Online Incremental Learning Algorithm Research Articles

OKO-SVM: Online kernel optimization-based support vector machine for the incremental learning and classification of the sentiments in the train reviews

Online incremental learning is one of the emerging research interests among the researchers in the recent years. The sentiment classification through the online incremental learning faces many challenges due to the limitations in the memory and the computing resources available for processing the online reviews. This work has introduced an online incremental learning algorithm for classifying the train reviews. The sentiments available in the reviews provided for the public services are necessary for improving the quality of the service. This work proposes the online kernel optimization-based support vector machine (OKO-SVM) classifier for the sentiment classification of the train reviews. This paper is the extension of the previous work kernel optimization-based support vector machine (KO-SVM). The OKO-SVM classifier uses the proposed fuzzy bound for modifying the weight for each incoming review database for the particular time duration. The simulation uses the standard train review and the movie review database for the classification. From the simulation results, it is evident that the proposed model has achieved a better performance with the values of 84.42%, 93.86%, and 74.56% regarding the accuracy, sensitivity, and specificity while classifying the train review database.

Between Exploration and Exploitation in Motor Babbling

Motor babbling allows an agent sampling trajectory data without a priori knowledge about self-body and environment dynamics. We discuss about the efficiency of motor babbling through the example of drawing task. In authors’ insight, natural motor babbling may be featured by exploration and exploitation processes. From this idea, we propose exploitation babbling and e-greedy babbling. In order to implement the proposed babblings, we developed dynamics learning tree (DLT). DLT is an online incremental learning algorithm that has constant calculation order O(1). The proposed exploitation babbling and e-greedy babbling improved the rate of effective data at 8 and 7 % from previous babbling respectively. e-greedy babbling converged its prediction error fastest among the three babblings. Using e-greedy babbling, a humanoid robot with wired flexible fingers successfully drew a figure without a priori knowledge about the dynamics among self-body, pen, and pen tablet.

FPGA implementation of a new interval type-2 Beta neuro-fuzzy system with on-chip learning for image denoising application

Hardware implementation of conventional interval type-2 neuro-fuzzy systems with on-chip learning is essential for real time applications. However, existing implementations are resource consuming due to the complexity of their architectures and the use of iterative procedure for system output estimation. To overcome this problem, we propose a new interval type-2 neuro-fuzzy architecture. Accordingly, the number of layers is reduced owing to using Beta membership functions. Moreover, a simplified output computing operation is applied. For implementing Beta functions, an accurate and compact Centered Recursive Interpolation (CRI) method is used. For on-chip learning system, a new on-line incremental learning algorithm with gradient descent technique is applied to adjust its parameters. Furthermore, a synthesis of the corresponding design on a Field Programmable Gate Array (FPGA) platform is achieved in image denoising application. Performances comparison with the existing implementations shows the effectiveness of our chip in terms of resource requirements, speed and denoising performances.

Time series learning network using fast dynamic time warping

Mining time series data has been revived in the last decade due to the increasing availability of time series datasets. This paper presents an online incremental learning algorithm for time series data using fast dynamic time warping (FastDTW), called time series learning network (TSLN), which is a neural network model for online incremental learning of time series. It automatically learns appropriate prototypes from input data series which are of equal or unequal length, and it does not ask to predefine the amount of prototypes and similarity thresholds. Besides, TSLN allows the arising of the new prototype, which is much superior to other existing unsupervised neural network algorithm. We test our model with UCR time series datasets, and experimental results show that, from the respect of classification accuracy, the proposed TSLN gets better or approximate results relative to the state-of-the-art approaches. Meanwhile, TSLN greatly mitigates demanding CPU time, which is much faster than most of the state-of-the-art methods.

Evolving granular neural networks from fuzzy data streams

This paper introduces a granular neural network framework for evolving fuzzy system modeling from fuzzy data streams. The evolving granular neural network (eGNN) is able to handle gradual and abrupt parameter changes typical of nonstationary (online) environments. eGNN builds interpretable multi-sized local models using fuzzy neurons for information fusion. An online incremental learning algorithm develops the neural network structure from the information contained in data streams. We focus on trapezoidal fuzzy intervals and objects with trapezoidal membership function representation. More precisely, the framework considers triangular, interval, and numeric types of data to construct granular fuzzy models as particular arrangements of trapezoids. Application examples in classification and function approximation in material and biomedical engineering are used to evaluate and illustrate the neural network usefulness. Simulation results suggest that the eGNN fuzzy modeling approach can handle fuzzy data successfully and outperforms alternative state-of-the-art approaches in terms of accuracy, transparency and compactness.

Learning fuzzy rules from iterative execution of games

This paper discusses the linguistic knowledge extraction from the iterative execution of a multi-player non-cooperative repeated game. Linguistic knowledge is automatically extracted in the form of fuzzy if–then rules. Our knowledge extraction is mainly based on the on-line incremental learning of fuzzy rule-based systems. In this sense, our linguistic knowledge extraction is the learning of fuzzy rules. We first briefly describe a market selection game, which is formulated as a non-cooperative repeated game with many players and several alternative actions. We also describe some simple strategies for our market selection game. In our market selection game, the payoff of each player depends on the actions of all players. When a particular action is chosen by many players, those players receive low payoff. High payoff is obtained from actions chosen by only a small number of players. This means that minority players with respect to their actions receive high payoff. Next we show how our market selection game can be handled as a pattern classification problem where a single training pattern is successively generated from every round of our game. A fuzzy rule-based classification system is used as a decision-making system by each player for choosing an action in every round. An on-line incremental learning algorithm is proposed for adjusting the fuzzy rule-based classification system. Then we show how our market selection game can be handled as a function approximation problem. A fuzzy rule-based approximation system is used as a value function for approximating the expected payoff from each action. Finally simulation results show that comprehensible linguistic knowledge is extracted by the learning of fuzzy rule-based systems.