Backpropagation without Multiplication

By Patrice Y. Simard, Hans Peter Graf. 1993

Subfields: Optimization & Training

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Optimization & Training

1. Matthieu Courbariaux
2. David Grangier
3. Zhouhan Lin
4. Michael Auli
5. Roland Memisevic
6. Jean-Pierre David
7. Yoshua Bengio
8. Sergey Edunov

Citations in the corpus

• Scaling Neural Machine Translation (2018, methodology): “Previous research considered training and inference with reduced numerical precision for neural networks (Simard and Graf, 1993; Courbariaux et al., 2015; Sa et al., 2018).”
• Neural Networks with Few Multiplications (2015, positive): “For instance, instead of reducing the precision of weights, Simard & Graf (1994) quantizes states, learning rates, and gradients to powers of two.”
• Training deep neural networks with low precision multiplications (2014, methodology): “Training neural networks with low precision arithmetic has already been done in previous works (Holt and Baker, 1991; Presley and Haggard, 1994; Simard and Graf, 1994; Wawrzynek et al., 1996; Savich et al., 2007) 2 .”
• Low precision storage for deep learning (2014, methodology): “Training neural networks with low precision has already been done in previous works (Holt and Baker, 1991; Presley and Haggard, 1994; Simard and Graf, 1994; Savich et al. , 2007).”