Download presentation
Presentation is loading. Please wait.
Published byCecily Cameron Modified over 9 years ago
1
Multi-Task Learning for Boosting with Application to Web Search Ranking Olivier Chapelle et al. Presenter: Wei Cheng
2
Outline Motivation Backgrounds Algorithm – From svm to boosting using L1 regularization – Є-boosting for optimization – Overall algorithm Evaluation Overall review and new research points discussion
3
Motivation Different/same search engine(s) for different countries? Domain specific engine is better! e.g. ‘gelivable’ (very useful) Domain specific engine is better! e.g. ‘gelivable’ (very useful)
4
Motivation Should we train ranking model separately? – Corps in some domains might be too small to train a good model – Solution: Multi-task learning
5
Backgrounds Sinno Jialin Pan and Qiang Yang, TKDE’2010
6
Backgrounds Sinno Jialin Pan and Qiang Yang, TKDE’2010
7
Backgrounds Why transfer learning works? Sinno Jialin Pan et al. At WWW’10
8
Backgrounds Why transfer learning works?(continue)
9
Backgrounds Why transfer learning works?(continue)
10
Backgrounds LearnerA Input: Target: LearnerB Dog/human Girl/boy traditional learning
11
Backgrounds Joint Learning Task Input: Target: Dog/human Girl/boy Multi-task learning
12
Algorithm The algorithm aims at designing an algorithm based on gradient boosted decision trees Inspired by svm based multi-task solution and boosting-trick. Using Є-boosting for optimization
13
Algorithm From svm to boosting using L1 regularization Previous svm based multi-task learning:
14
Algorithm Svm(kernel-trick)--- boosting (boosting trick) Pick set of non-linear functions(e.g., decision trees, regression trees,….) H |H|=J Apply every single function to each data point X ф(X)
15
Є-boosting for optimization Using L1 regulization Using Є-boosting
16
Algorithm
17
Evaluation Datasets
18
Evaluation
23
Overall review and new research point discussion Contributions: – Propose a novel multi-task learning method based on gradient boosted decision tree, which is useful for web-reranking applications. (e.g., personalized search). – Have a thorough evaluation on we-scale datasets. New research points: – Negative transfer: – Effective grouping: flexible domain adaptation
24
Q&A Thanks!
Similar presentations
© 2024 SlidePlayer.com. Inc.
All rights reserved.