Essential Loss: Bridge the Gap between Ranking Measures and Loss Functions in Learning to Rank

نویسندگان

  • Wei Chen
  • Tie-Yan Liu
  • Yanyan Lan
  • Hang Li
چکیده

Learning to rank has become an important research topic in machine learning. While most learning-to-rank methods learn the ranking functions by minimizing the loss functions, it is the ranking measures (such as NDCG and MAP) that are used to evaluate the performance of the learned ranking functions. In this work, we reveal the relationship between ranking measures and loss functions in learningto-rank methods, such as Ranking SVM, RankBoost, RankNet, and ListMLE. We show that the loss functions of these methods are upper bounds of the measurebased ranking errors. As a result, the minimization of these loss functions will lead to the maximization of the ranking measures. The key to obtaining this result is to model ranking as a sequence of classification tasks, and define a so-called essential loss for ranking as the weighted sum of the classification errors of individual tasks in the sequence. We have proved that the essential loss is both an upper bound of the measure-based ranking errors, and a lower bound of the loss functions in the aforementioned methods. Our proof technique also suggests a way to modify existing loss functions to make them tighter bounds of the measure-based ranking errors. Experimental results on benchmark datasets show that the modifications can lead to better ranking performances, demonstrating the correctness of our theoretical analysis.

برای دانلود رایگان 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

Ranking Measures and Loss Functions in Learning to Rank

Learning to rank has become an important research topic in machine learning. While most learning-to-rank methods learn the ranking functions by minimizing loss functions, it is the ranking measures (such as NDCG and MAP) that are used to evaluate the performance of the learned ranking functions. In this work, we reveal the relationship between ranking measures and loss functions in learningto-r...

متن کامل

A Unified View of Loss Functions in Learning to Rank

This paper provides a unified view of loss functions used in learning to rank. Loss function is a key component in learning to rank, because it encodes human knowledge on evaluation of ranking and guides the process of learning. Many loss functions have been proposed in the literature of learning to rank, with different forms and different motivations, and have been exploited in the development...

متن کامل

Directly Optimizing Evaluation Measures in Learning to Rank

One of the central issues in learning to rank for information retrieval is to develop algorithms that construct ranking models by directly optimizing evaluation measures used in information retrieval such as Mean Average Precision (MAP) and Normalized Discounted Cumulative Gain (NDCG). Several such algorithms including SVMmap and AdaRank have been proposed and their effectiveness has been verif...

متن کامل

Identifying natural gas loss risks and ranking of corrective actions

The aim of this study was to provide a new model for identifying the sources and sources of waste gas in Mahdishahr city gas department and to define corrective measures and prioritize measures to help managers to make appropriate decisions to reduce waste gas. The research method is descriptive-analytical in terms of nature and is applied in terms of purpose. The statistical sample of the rese...

متن کامل

Label Ranking by Directly Optimizing Performance Measures

Label ranking aims to map instances to an order over a predefined set of labels. It is ideal that the label ranking model is trained by directly maximizing performance measures on training data. However, existing studies on label ranking models mainly based on the minimization of classification errors or rank losses. To fill in this gap in label ranking, in this paper a novel label ranking mode...

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

عنوان ژورنال:

دوره   شماره 

صفحات  -

تاریخ انتشار 2009