arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 1609.08810cs.CL
收藏

通过模型组合和R-CCA方法实现语言与视觉的有效结合

Effective Combination of Language and Vision Through Model Composition and the R-CCA Method

Technion, IIT(以色列理工学院)

Hagar Loeub, Roi Reichart

摘要:

我们解决的问题,整合文本和视觉信息的向量空间模型的词义表示。我们首先提出了残差CCA(R-CCA)方法,该方法通过表示原始信号和投影到共享的最大相关空间的信号之间的差异来补充标准CCA方法。然后,我们表明,构建视觉和文本表示,然后通过PCA,CCA,R-CCA和线性插值(又名顺序建模)等常见建模主题的组合进行后处理,产生高质量的模型。在五个标准语义基准上,我们的顺序模型优于最近的多模态表示学习替代方案,包括依赖于联合表示学习的替代方案。对于其中两个基准测试,我们的R-CCA方法是我们算法产生的最佳配置的一部分。

英文摘要:

We address the problem of integrating textual and visual information in vector space models for word meaning representation. We first present the Residual CCA (R-CCA) method, that complements the standard CCA method by representing, for each modality, the difference between the original signal and the signal projected to the shared, max correlation, space. We then show that constructing visual and textual representations and then post-processing them through composition of common modeling motifs such as PCA, CCA, R-CCA and linear interpolation (a.k.a sequential modeling) yields high quality models. On five standard semantic benchmarks our sequential models outperform recent multimodal representation learning alternatives, including ones that rely on joint representation learning. For two of these benchmarks our R-CCA method is part of the Best configuration our algorithm yields.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑