Hadamard Product for Low-rank Bilinear Pooling. (arXiv:1610.04325v1 [cs.CV])

Bilinear models provide rich representations compared to linear models. They have been applied in various visual tasks, such as object recognition, segmentation, and visual question-answering, to get state-of-the-art performances taking advantage of the expanded representations. However, bilinear representations tend to be high-dimensional, limiting the applicability to computationally complex tasks. We propose low-rank bilinear neural networks using Hadamard product (element-wise multiplication), commonly implemented in many scientific computing frameworks. We show that our model outperforms compact bilinear pooling in visual question-answering tasks, having a better parsimonious property.

from cs.AI updates on arXiv.org http://ift.tt/2ed1inE
via IFTTT