Authors argue by analogy that localised hardware ML architectures may be more interpretable and efficient than deep neural networks on GPUs for smaller datasets and evaluate candidate hardware paradigms.
A Brief History of Artificial Intelligence: On the Past, Present, and Future of Artificial Intelligence
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Enhancing AI Interpretability and Safety through Localised Architectures
Authors argue by analogy that localised hardware ML architectures may be more interpretable and efficient than deep neural networks on GPUs for smaller datasets and evaluate candidate hardware paradigms.