Extreme Multi - label Loss Functions for Recommendation , Tagging , Ranking & Other Missing Label Applications – Supplementary Material

نویسندگان

  • Himanshu Jain
  • Yashoteja Prabhu
  • Manik Varma
چکیده

1. PROPENSITY SCORED LOSSES This section provides proofs for all the theorems stated in Section 4 of the paper. Theorem 4.1. The loss function L(y, ŷ) evaluated on the observed ground truth y is an unbiased estimator of the true loss function L∗(y∗, ŷ) evaluated on complete ground truth y∗. Thus, Ey[L(y, ŷ)] = Ey∗ [L∗(y∗, ŷ)], for any P (y∗) and P (y) related through propensities pl and any fixed ŷ. Proof. Ey[L(y, ŷ)] = ∑

برای دانلود متن کامل این مقاله و بیش از 32 میلیون مقاله دیگر ابتدا ثبت نام کنید

ثبت نام

اگر عضو سایت هستید لطفا وارد حساب کاربری خود شوید

منابع مشابه

Extreme Multi - label Loss Functions for Recommendation , Tagging , Ranking & Other Missing Label Applications – Supplementary

1. PROPENSITY SCORED LOSSES Theorem 4.1. The loss function L(y, ˆ y) evaluated on the observed ground truth y is an unbiased estimator of the true loss function L * (y * , ˆ y) evaluated on complete ground truth y *. Thus, Ey[L(y, ˆ y)] = Ey * [L * (y * , ˆ y)], for any P (y *) and P (y) related through propensities p l and any fixedˆy. Proof.

متن کامل

Supplementary : Extreme Multi-label Learning with Label Features for Warm-start Tagging, Ranking & Recommendation

Section 1 presents the pseudocodes for SwiftXML training and prediction algorithms. Section 2 reports complete set of experimental results comparing SwiftXML to various baselines in terms of both propensity-scored precisions (PSP1,PSP3,PSP5) as well as standard precisions (P1,P3,P5). Section 3 shows the derivations for individual steps of the alternating minimization algorithm used for node par...

متن کامل

On the Consistency of Multi-Label Learning

Multi-label learning has attracted much attention during the past few years. Many multilabel learning approaches have been developed, mostly working with surrogate loss functions since multi-label loss functions are usually difficult to optimize directly owing to non-convexity and discontinuity. Though these approaches are effective, to the best of our knowledge, there is no theoretical result ...

متن کامل

Learning Extreme Multi-label Tree-classifier via Nearest Neighbor Graph Partitioning

Web scale classification problems, such as Web page tagging and E-commerce product recommendation, are typically regarded as multi-label classification with an extremely large number of labels. In this paper, we propose GPT, which is a novel tree-based approach for extreme multi-label learning. GPT recursively splits a feature space with a hyperplane at each internal node, considering approxima...

متن کامل

Exploiting Associations between Class Labels in Multi-label Classification

Multi-label classification has many applications in the text categorization, biology and medical diagnosis, in which multiple class labels can be assigned to each training instance simultaneously. As it is often the case that there are relationships between the labels, extracting the existing relationships between the labels and taking advantage of them during the training or prediction phases ...

متن کامل

ذخیره در منابع من


  با ذخیره ی این منبع در منابع من، دسترسی به آن را برای استفاده های بعدی آسان تر کنید

برای دانلود متن کامل این مقاله و بیش از 32 میلیون مقاله دیگر ابتدا ثبت نام کنید

ثبت نام

اگر عضو سایت هستید لطفا وارد حساب کاربری خود شوید

عنوان ژورنال:

دوره   شماره 

صفحات  -

تاریخ انتشار 2016