Deep neural networks are commonly developed and trained in 32-bit floating point format. Significant gains in performance and energy efficiency could be realized by training and inference in numerical formats optimized for deep learning. Despite advances in limited precision inference in recent years, training of neural networks in low bit-width remains a challenging problem...
Authors
Naveen Rao
Vice President & General Manager, Artificial Intelligence Products Group
William H. Constable
Scott Gray
Stewart Hall
Luke Hornof
Related Content
Anatomy Of High-Performance Deep Learning Convolutions On SIMD...
Convolution layers are prevalent in many classes of deep neural networks, including Convolutional Neural Networks (CNNs) which provide state-of-the-art results...
Sequence Modeling Using a Memory Controller Extension for...
The Long Short-term Memory (LSTM) recurrent neural network is a powerful model for time series forecasting and various temporal tasks....
Unsupervised Deep Structure Learning by Recursive Independence Testing
We introduce a principled approach for unsupervised structure learning of deep, feed-forward, neural networks. We propose a new interpretation for...
Precision and Recall for Range-Based Anomaly Detection
Classical anomaly detection is principally concerned with point-based anomalies, anomalies that occur at a single data point. In this paper,...