On transfer learning using a MAC model variant
By Vincent Marois, T. S. Jayram, Vincent Albouy, Tomasz Kornuta, Younes Bouhadjar, Ahmet S. Ozcan. 2018
9 citations recorded by the source corpus.
We introduce a variant of the MAC model (Hudson and Manning, ICLR 2018) with a simplified set of equations that achieves comparable accuracy, while training faster. We evaluate both models on CLEVR and CoGenT, and show that, transfer learning with fine-tuning results in a 15 point increase in accuracy, matching the state of the art. Finally, in contrast, we demonstrate that improper fine-tuning can actually reduce a model's accuracy as well.
Subfields: Artificial Intelligence, Computer Vision and Pattern Recognition
How researchers see this paper
The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.
In Artificial Intelligence
- Christopher Kanan
- Tyler L. Hayes
- Robik Shrestha
- Manoj Acharya
- Kushal Kafle
- Vincenzo Lomonaco
- Davide Maltoni
- Christopher D. Manning
In Computer Vision and Pattern Recognition
- Christopher Kanan
- Tyler L. Hayes
- Robik Shrestha
- Kushal Kafle
- Manoj Acharya
- Christopher D. Manning
- Drew A. Hudson
- Scott D. Cohen
Citations in the corpus
- REMIND Your Neural Network to Prevent Catastrophic Forgetting (2019, methodology): “ We combine REMIND with two popular VQA algorithms, using a modied version of the stacked attention network (SAN) [42,78] for TDIUC, and a simplied version of the Memory Attention and Control (MAC) [28,55] network for CLEVR. A ResNet-101 model pre-trained on ImageNet is used to extract features for ”