Search

This paper presents three-layer dynamic binary neural networks characterized by ternary connection parameters and the signum activation function. The dynamics is described by a difference equation of binary state variables. Depending on the parameters, the network can generate various binary periodic orbits. We give two main theoretical results. First, when a desired periodic orbit is given, we can set the parameters that guarantee storage and local stability of the periodic orbit. The stability is related to error correction of various binary signals in engineering applications. Second, if a part of the connection parameters becomes zero then stability of the periodic orbit becomes very strong. In this case, all the initial states fall directly into the periodic orbit. [ABSTRACT FROM AUTHOR]

With the growing penetration of wind power, the wind power forecasting is fundamental in aiding the grid scheduling and electricity trading. In this paper, a numerical weather prediction wind speed error correction model based on gated recurrent unit neural networks is proposed for short-term wind power forecasting. Firstly, the standard deviation of numerical weather prediction wind speed error is extracted as weights, and these weights are rearranged according to the numerical weather prediction wind speed time series to get the weight time series. Then, the bidirectional gated recurrent unit neural networks based error correction model is proposed to correct error of numerical weather prediction wind speed with the inputs as numerical weather prediction wind speed, trend and detail terms of the weight time series. The wind power curve model is applied to forecast short-term wind power by using corrected numerical weather prediction wind speed. Finally, the effectiveness of the proposed method is compared with benchmark models by using actual data of wind farm, and the results show that the proposed model outperforms these benchmark models. [ABSTRACT FROM AUTHOR]

In this paper, we present some results on error detection and correction in a modular neurocomputer that are based on redundant residue number systems. The error correction method developed below involves the modified Chinese Remainder Theorem with fractions and uses a Hopfield neural network to correct the errors. The suggested approach eliminates the need for extending the bases of a residue number system, a costly operation required in case of syndrome decoding with error syndromes calculation on the control bases of the system. Also the approach does not utilize the projection method, another costly operation intended to localize errors (i.e., to detect the moduli associated with faulty digits). The well-known procedures mentioned above seem inefficient in terms of practical implementation, as they employ a mixed radix number system: transition to such a system is iterative and may affect the performance of a whole neurocomputer. Owing to the exclusion of these costly operations, the suggested approach significantly simplifies error correction procedures for integer numbers. [ABSTRACT FROM AUTHOR]

Occlusion is a common yet challenging problem in face recognition. Most of the existing approaches cannot achieve the accuracy of the recognition with high efficiency in the occlusion case. To address this problem, this paper proposes a novel algorithm, called efficient locality-constrained occlusion coding (ELOC), improving the previous sparse error correction with Markov random fields (SEC_MRF) algorithm. The proposed approach estimates and excludes occluded region by locality-constrained linear coding (LLC), which avoids the time-consuming l 1 -minimization and exhaustive subject-by-subject search during the occlusion estimation, and greatly reduces the running time of recognition. Moreover, by simplifying the regularization, the ELOC can be further accelerated. Experimental results on several face databases show that our algorithms significantly improve the previous algorithms in efficiency without losing too much accuracy. [ABSTRACT FROM AUTHOR]

In this paper, we propose the architecture of a fault-tolerant unit in a modular neurocomputer that is based on decoding with computation of errors syndromes on redundant moduli and implemented using FPGA and a finite ring neural network. The computational complexity of the proposed architecture is about 80% less in comparison with that of the architecture based on number projections in the mixed radix number system. [ABSTRACT FROM AUTHOR]

In this paper, an enhanced M-ary SVM algorithm in combination with error correction coding method is presented for multi-category classification. The main process of the approach is divided into three steps. It is firstly to generate a group of best codes based on information codes derived from the original category flags information. Secondly, it is to utilize such codes as the basis for training the classifier. The third step is the final feed-forward phase. The output codes composed of each sub-classifier are corrected by error detection and correction principle whenever any identifying error occurs. The improved algorithm not only maintains the highly simplified architecture of the standard M-ary algorithm, but also improves its generalization ability. The experimental results confirm s the effectiveness of the improved algorithm brought about by introducing as few sub-classifiers. Finally, an application example for shuttle kiln control system is shown. [ABSTRACT FROM AUTHOR]

In this paper, we propose a new online learning method for measuring affinity between tracklets in multi-target tracking. As targets and background usually keep changing in the video, fixed affinity measurement could not adapt to their variations. Most existing affinity learning methods construct labeled samples based on the obtained tracklets, and then minimize a predefined loss function to get an optimal affinity measurement. However, those methods simply assume that the training error equals to testing error which is not true in many of real time tracking scenarios. Differently, we propose to learn affinity measurement through CovBoosting, which considers the evolution of the tracklets and could obtain affinity measurement with more discriminative ability. To deal with targets׳ disappearance and new targets׳ appearance, we combine tracklet affinity with contextual information to do an optimal inference. Moreover, an online updating algorithm is developed to guarantee that the learned tracklet affinity is always optimal for tracking targets in current sliding window. Experimental results on benchmark datasets demonstrate that tracklet affinity learned with our method is more discriminative and could greatly improve the performance of the multi-target tracker. [ABSTRACT FROM AUTHOR]

The classical AdaBoost algorithm is an ensemble of weak learners, and can be used to construct a strong classifier. A weak learner is incorporated into the ensemble at each step, and the classification of the derived ensemble is improved by properly adjusting the weight of each weak learner. The classical AdaBoost algorithm has some limitations; for example, it is sensitive to noisy data, which may impede the generalization capability of the derived classifier and lead to overfitting problem. A selective boosting, sBoost, technique is proposed in this paper to tackle these problems. The proposed sBoost classifier focuses on the generalized classification performance rather than those hard-to-learn samples, and the penalties of hard-to-learn samples are mitigated to the degree associated with their noise level. An error correction method is suggested to detect potential clean samples and prevent them from misclassification to further alleviate the overfitting problem. The effectiveness of the developed sBoost technique is tested by a series of simulation tests. Test results show that the developed sBoost technique can improve the classification accuracy and prevent the overfitting problem effectively. [ABSTRACT FROM AUTHOR]