With the advent of high-dimensional stored big data and streaming data, suddenly machine learning on a very large scale has become a critical need. Such machine learning should be extremely fast, should scale up easil...
详细信息
With the advent of high-dimensional stored big data and streaming data, suddenly machine learning on a very large scale has become a critical need. Such machine learning should be extremely fast, should scale up easily with volume and dimension, should be able to learn from streaming data, should automatically perform dimension reduction for high-dimensional data, and should be deployable on hardware. Neural networks are well positioned to address these challenges of large scale machine learning. In this paper, we present a method that can effectively handle large scale, high-dimensional data. It is an online method that can be used for both streaming and large volumes of stored big data. It primarily uses Kohonen nets, although only a few selected neurons (nodes) from multiple Kohonen nets are actually retained in the end;we discard all Kohonen nets after training. We use Kohonen nets both for dimensionality reduction through feature selection and for building an ensemble of classifiers using single Kohonen neurons. The method is meant to exploit massive parallelism and should be easily deployable on hardware that implements Kohonen nets. Some initial computational results are presented.
In order to improve the performance of text classification and information retrieval in big data of electric power domain, we propose a novel Chinese language classification algorithm-De-word classification algorithm....
详细信息
ISBN:
(纸本)9789401796187;9789401796170
In order to improve the performance of text classification and information retrieval in big data of electric power domain, we propose a novel Chinese language classification algorithm-De-word classification algorithm. Focusing on the key role played by the De-word in modern Chinese language, this algorithm examines Chinese text classification method from a unique angle. Besides, on the basis of traditional weighted algorithm, it designs a novel relevance weighting model-De-TFIDF, and achieves a higher correlation in text information retrieval. Experiments show that, De-word classification algorithm significantly improves the efficiency of text classification, significantly improved information retrieval performance.
To solve the unascertainty in water environmental risk assessment, a classification algorithm that based on unascertained mathematics is established is proposed. And the unascertained information of pollutant concentr...
详细信息
ISBN:
(纸本)9781538653739
To solve the unascertainty in water environmental risk assessment, a classification algorithm that based on unascertained mathematics is established is proposed. And the unascertained information of pollutant concentration in water environment systems, population health life and body weight are considered in the algorithm. Among it, ammonia nitrogen (NH3N) and arsenic (As) are selected as a non-carcinogenic chemical indicator and a chemical carcinogen index respectively. Then the risk degree and risk index can be calculated separately according to the proposed risk level classifications. The case study in Taoge River System shows that the proposed algorithm is capable of evaluating the health risk level in water environment and could provide useful information for the regional water environmental management.
To meet the demands on wideband unidirectional antennas in distinguishing ICH and IS stroke based on classification algorithm, a novel folded antenna is presented. The design of the antenna starts with a planar diplo ...
详细信息
ISBN:
(纸本)9781538653739
To meet the demands on wideband unidirectional antennas in distinguishing ICH and IS stroke based on classification algorithm, a novel folded antenna is presented. The design of the antenna starts with a planar diplo antenna with bridge-shape ground. Then some slots are added to the ground to expand the bandwidth. Next, a folding technology is applied to the monopole to enhance its directivity while the overall size is reduced markedly at the same time. At last, a reflector and a pair of top-facing p-shape parasitic patches are applied to the antenna to increase the gain and decrease the voltage standing wave ratio (VSWR) at 1.7GHz. The final designed antenna has a dimensions of 0.293 lambda x0.107 lambda x0.053. (lambda is the wavelength at the lowest operating frequency). It has a measured reflection coefficient less than -10dB fractional bandwidth of 76% (1.1-2.45 GHz) and a peak gain of 4.5 dBi. In order to validate the effectiveness of the proposed antenna in the application, the proposed folded antenna is simulated with the human brain model and the result is classified by PCA and LDA classification algorithms. Finally, the accuracy of the final classification result is 100%, which indicates that the proposed folded antenna can be well applied to the brain stroke detection system.
This paper proposes the classification algorithm of news pages based on domain Ontology. In order to improve the shortage of current classification algorithm that only considers the content similarity, this paper pres...
详细信息
ISBN:
(纸本)9783037857519
This paper proposes the classification algorithm of news pages based on domain Ontology. In order to improve the shortage of current classification algorithm that only considers the content similarity, this paper presents the semantic classification method which considers both content similarity and structural correlation. Firstly, it parses the Ontology to get Ontology category vector, extracts keywords of news pages' texts and drops semantic dimension. At this time, finding out the same vocabulary and ontology category vector in page texts to constitute the text expectation vector, and then calculating the content similarity between ontology category vector and expectation vector of text by using the law of cosines. Secondly, the common vocabularies are mapped to the ontology hierarchy chart, and the structural relevancy is obtained by calculating weighted path of this directed acyclic graph. Finally, it calculates the correlation degree of the news pages and Ontology by combining both, and determines the category of news pages by judging the size relationship between the result and the initial threshold value.
A novel classification algorithm, OCEC, based on evolutionary computation for data mining is proposed. It is compared to GA-based and non GA-based algorithms on 8 datasets from the UCI machine learning repository. Res...
详细信息
ISBN:
(纸本)0780374886
A novel classification algorithm, OCEC, based on evolutionary computation for data mining is proposed. It is compared to GA-based and non GA-based algorithms on 8 datasets from the UCI machine learning repository. Results show OCEC can achieve higher prediction accuracy, smaller number of rules and more stable performance.
As a social network, microblog has obtained great attention and gotten wide application. Applications of microblog need to retrieve quickly information with the support of real-time search technology in order to imple...
详细信息
ISBN:
(纸本)9783037858646
As a social network, microblog has obtained great attention and gotten wide application. Applications of microblog need to retrieve quickly information with the support of real-time search technology in order to implement information sharing. A query classification algorithm of microblog for real-time search was put forward. Based on question classification mechanism, the algorithm divides queries into two categories: the candidate queries and the popular queries, and takes separate storage strategy. Test results show that the classification algorithm can reduce real-time search time and improve the efficiency of retrieval.
In this paper, we study on the problem of statistical machine translation (SMT) for English language, and the main innovation is that we introduce the SVM classifier to solve this problem. Particularly, the process of...
详细信息
ISBN:
(纸本)9781509004645
In this paper, we study on the problem of statistical machine translation (SMT) for English language, and the main innovation is that we introduce the SVM classifier to solve this problem. Particularly, the process of statistical machine translation is constructed of two modules, that is, 1) constructing the target language sentence, and 2) translating from target texts to source texts language. Afterwards, framework of the statistical machine translation system is provided. The main idea of statistical machine translation is in that the sentence with highest probability is selected by optimizing an objective function, which describes the relationships between source language sentences and target language sentences. Finally, experimental results demonstrate that our proposed algorithm outperforms Topic-Based SMT and Rule Based SMT.
This study explores the integration of supervised machine learning using decision tree algorithms within the framework of the Structure of Observed Learning Outcomes for diagnosing and customizing learning pathways in...
详细信息
ISBN:
(纸本)9798350379068;9798350379051
This study explores the integration of supervised machine learning using decision tree algorithms within the framework of the Structure of Observed Learning Outcomes for diagnosing and customizing learning pathways in middle school mathematics education. Focusing on seventh-grade students' proficiency in numbers and operations, the research employs a large dataset of student responses to develop a real-time adaptive diagnostic tool. The tool classifies students into five proficiency levels-starter, basic, medium, high, and advanced-based on their responses. Initial findings demonstrate an overall classification accuracy of 83%, with further analysis revealing specific strengths and weaknesses across different proficiency levels. This research underscores the potential of supervised machine learning to enhance educational diagnostics and contribute to personalized learning experiences, suggesting that such technology can significantly improve educational outcomes by dynamically adjusting to individual student needs.
This article proposes a design and implementation scheme for an intelligent community product service system based on classification algorithms. The system aims to solve the problem of product service recommendation a...
详细信息
暂无评论