Gradient Boosting Decision Tree (GBDT) is a popular machine learning algorithm, and has quite a few effective implementations such as XGBoost and pGBRT. Although many engineering optimizations have been adopted in these implementations, the efficiency and scalability are still unsatisfactory when the feature dimension is high and data size is large. A major reason is that for each feature, they need to scan all the data instances to estimate the information gain of all possible split points, which is very time consuming. To tackle this problem, we propose two novel techniques: Gradient-based One-Side Sampling (GOSS) and Exclusive Feature Bundling (EFB). With GOSS, we exclude a significant proportion of data instances with small gradients, and only use the rest to estimate the information gain. We prove that, since the data instances with larger gradients play a more important role in the computation of information gain, GOSS can obtain quite accurate estimation of the information gain with a much smaller data size. With EFB, we bundle mutually exclusive features (i.e., they rarely take nonzero values simultaneously), to reduce the number of features. We prove that finding the optimal bundling of exclusive features is NP-hard, but a greedy algorithm can achieve quite good approximation ratio (and thus can effectively reduce the number of features without hurting the accuracy of split point determination by much). We call our new GBDT implementation with GOSS and EFB LightGBM. Our experiments on multiple public datasets show that, LightGBM speeds up the training process of conventional GBDT by up to over 20 times while achieving almost the same accuracy.

/pdf/lightgbm-a-highly-efficient-gradient-boosting-decision-tree-f835dtmi7b.pdf

LightGBM: a highly efficient gradient boosting decision tree

General Strategies.- Regression Models.- Classification Models.- Other Considerations.- Appendix.- References.- Indices.

Applied Predictive Modeling

This tutorial is concerned with a comprehensive introduction to the research area of learning to rank for information retrieval. In the first part of the tutorial, we will introduce three major approaches to learning to rank, i.e., the pointwise, pairwise, and listwise approaches, analyze the relationship between the loss functions used in these approaches and the widely-used IR evaluation measures, evaluate the performance of these approaches on the LETOR benchmark datasets, and demonstrate how to use these approaches to solve real ranking applications. In the second part of the tutorial, we will discuss some advanced topics regarding learning to rank, such as relational ranking, diverse ranking, semi-supervised ranking, transfer ranking, query-dependent ranking, and training data preprocessing. In the third part, we will briefly mention the recent advances on statistical learning theory for ranking, which explain the generalization ability and statistical consistency of different ranking methods. In the last part, we will conclude the tutorial and show several future research directions.

/pdf/learning-to-rank-for-information-retrieval-2q6t3viyug.pdf

Learning to Rank for Information Retrieval

Learning sophisticated feature interactions behind user behaviors is critical in maximizing CTR for recommender systems. Despite great progress, existing methods seem to have a strong bias towards low- or high-order interactions, or require expertise feature engineering. In this paper, we show that it is possible to derive an end-to-end learning model that emphasizes both low- and high-order feature interactions. The proposed model, DeepFM, combines the power of factorization machines for recommendation and deep learning for feature learning in a new neural network architecture. Compared to the latest Wide & Deep model from Google, DeepFM has a shared input to its "wide" and "deep" parts, with no need of feature engineering besides raw features. Comprehensive experiments are conducted to demonstrate the effectiveness and efficiency of DeepFM over the existing models for CTR prediction, on both benchmark data and commercial data.

/pdf/deepfm-a-factorization-machine-based-neural-network-for-ctr-xde2g2fq1v.pdf

DeepFM: a factorization-machine based neural network for CTR prediction

Search engine click logs provide an invaluable source of relevance information, but this information is biased. A key source of bias is presentation order: the probability of click is influenced by a document's position in the results page. This paper focuses on explaining that bias, modelling how probability of click depends on position. We propose four simple hypotheses about how position bias might arise. We carry out a large data-gathering effort, where we perturb the ranking of a major search engine, to see how clicks are affected. We then explore which of the four hypotheses best explains the real-world position effects, and compare these to a simple logistic regression model. The data are not well explained by simple position models, where some users click indiscriminately on rank 1 or there is a simple decay of attention over ranks. A 'cascade' model, where users view results from top to bottom and leave as soon as they see a worthwhile document, is our best explanation for position bias in early ranks

An experimental comparison of click position-bias models

Search engine advertising has become a significant element of the Web browsing experience. Choosing the right ads for the query and the order in which they are displayed greatly affects the probability that a user will see and click on each ad. This ranking has a strong impact on the revenue the search engine receives from the ads. Further, showing the user an ad that they prefer to click on improves user satisfaction. For these reasons, it is important to be able to accurately estimate the click-through rate of ads in the system. For ads that have been displayed repeatedly, this is empirically measurable, but for new ads, other means must be used. We show that we can use features of ads, terms, and advertisers to learn a model that accurately predicts the click-though rate for new ads. We also show that using our model improves the convergence and performance of an advertising system. As a result, our model increases both revenue and user satisfaction.

/pdf/predicting-clicks-estimating-the-click-through-rate-for-new-5b9f38ivaq.pdf

Predicting clicks: estimating the click-through rate for new ads

While interacting with a messaging interface, keywords or other search criteria are automatically identified and used to perform a search. Search results and sponsored links (or advertisements) are displayed to the user within the messaging interface. To alleviate privacy concerns, this process will not be performed unless the user has explicitly opted-in to the search feature. In another embodiment, the user can highlight keywords in an outbound or inbound message to trigger a search without leaving the messaging interface. In another embodiment, the user can input a search keyword or phrase to trigger a search without leaving the messaging interface.

Searching based on messages

A contextual multimedia advertisement system and associated methods are disclosed. The architecture includes a component that determines content of a multimedia item, and a distribution component that facilitates presentation of one or more advertisements associated with the content. Additionally or alternatively, advertisement presentation, selection and/or generation can be based on context information pertaining to the item and/or end user as well as ratings influenced by advertisers, among other things. The one or more identified advertisements can be presented with the item at valuable points therein.

Contextual multimedia advertisement presentation

Systems, methods, and computer-readable media for estimating expected advertisement performance of advertisements are provided. An advertisement performance prediction model is developed using features extracted from a sample set. Once developed, advertisements that are not a part of the sample set are identified and features are extracted there from. The features are then input into the advertisement performance prediction model and expected performance of the corresponding advertisement is estimated. In embodiments, the estimated expected advertisement performance may be used to appropriately rank the advertisement relative to a plurality of other advertisements such that the advertisement will be displayed according to the advertisement ranking.

Estimating expected performance of advertisements

A predictive travel log system using one or more ranking schemes to predict the location that a user would seek to navigate to through back and forward navigation in a web browser is provided. The navigation functionality can be exposed through traditional back and forward buttons with drop down travel log menus found in a browser user interface.

Ewa Dominowska

Papers

Predicting clicks: estimating the click-through rate for new ads

Searching based on messages

Contextual multimedia advertisement presentation

Estimating expected performance of advertisements

Browsing web content using predictive navigation links