Online sequential extreme learningmachine (OS-ELM) proposed by Liang et al. employ sequential learning strategy to learn the target concept from the data. Compared with the original ELM, OS-ELM can learn data one-by-...
详细信息
ISBN:
(纸本)9781479938414
Online sequential extreme learningmachine (OS-ELM) proposed by Liang et al. employ sequential learning strategy to learn the target concept from the data. Compared with the original ELM, OS-ELM can learn data one-by-one or chunk-by-chunk with fixed or varying chunk size with almost same performance as ELM. While compared with other state-of-the-art sequential algorithms such as SGBP, RAN and GAP-RBF, OS-ELM has faster learning speed and better generalization ability. However, similar to ELM, OS-ELM also has instability in different trials of simulations. In addition, for large data sets, OS-ELM will not halt when there are training samples not be learned, this phenomenon results in long learning time. In order to deal with the problems, this paper proposes an algorithm named E-OS-ELM for integrating OS-ELM to classify large data sets. The experimental results show that the proposed method is effective and efficient;it can effectively overcome the drawbacks mentioned above.
Average neighborhood margin maximization(ANMM) is a feature extraction method to make homogeneous points collect as near as possible and heterogeneous points disperse as far away as possible. To enhance the anti-noise...
详细信息
ISBN:
(纸本)9781479937097
Average neighborhood margin maximization(ANMM) is a feature extraction method to make homogeneous points collect as near as possible and heterogeneous points disperse as far away as possible. To enhance the anti-noise ability of ANMM, correntropy based average neighborhood margin maximization(CANMM) is proposed in this paper. This method utilizes correntropy to substitute the Euclidean distance for measuring the similarity between the given data, and uses the maximum correntropy criterion to replace the maximum distance criterion, which makes CANMM more robust. The experimental results on three benchmark face databases validate the effectiveness of the proposed method.
Libraries are recently changing their classical role of providing stored information into new virtual communities, which involve large number of users sharing real time information. Despite of those good features, the...
详细信息
Libraries are recently changing their classical role of providing stored information into new virtual communities, which involve large number of users sharing real time information. Despite of those good features, there is still a necessity of developing tools to help users to reach decisions with a high level of consensus in those new virtual environments. In this contribution we present a new consensus reaching tool with linguistic preferences designed to minimize the main problems that this kind of organization presents (low and intermittent participation rates, difficulty of establishing trust relations and so on) while incorporating the benefits that a new digital library offers (rich and diverse knowledge due to a large number of users, real-time communication and so on). The tool incorporates some delegation and feedback mechanisms to improve the speed of the process and its convergence towards a consensual solution.
Patch-level features are essential for achieving good performance in computer vision tasks. Besides well-known pre-defined patch-level descriptors such as scaleinvariant feature transform (SIFT) and histogram of orien...
详细信息
It is quite inadequate in providing formula retrieval function by traditional retrieval techniques used in full-text information retrieval system. The main reason is that there are many difficulties to extract the key...
详细信息
In this paper, we propose a new method based on Chinese keyword search to select the WAV or MP3 files in audio post-production. First, we listen to each file and label it with Chinese characters, and then classify and...
详细信息
Multi-label text classification deals with problems in which each document is associated with a subset of categories. These documents often consist of a large number of words, which can hinder the performance of learn...
详细信息
Multi-label text classification deals with problems in which each document is associated with a subset of categories. These documents often consist of a large number of words, which can hinder the performance of learning algorithms. Feature selection is a popular task to find representative words and remove unimportant ones, which could speed up learning and even improve learning performance. This work evaluates eight feature selection algorithms in text benchmark datasets. The best algorithms are subsequently compared with random feature selection and classifiers built using all features. Results agree with literature by finding that well-known approaches, such as maximum chi-squared scoring across all labels, are good choices to reduce text dimensionality while reaching competitive multi-label classification performance.
This paper proposes to employ a detailed tumor growth model to synthesize labelled images which can then be used to train an efficient data-driven machinelearning tumor predictor. Our MR image synthesis step generate...
详细信息
Pathfinding is a typical task in many computer games, and its performance will affect the quality of game AI. In order to enhance the efficiency of multi-task pathfinding, case-based reasoning has been introduced in t...
详细信息
Pathfinding is a typical task in many computer games, and its performance will affect the quality of game AI. In order to enhance the efficiency of multi-task pathfinding, case-based reasoning has been introduced in traditional A* algorithm, called the CBMT method. The method needs to select representative paths which can cover the whole map to build a compact case base, which is difficult in large maps. Besides, repeatedly searching for similar cases for each pathfinding task would be a time consuming process. To address these problems, we provide a kd-tree case storage structure and case retrieval mechanical in the CBMT method. The pre-stored cases(previously found paths) are generated randomly and incrementally. The original flat storage structure of the cases is changed into the kd-tree structure. Since the searching space can be reduced by branch pruning in case retrieval, the pathfinding efficiency has been improved obviously, and the number of searched nodes is also reduced.
As the kernel component of scientific documents, mathematical expressions are becoming a new object of searching engines. Different from normal text, mathematical expressions are composed of various kinds of symbols a...
详细信息
ISBN:
(纸本)9781479905607
As the kernel component of scientific documents, mathematical expressions are becoming a new object of searching engines. Different from normal text, mathematical expressions are composed of various kinds of symbols arranged in nonlinear mode, which results in the limitations of traditional full-text information retrieval used for expression searching. In this paper, we discuss the existing search engine of mathematical expressions and introduce the two-dimensional characteristics of mathematical expressions firstly. Then, a data structure of expressing mathematical formulas is designed which contains not only the symbol code but also the mathematical information among symbols. Finally, the indexing algorithm of mathematical expressions is put forward on the basis of the expression data structure. The experimental result shows the effectiveness of the indexing method proposed in this paper.
暂无评论