Set Prediction without Imposing Structure as Conditional Density Estimation
This work addresses set prediction tasks like point-cloud reconstruction and molecular state prediction, offering a novel approach for cases with ambiguous outcomes, though it is incremental in extending existing methods.
The paper tackles the problem of set prediction in stochastic and underdefined scenarios, where traditional set losses can lead to implausible predictions, by proposing a conditional density estimation framework that learns multi-modal densities and produces multiple plausible predictions, achieving competitive results on standard benchmarks.
Set prediction is about learning to predict a collection of unordered variables with unknown interrelations. Training such models with set losses imposes the structure of a metric space over sets. We focus on stochastic and underdefined cases, where an incorrectly chosen loss function leads to implausible predictions. Example tasks include conditional point-cloud reconstruction and predicting future states of molecules. In this paper, we propose an alternative to training via set losses by viewing learning as conditional density estimation. Our learning framework fits deep energy-based models and approximates the intractable likelihood with gradient-guided sampling. Furthermore, we propose a stochastically augmented prediction algorithm that enables multiple predictions, reflecting the possible variations in the target set. We empirically demonstrate on a variety of datasets the capability to learn multi-modal densities and produce different plausible predictions. Our approach is competitive with previous set prediction models on standard benchmarks. More importantly, it extends the family of addressable tasks beyond those that have unambiguous predictions.