Loss Functions for Multiset Prediction

TL;DR

This paper introduces a new loss function for multiset prediction, addressing the challenge of unordered and multi-occurrence items, and demonstrates its effectiveness through experiments on synthetic and real datasets.

Contribution

A novel multiset loss function is proposed, framing the problem as sequential decision making, and shown to outperform existing baseline loss functions.

Findings

The proposed loss function outperforms reinforcement learning, sequence, and distribution matching losses.

Experiments on synthetic and real datasets validate the effectiveness of the new loss.

The approach handles unordered multisets with repeated items more effectively than existing methods.

Abstract

We study the problem of multiset prediction. The goal of multiset prediction is to train a predictor that maps an input to a multiset consisting of multiple items. Unlike existing problems in supervised learning, such as classification, ranking and sequence generation, there is no known order among items in a target multiset, and each item in the multiset may appear more than once, making this problem extremely challenging. In this paper, we propose a novel multiset loss function by viewing this problem from the perspective of sequential decision making. The proposed multiset loss function is empirically evaluated on two families of datasets, one synthetic and the other real, with varying levels of difficulty, against various baseline loss functions including reinforcement learning, sequence, and aggregated distribution matching loss functions. The experiments reveal the effectiveness of the proposed loss function over the others.