Solving Visual Analogies Using Neural Algorithmic Reasoning (Student Abstract)

Atharv Sonwane, Gautam Shroff, Lovekesh Vig, Ashwin Srinivasan, Tirtharaj Dash

AAAI 2022 pp. 13055-13056

doi:10.1609/AAAI.V36I11.21664 /aaai/2022/sonwane2022aaai-solving/

Abstract

We consider a class of visual analogical reasoning problems that involve discovering the sequence of transformations by which pairs of input/output images are related, so as to analogously transform future inputs. This program synthesis task can be easily solved via symbolic search. Using a variation of the ‘neural analogical reasoning’ approach, we instead search for a sequence of elementary neural network transformations that manipulate distributed representations derived from a symbolic space, to which input images are directly encoded. We evaluate the extent to which our ‘neural reasoning’ approach generalises for images with unseen shapes and positions.

PDF AAAI Semantic Scholar

Cite

Text

Sonwane et al. "Solving Visual Analogies Using Neural Algorithmic Reasoning (Student Abstract)." AAAI Conference on Artificial Intelligence, 2022. doi:10.1609/AAAI.V36I11.21664

Markdown

[Sonwane et al. "Solving Visual Analogies Using Neural Algorithmic Reasoning (Student Abstract)." AAAI Conference on Artificial Intelligence, 2022.](https://mlanthology.org/aaai/2022/sonwane2022aaai-solving/) doi:10.1609/AAAI.V36I11.21664

BibTeX

@inproceedings{sonwane2022aaai-solving,
  title     = {{Solving Visual Analogies Using Neural Algorithmic Reasoning (Student Abstract)}},
  author    = {Sonwane, Atharv and Shroff, Gautam and Vig, Lovekesh and Srinivasan, Ashwin and Dash, Tirtharaj},
  booktitle = {AAAI Conference on Artificial Intelligence},
  year      = {2022},
  pages     = {13055-13056},
  doi       = {10.1609/AAAI.V36I11.21664},
  url       = {https://mlanthology.org/aaai/2022/sonwane2022aaai-solving/}
}