A Nonparametric Bayesian Approach Toward Stacked Convolutional Independent Component Analysis

TL;DR

This paper introduces a nonparametric Bayesian convolutional ICA model that automatically infers the number of features, enabling scalable, deep unsupervised learning without tedious parameter tuning, and demonstrates superior performance on action recognition tasks.

Contribution

The paper presents a novel convolutional nonparametric Bayesian sparse ICA with Indian buffet process prior, allowing automatic feature number inference and scalable deep hierarchical feature learning.

Findings

Outperforms state-of-the-art in action recognition benchmarks

Automatically infers the number of latent features

Heuristics-free inference process

Abstract

Unsupervised feature learning algorithms based on convolutional formulations of independent components analysis (ICA) have been demonstrated to yield state-of-the-art results in several action recognition benchmarks. However, existing approaches do not allow for the number of latent components (features) to be automatically inferred from the data in an unsupervised manner. This is a significant disadvantage of the state-of-the-art, as it results in considerable burden imposed on researchers and practitioners, who must resort to tedious cross-validation procedures to obtain the optimal number of latent features. To resolve these issues, in this paper we introduce a convolutional nonparametric Bayesian sparse ICA architecture for overcomplete feature learning from high-dimensional data. Our method utilizes an Indian buffet process prior to facilitate inference of the appropriate number of latent features under a hybrid variational inference algorithm, scalable to massive datasets. As we show, our model can be naturally used to obtain deep unsupervised hierarchical feature extractors, by greedily stacking successive model layers, similar to existing approaches. In addition, inference for this model is completely heuristics-free; thus, it obviates the need of tedious parameter tuning, which is a major challenge most deep learning approaches are faced with. We evaluate our method on several action recognition benchmarks, and exhibit its advantages over the state-of-the-art.