to explain behavior of the target chaotic system observed in response to all tests carried out so far; the fitness of candidate tests is their ability to make the models disagree in their predictions. The generality of this modeling-evaluation algorithm is demonstrated by applying the chosen model of this algorithm to predict sunspot number and the results depict the power of this training method which yields proper model to predict such chaotic time series.

This paper presents a systematic design procedure of fuzzy controllers for exponentially stabilizing affine nonlinear systems, which are subject to vanishing perturbations. The modeling error, as a key factor in the validity of the stability analysis is also taken into consideration. A unified framework is presented, dealing with the modeling error as the system perturbation. It is shown that the constraints imposed on the modeling error are always satisfied for affine nonlinear systems, when they are modeled by Takagi- Sugeno fuzzy model through currently available methods. The constraints required to guarantee the exponential stability of the original nonlinear system and design the controller are transformed into LMIs. Finally, the way to utilize the presented method for a stabilizing problem is demonstrated using two examples.

Cooperation in learning (CL) can be realized in a multiagent system, if agents are capable of learning from both their own experiments and other agents' knowledge and expertise. Extra resources are exploited into higher efficiency and faster learning in CL as compared to that of individual learning (IL). In the real world, however, implementation of CL is not a straightforward task, in part due to possible differences in area of expertise (AOE). In this paper, reinforcement-learning homogenous agents are considered in an environment with multiple goals or tasks. As a result, they become expert in different domains with different amounts of expertness. Each agent uses a one-step Q-learning algorithm and is capable of exchanging its Q-table with those of its teammates. Two crucial questions are addressed in this paper: "How the AOE of an agent can be extracted?" and "How agents can improve their performance in CL by knowing their AOEs?" An algorithm is developed to extract the AOE based on state transitions as a gold standard from a behavioral point of view. Moreover, it is discussed that the AOE can be implicitly obtained through agents' expertness in the state level. Three new methods for CL through the combination of Q-tables are developed and examined for overall performance after CL. The performances of developed methods are compared with that of IL, strategy sharing (SS), and weighted SS (WSS). Obtained results show the superior performance of AOE-based methods as compared to that of existing CL methods, which do not use the notion of AOE. These results are very encouraging in support of the idea that "cooperation based on the AOE" performs better than the general CL methods.

Designing an intelligent situated agent is a difficult task as the designer must see the problem from the agent's standpoint considering all its sensors and actuators. We have devised a co-evolutionary/reinforcement learning hybrid method to automate the design of hierarchical behavior-based systems. In our approach, the design problem is decomposed into two separate parts: developing a repertoire of behaviors and organizing those behaviors in a structure. Mathematical formulation shows how to decompose the value of the structure to simpler components. These components can be estimated and used to find the optimal organization of behaviors during the agent's lifetime. Moreover, a novel co-evolutionary mechanism is suggested that evolves each type of behavior separately in their own genetic pool. Our method is applied to the decentralized multi-robot object lifting task which results in human-competitive performance.

This paper presents a novel method for detecting multiple frontal faces in still images using multi-scale processing. The main characteristic of this algorithm is its stability in detecting faces with seldom false detections and a high correct detection rate. The novelty of this work comes from the utilization of multiscale detection and using two classifiers to reduce false detections. The algorithm generally has two stages: in the first stage, a face is detected in a unique scale and in the second stage, only the faces that are located in the neighbor scales are accepted as real faces. Consequently, a still image is first resized and scanned block wise, and then each enhanced block is tested for being face. One dimensional Harr wavelet is used for feature extraction, which gives appropriate discriminating features between the face and nonface classes. Detection results at each scale are accumulated in an internal database, so the ultimate detection is prepared based on the mutual detection information between consequent scales. To parameterize both the Bayesian and the simple proposed classifier, 2,643 faces were congregated from famous face databases and more than 10,000 non-face samples were selected from nature images. Experimental results using images gathered from known databases like MIT-CMU show great ability of the proposed algorithm in detecting faces.

Behavior-based systems have been successfully used in control and robotics applications. In traditional behavior-based systems, only a single behavior controls the agent in any time step. However, this behavior arbitration is not appropriate for many complex tasks. In this paper, we propose Hierarchical Soft Behavior-based Architecture that uses the concept of soft suppression to coordinate flexibly between behaviors. In our method, we use reinforcement learning to find an appropriate amount of suppression for each behavior in the architecture, in addition to learn the internal mechanism of each behavior. Several experiments are provided to show the effectiveness of our method in the mobile robot navigation task.

Using each other's knowledge and expertise in learning - what we call cooperation in learning- is one of the major existing methods to reduce the number of learning trials, which is quite crucial for real world applications. In situated systems, robots become expert in different areas due to being exposed to different situations and tasks. As a consequence, areas of expertise (AOE) of the other agents must be detected before using their knowledge, especially when the exchanged knowledge is not abstract, and simple information exchange might result in incorrect knowledge, which is the case for Q-learning agents. In this paper we introduce an approach for extraction of AOE of agents for cooperation in learning using their Q-tables. The evaluating robot uses a behavioral measure to evaluate itself, in order to find a set of states it is expert in. That set is used, then, along with a Q-table-based feature for extraction of areas of expertise of other robots by means of a classifier. Extracted areas are merged in the last stage. The proposed method is tested both in extensive simulations and in real world experiments using mobile robots. The results show effectiveness of the introduced approach, both in accurate extraction of areas of expertise and increasing the quality of the combined knowledge, even when, there are uncertainty and perceptual aliasing in the application and the robot

NnSP is a stream-based programmable and code-level statically reconfigurable processor for realization of neural networks in embedded systems. NnSP is provided with a neural-network-to-stream compiler and a hardware core builder. The NnSP stream compiler makes it possible to realize various neural networks using NnSP. On the other hand, the NnSP builder makes the NnSP processor an IP core that can be restructured to satisfy different demands and constraints. This paper presents the architecture of the NnSP processor, the streaming mechanism, and the builder facilities. Also, synthesis results of a 64-PE NnSP on a 0.18 mum standard-cell library are presented. The obtained results show that a 64-PE NnSP can perform computations of 25.6 giga connections in a second, while its throughput is upto 51.2 giga 32-bit fixed point operations per second. Comparing with high performance parallel architectures locates 64-PE NnSP among the best state of the art parallel processors

This paper proposes a cooperative evolutionary method for optimizing the properties of an ANFIS-architecture-based model where only the input-output data of the identified system are available. The primary tasks of fuzzy modeling are structure identification and parameter optimization: the former determines the numbers of membership functions and fuzzy if-then rules while the latter identifies a feasible set of parameters under the given structure. The proposed approach manages all mentioned attributes simultaneously. Particularly, number of rules and parameters of membership functions are realized by applying a novel approach using genetic programming and genetic algorithm whereas consequent parameters are tuned by using least-squares estimation. Finally, two examples of nonlinear system are given to illustrate the effectiveness of the proposed approach.

Imitation equips robots with a simple and natural interface to learn new tasks. Although abstraction is a remarkable feature of imitation that discriminates it from mimicking, there has been no enough research on this dimension of imitation. Relational concepts are the simplest type of abstract concepts and can be an appropriate start point. These concepts may be learned by combining perceptual categorization and classical conditioning. The paper will first formalize relational concept learning within an imitative context. Internal modules of the learning agent are considered to be functions. We will prove that in this case the concept-motor mapping becomes one-to-one which simplifies learning. A learning algorithm for the model will be also proposed and evaluated in a phoneme acquisition experiment with a large number of highly overlapped samples.

Iris detection is a crucial part of an iris recognition system. One of the main issues in iris segmentation is coping with occlusions that happen due to eyelids and eyelashes. In This work, only the lower part of the iris is utilized for recognition. Wavelet based texture features along with a mixed Hamming; harmonic mean distance classifier is used for identification. It is observed that relying in a smaller but more reliable part of the iris, though reducing the net amount of information, improves the overall performance. Experimental results on CASIA database show that the method has a promising performance with an accuracy of more than 99%.

This paper proposes a new method of face representation which is used for face recognition by SVM. For face representation we have used a two-step method, first two-dimensional discrete wavelet transform (DWT) is used to transform the faces to a more discriminated space and then principal component analysis (PCA) is applied. The proposed method produced a significant improvement which includes a substantial reduction in error rate and in time of processing during the obtaining PCA orthonormal basis.

Many real world applications would involve grasp of large objects in unstructured environments. Agent-based approach to multi-robot grasp of objects would prove useful under the above circumstances. In this paper, the problem of form closure grasp for planar convex objects by multiple robots is tackled. Contrary to the previous approaches, no a priori information about the shape of the object is assumed, and the robots are not allowed to fully communicate among themselves. A distributed multi-agent based approach using Q-learning is proposed. The state space, action set and learning algorithm are formulated. The results are verified through simulations using a developed Q-learning test bed.

Hand-design of an intelligent agent's behaviors and their hierarchy is a very hard task. One of the most important steps toward creating intelligent agents is providing them with capability to learn the required behaviors and their architecture. Architecture learning in a behavior-based agent with subsumption architecture is considered in this paper. Overall value function is decomposed into easily calculate-able parts in order to learn the behavior hierarchy. Using probabilistic formulations, two different decomposition methods are discussed: storing the estimated value of each behavior in each layer, and storing the ordering of behaviors in the architecture. Using defined decompositions, two appropriate credit assignment methods are designed. Finally, the proposed methods are tested in a multi-robot object-lifting task that results in satisfactory performance.

The field of robotics is currently undergoing a change toward creation of robots that can naturally interact with humans. For achieving this, interactive robots must be endowed with natural interfaces that can sense and respond in real-time. Vision can provide handy information for this purpose by detecting and tracking human limbs to analyze gestures, actions and even emotions. However, real-time processing of visual information is a challenging bottleneck. In this paper, we introduce a novel method, namely "self-organized contours", that can distinctly accelerate contour initialization, which is the slowest phase in visual tracking. Although the proposed method is general-purpose, it allows immediate initialization of active contours due to its similarity with snake structure. The proposed method is inspired from group behavior in insects and animals, particularly fishes.

Individual identification of dolphins is of much interest to marine mammologists. Natural markings and notches on the trailing edge of the dorsal fin provide an almost unique signature within a study population for individual recognition. The dorsal fin is an almost rigid, planar object, and, as a result, the problem boils down to planar curve matching. Invariants under three groups of similarity, affine and projective transformations are examined in this paper. Three families of differential, algebraic and semi-differential invariants are studied and evaluated for their applicability and performance in dolphin identification from photographs. It is observed that, under practical conditions, for smooth quasi-rigid curves like the trailing edge of the dorsal fin, curvature-based matching methods outperform almost all of the other techniques which offer a higher level of invariance.

Elastography is a new imaging modality which provides an image of tissue elastic properties. Simulated elastograms are used to study the ability of textural features to discriminate between abnormalities with different levels of hardness and density. Elastograms are simulated for different signal processing parameters as well as for a non-uniform stress field. 15 selected texture features are extracted. The Gaussian classifier does not perform well due to the complexity of clusters. Hence, a neural network classifier is trained and tested to discriminate between the classes. The results indicate that the texture features perform well under variations of the signal processing parameters and non-uniformity of the applied stress field. This simulation study provides a guiding mechanism to discriminate between malignant and benign tissue abnormalities once clinical data becomes available.

The Dempster-Shafer theory (DST) may be considered as a generalization of the probability theory, which assigns mass values to the subsets of the referential set and suggests an interval-valued probability measure. There have been several attempts for fuzzy generalization of the DST by assigning mass (probability) values to the fuzzy subsets of the referential set. The interval-valued probability measures thus obtained are not equivalent to the original fuzzy body of evidence. In this paper, a new generalization of the DST is put forward that gives a fuzzy-valued definition for the belief, plausibility, and probability functions over a finite referential set. These functions are all equivalent to one another and to the original fuzzy body of evidence. The advantage of the proposed model is shown in three application examples. It can be seen that the proposed generalization is capable of modeling the uncertainties in the real world and eliminate the need for extra preassumptions and preprocessing.