MLJun 7, 2017

Improving Variational Auto-Encoders using convex combination linear Inverse Autoregressive Flow

arXiv:1706.02326v226 citations
Originality Synthesis-oriented
AI Analysis

This is an incremental improvement for machine learning practitioners working with generative models.

The paper tackled improving Variational Auto-Encoders by proposing a new volume-preserving flow using a convex combination of linear Inverse Autoregressive Flow matrices, and it showed competitive performance with state-of-the-art methods on MNIST and Histopathology data.

In this paper, we propose a new volume-preserving flow and show that it performs similarly to the linear general normalizing flow. The idea is to enrich a linear Inverse Autoregressive Flow by introducing multiple lower-triangular matrices with ones on the diagonal and combining them using a convex combination. In the experimental studies on MNIST and Histopathology data we show that the proposed approach outperforms other volume-preserving flows and is competitive with current state-of-the-art linear normalizing flow.

Code Implementations1 repo
Foundations

The foundational work for this paper's niche, ranked by how specifically the neighbourhood builds on it — not by global fame.

Your Notes