【摘要】：針對傳統(tǒng)文本分類算法準確率低和正確率分布不均勻的問題,提出了基于深度學習的文本分類算法。深度信念網(wǎng)絡具有強大的學習能力,可以從高維的原始特征中提取高度可區(qū)分的低維特征,不僅能夠更全面的考慮到文本信息量,而且能夠進行快速分類。采用TF-IDF方法計算文本特征值,利用深度信念網(wǎng)絡構造分類器進行精準分類。實驗結果表明,與支持向量機、神經(jīng)網(wǎng)絡和極端學習機等常用分類算法相比,該算法有更高的準確率和實用性,為文本的分類研究開拓了新思路。
[Abstract]:Aiming at the problems of low accuracy and uneven distribution of correct rate in traditional text classification algorithms, a text classification algorithm based on depth learning is proposed. Deep belief network has a strong learning ability, it can extract highly distinguishable low-dimensional features from high-dimensional original features, which can not only take into account the amount of text information more comprehensively, but also can be classified quickly. The TF-IDF method is used to calculate the text eigenvalues and the depth belief network is used to construct a classifier for accurate classification. The experimental results show that this algorithm has higher accuracy and practicability than other common classification algorithms such as support vector machine, neural network and extreme learning machine, and opens up a new idea for text classification research.
【作者單位】： 東北林業(yè)大學信息與計算機工程學院;
【基金】：中央高�；究蒲袠I(yè)務費專項資金(2572015DY07) 黑龍江省自然科學基金(F201347) 哈爾濱市科技創(chuàng)新人才專項資金(2013RFQXJ100) 國家自然科學基金(61300098) 教育部大學生創(chuàng)新訓練計劃項目(201510225043)
【分類號】：TP391.1

【相似文獻】

相關期刊論文 前10條

1 鄭智捷;幻序合并分類算法[J];計算機學報;1984年05期

2 劉t，

本文編號：2342154