DFM: Interpolant-free Dual Flow Matching

TL;DR

The paper introduces DFM, a new approach for continuous normalizing flows that eliminates the need for interpolants and explicitly models both forward and reverse vector fields, improving efficiency and performance.

Contribution

DFM is the first interpolant-free dual flow matching method that jointly optimizes forward and reverse vector fields without explicit assumptions, enhancing model bijectivity.

Findings

DFM outperforms traditional CNFs in anomaly detection tasks.

DFM achieves state-of-the-art performance metrics.

DFM simplifies training by removing the need for interpolants.

Abstract

Continuous normalizing flows (CNFs) can model data distributions with expressive infinite-length architectures. But this modeling involves computationally expensive process of solving an ordinary differential equation (ODE) during maximum likelihood training. Recently proposed flow matching (FM) framework allows to substantially simplify the training phase using a regression objective with the interpolated forward vector field. In this paper, we propose an interpolant-free dual flow matching (DFM) approach without explicit assumptions about the modeled vector field. DFM optimizes the forward and, additionally, a reverse vector field model using a novel objective that facilitates bijectivity of the forward and reverse transformations. Our experiments with the SMAP unsupervised anomaly detection show advantages of DFM when compared to the CNF trained with either maximum likelihood or FM objectives with the state-of-the-art performance metrics.