Comparing Aggregators for Relational Probabilistic Models
This work addresses a foundational challenge in relational probabilistic models, which is incremental but promises to improve representations for tasks like predicting gender from movie ratings.
The paper tackled the problem of aggregation in relational probabilistic models, which is poorly understood and leads to information loss or overconfidence in existing methods, and proposed new simple aggregators and modifications that empirically outperform existing ones.
Relational probabilistic models have the challenge of aggregation, where one variable depends on a population of other variables. Consider the problem of predicting gender from movie ratings; this is challenging because the number of movies per user and users per movie can vary greatly. Surprisingly, aggregation is not well understood. In this paper, we show that existing relational models (implicitly or explicitly) either use simple numerical aggregators that lose great amounts of information, or correspond to naive Bayes, logistic regression, or noisy-OR that suffer from overconfidence. We propose new simple aggregators and simple modifications of existing models that empirically outperform the existing ones. The intuition we provide on different (existing or new) models and their shortcomings plus our empirical findings promise to form the foundation for future representations.