abduction
发表于 2025-3-25 05:22:32
http://reply.papertrans.cn/63/6206/620513/620513_21.png
放逐某人
发表于 2025-3-25 10:38:25
Fast Label Embeddings via Randomized Linear Algebrasettings. The result is a randomized algorithm whose running time is exponentially faster than naive algorithms. We demonstrate our techniques on two large-scale public datasets, from the Large Scale Hierarchical Text Challenge and the Open Directory Project, where we obtain state of the art results.
perjury
发表于 2025-3-25 13:06:06
Regression with Linear Factored FunctionsWe derive a regularized greedy optimization scheme, that learns factored basis functions during training. The novel regression algorithm performs competitively to . on benchmark tasks, and the learned LFF functions are with 4-9 factored basis functions on average very compact.
Apoptosis
发表于 2025-3-25 18:32:44
Ridge Regression, Hubness, and Zero-Shot Learningprove that the proposed approach indeed reduces hubness. This was verified empirically on the tasks of bilingual lexicon extraction and image labeling: hubness was reduced with both of these tasks and the accuracy was improved accordingly.
懒惰人民
发表于 2025-3-25 23:59:25
http://reply.papertrans.cn/63/6206/620513/620513_25.png
Erythropoietin
发表于 2025-3-26 01:29:25
http://reply.papertrans.cn/63/6206/620513/620513_26.png
安抚
发表于 2025-3-26 07:30:09
http://reply.papertrans.cn/63/6206/620513/620513_27.png
富足女人
发表于 2025-3-26 09:33:18
http://reply.papertrans.cn/63/6206/620513/620513_28.png
Customary
发表于 2025-3-26 14:02:33
ConDist: A Context-Driven Categorical Distance Measureal distance measures and evaluate on different data sets from the UCI machine-learning repository. The experiments show that our distance measure is recommendable, since it achieves similar or better results in a more robust way than previous approaches.
gangrene
发表于 2025-3-26 17:23:36
Conference proceedings 2015overy in Databases, ECML PKDD 2015, held in Porto, Portugal, in September 2015. .The 131 papers presented in these proceedings were carefully reviewed and selected from a total of 483 submissions. These include 89 research papers, 11 industrial papers, 14 nectar papers, and 17 demo papers. They were