no code implementations • 26 Jan 2024 • Wen Ma, Qiuwen Lou, Arman Kazemi, Julian Faraone, Tariq Afzal
Video quality can suffer from limited internet speed while being streamed by users.
no code implementations • ICLR 2021 • Sean Fox, Seyedramin Rasoulinezhad, Julian Faraone, David Boland, Philip Leong
Training Deep Neural Networks (DNN) with high efficiency can be difficult to achieve with native floating point representations and commercially available hardware.
no code implementations • 19 Nov 2019 • Julian Faraone, Martin Kumm, Martin Hardieck, Peter Zipf, Xueyuan Liu, David Boland, Philip H. W. Leong
Low-precision arithmetic operations to accelerate deep-learning applications on field-programmable gate arrays (FPGAs) have been studied extensively, because they offer the potential to save silicon area or increase throughput.
no code implementations • 25 Sep 2019 • Julian Faraone, Philip Leong
We present a novel technique, Monte Carlo Deep Neural Network Arithmetic (MCA), for determining the sensitivity of Deep Neural Networks to quantization in floating point arithmetic. We do this by applying Monte Carlo Arithmetic to the inference computation and analyzing the relative standard deviation of the neural network loss.
1 code implementation • CVPR 2018 • Julian Faraone, Nicholas Fraser, Michaela Blott, Philip H. W. Leong
An efficient way to reduce this complexity is to quantize the weight parameters and/or activations during training by approximating their distributions with a limited entry codebook.
no code implementations • 19 Sep 2017 • Julian Faraone, Nicholas Fraser, Giulio Gambardella, Michaela Blott, Philip H. W. Leong
A low precision deep neural network training technique for producing sparse, ternary neural networks is presented.