Scholarlink

<p>Multi-view learning aims to improve learning capability by leveraging diverse sources of information. However, effectively integrating and exploiting these views remains a formidable task. In this work, we introduce a novel methodology that leverages the 2-level ensemble concept within the feature set partitioning (FSP)-based multi-view ensemble learning (MEL) framework to address this challenge. Our method involves employing different FSP techniques to divide the dataset’s feature set into subsets, and then training base models on these subsets. In the first ensemble level, predictions from these base models, applied to individual FSPs, are amalgamated using ensemble techniques. Subsequently, in the second ensemble level, predictions obtained from different FSP techniques of the first-level ensemble models are further amalgamated to derive the final prediction. This 2-level ensemble hybrid FSP (<InlineEquation ID="IEq1"> <InlineMediaObject> <ImageObject Color="BlackWhite" FileRef="42979_2025_4229_Article_IEq1.gif" Format="GIF" Height="14" Rendition="HTML" Resolution="72" Type="Linedraw" Width="19" /> </InlineMediaObject> <EquationSource Format="TEX">\({\mathcal {H}}\)</EquationSource> <EquationSource Format="MATHML"><math> <mi mathvariant="script">H</mi> </math></EquationSource> </InlineEquation>-FSP) strategy captures diverse and complementary information from each FSP, thereby enhancing the overall learning algorithm’s efficacy in MEL settings. We evaluate this <InlineEquation ID="IEq2"> <InlineMediaObject> <ImageObject Color="BlackWhite" FileRef="42979_2025_4229_Article_IEq1.gif" Format="GIF" Height="14" Rendition="HTML" Resolution="72" Type="Linedraw" Width="19" /> </InlineMediaObject> <EquationSource Format="TEX">\({\mathcal {H}}\)</EquationSource> <EquationSource Format="MATHML"><math> <mi mathvariant="script">H</mi> </math></EquationSource> </InlineEquation>-FSP approach on ten high-dimensional datasets, demonstrating its effectiveness in improving classification performance compared to both single-view learning and other cutting-edge FSP-based MEL methods. Furthermore, statistical analysis utilizing Friedman ranking corroborates the robustness of our proposed method. Our approach offers a promising solution for effectively harnessing multiple FSPs in MEL scenarios, thereby paving the way for further advancements in this field.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Enhancing the Performance of Multi-View Learning by Hybrid Feature Partitioning Method

  • 期刊论文

  • 发表日期：

    2025年7月31日

查看全文
  • Aditya Kumar,
  • Jainath Yadav

摘要

Multi-view learning aims to improve learning capability by leveraging diverse sources of information. However, effectively integrating and exploiting these views remains a formidable task. In this work, we introduce a novel methodology that leverages the 2-level ensemble concept within the feature set partitioning (FSP)-based multi-view ensemble learning (MEL) framework to address this challenge. Our method involves employing different FSP techniques to divide the dataset’s feature set into subsets, and then training base models on these subsets. In the first ensemble level, predictions from these base models, applied to individual FSPs, are amalgamated using ensemble techniques. Subsequently, in the second ensemble level, predictions obtained from different FSP techniques of the first-level ensemble models are further amalgamated to derive the final prediction. This 2-level ensemble hybrid FSP ( \({\mathcal {H}}\) H -FSP) strategy captures diverse and complementary information from each FSP, thereby enhancing the overall learning algorithm’s efficacy in MEL settings. We evaluate this \({\mathcal {H}}\) H -FSP approach on ten high-dimensional datasets, demonstrating its effectiveness in improving classification performance compared to both single-view learning and other cutting-edge FSP-based MEL methods. Furthermore, statistical analysis utilizing Friedman ranking corroborates the robustness of our proposed method. Our approach offers a promising solution for effectively harnessing multiple FSPs in MEL scenarios, thereby paving the way for further advancements in this field.