研究人员引入了一种新颖的结构化稀疏非负低秩分解框架，以改进二分网络（尤其是在生态学研究中使用的网络）中潜在结构的推断。该方法通过纳入检测概率估计和施加非凸 $\ell_{1/2}$ 正则化来促进稀疏性和更好的相对尺度，从而解决了现有模型的局限性。开发了一种基于ADMM的算法来解决由此产生的非凸和非光滑优化问题，实验表明在合成和真实生态数据集上均能增强潜在因子和网络结构的恢复。 AI
排序理由 这是一篇发表在arXiv上的研究论文，详细介绍了一个新的网络推断框架。
AI 生成摘要 · Google Gemini · 来自 4 个来源。 我们如何撰写摘要 →
Recovering latent structure from count data has received considerable attention in network inference, particularly when one seeks both cross-group interactions and within-group similarity patterns in bipartite networks, which is widely used in ecology research. Such networks are …
arXiv:2604.18820v2 Announce Type: replace Abstract: Recovering latent structure from count data has received considerable attention in network inference, particularly when one seeks both cross-group interactions and within-group similarity patterns in bipartite networks, which is…
Recovering latent structure from count data has received considerable attention in network inference, particularly when one seeks both cross-group interactions and within-group similarity patterns in bipartite networks, which is widely used in ecology research. Such networks are …
Recovering latent structure from count data has received considerable attention in network inference, particularly when one seeks both cross-group interactions and within-group similarity patterns in bipartite networks, which is widely used in ecology research. Such networks are …