Weight discretization paradigm for optical neural networks

By E. Fiesler, A. Choudry, H. John Caulfield. 1990

How researchers see this paper

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

Citations in the corpus

Secure Evaluation of Quantized Neural Networks (2019, background): “4 This is a long-standing research area, with initial works already dating back to the 1990s [5, 26, 49, 62], and this extensive research body have enabled modern quantized neural networks to have essentially the same accuracy as their full-precision counterparts [14, 29, 32, 53, 65].”
Bitwise Neural Networks (2016, methodology): “Most of the effort is focused on training networks whose weights can be transformed into some quantized representations with a minimal loss of performance (Fiesler et al., 1990; Hwang & Sung, 2014).”
Fixed-point feedforward deep neural network design using weights +1, 0, and −1 (2014, methodology): “Instead of direct weight quantization, retraining with backpropagation was developed in [7], [8].”
Training Spiking Neural Networks Using Lessons From Deep Learning (2021, background): “Training quantized neural networks involves adjusting the weights and activations to use lower-precision fixed-point representations while maintaining acceptable performance and accuracy [179, 180].”
Pruning and Quantization for Deep Neural Network Acceleration: A Survey (2021, background): “Converting floating-point parameters into low numerical precision datatypes for quantizing neural network was proposed as far back as the 1990s [63, 14].”
A Survey on Methods and Theories of Quantized Neural Networks (2018, background): “Quantizing neural networks dates back to the 1990s [ Fiesler et al. , 1990; Balzer et al. , 1991; Tang and Kwan, 1993; Marchesi et al. , 1993 ] .”
A comprehensive survey on model compression and acceleration (2020, background): “The idea of quantizing neural networks is not new, it was introduced earlier by Fiesler et al. (1990) and Balzer et al. (1991) to make the implementation of neural networks easier on hardware.”
Backpropagation for Energy-Efficient Neuromorphic Computing (2015, methodology): “Backpropagation has been used to train networks with spiking neurons but with high-precision weights [9][10][11][12], and the converse, networks with trinary synapses but with continuous output neurons [13].”
Beyond Transmitting Bits: Context, Semantics, and Task-Oriented Communications (2022, unknown): “Quantization has long been employed to reduce the network size for efficient storage [119], [120].”
Extremely Low Bit Neural Network: Squeeze the Last Bit Out with ADMM (2017, background): “The research of low bits quantization of neural network can be traced back to 1990s [19, 20].”
X1000 real-time phoneme recognition VLSI using feed-forward deep neural networks (2014, methodology): “Therefore, we employ the weight quantization strategy similar to the algorithms proposed in [7, 13] to retrain the weights after the direct quantization.”
Resiliency of Deep Neural Networks under Quantization (2015, methodology): “An implementation with ternary weights were reported for neural network design with optical ﬁber networks (Fiesler et al., 1990).”