Energy-Based Spherical Sparse Coding
This work addresses image classification challenges by integrating class information into sparse coding, though it appears incremental as it builds on existing convolutional sparse coding methods.
The paper tackles the problem of using spherical sparse codes for discriminative classification by introducing Energy-Based Spherical Sparse Coding (EB-SSC), which incorporates a learned linear bias based on class labels, and demonstrates its performance in a deep layered model for image classification.
In this paper, we explore an efficient variant of convolutional sparse coding with unit norm code vectors where reconstruction quality is evaluated using an inner product (cosine distance). To use these codes for discriminative classification, we describe a model we term Energy-Based Spherical Sparse Coding (EB-SSC) in which the hypothesized class label introduces a learned linear bias into the coding step. We evaluate and visualize performance of stacking this encoder to make a deep layered model for image classification.