Ranking via Robust Binary Classification and Parallel Parameter Estimation in Large-Scale Data
This addresses the challenge of efficient and high-quality ranking in large-scale data for applications like search engines or recommendation systems, with incremental improvements in parallelization.
The authors tackled the problem of learning to rank by proposing RoBiRank, an algorithm that connects ranking evaluation metrics to robust classification loss functions, achieving competitive performance on standard benchmarks and dramatically higher quality solutions in large-scale tasks compared to a state-of-the-art competitor given the same computation time.
We propose RoBiRank, a ranking algorithm that is motivated by observing a close connection between evaluation metrics for learning to rank and loss functions for robust classification. The algorithm shows a very competitive performance on standard benchmark datasets against other representative algorithms in the literature. On the other hand, in large scale problems where explicit feature vectors and scores are not given, our algorithm can be efficiently parallelized across a large number of machines; for a task that requires 386,133 x 49,824,519 pairwise interactions between items to be ranked, our algorithm finds solutions that are of dramatically higher quality than that can be found by a state-of-the-art competitor algorithm, given the same amount of wall-clock time for computation.