Parameter prediction for unseen deep architectures
Deep learning has been successful in automating the design of features in machine learning
pipelines. However, the algorithms optimizing neural network parameters remain largely …
pipelines. However, the algorithms optimizing neural network parameters remain largely …
Flexpoint: An adaptive numerical format for efficient training of deep neural networks
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 …
Significant gains in performance and energy efficiency could be realized by training and …
1.1 deep learning hardware: Past, present, and future
Y LeCun - 2019 IEEE International Solid-State Circuits …, 2019 - ieeexplore.ieee.org
Historically, progress in neural networks and deep learning research has been greatly
influenced by the available hardware and software tools. This paper identifies trends in deep …
influenced by the available hardware and software tools. This paper identifies trends in deep …
Designing hardware for machine learning: The important role played by circuit designers
V Sze - IEEE Solid-State Circuits Magazine, 2017 - ieeexplore.ieee.org
Machine learning is becoming increasingly important in this era of big data. It enables us to
extract meaningful information from the overwhelming amount of data being generated and …
extract meaningful information from the overwhelming amount of data being generated and …
Discovering low-precision networks close to full-precision networks for efficient embedded inference
JL McKinstry, SK Esser, R Appuswamy… - arXiv preprint arXiv …, 2018 - arxiv.org
To realize the promise of ubiquitous embedded deep network inference, it is essential to
seek limits of energy and area efficiency. To this end, low-precision networks offer …
seek limits of energy and area efficiency. To this end, low-precision networks offer …
Adaptive deep learning model selection on embedded systems
The recent ground-breaking advances in deep learning networks (DNNs) make them
attractive for embedded systems. However, it can take a long time for DNNs to make an …
attractive for embedded systems. However, it can take a long time for DNNs to make an …
[PDF][PDF] Lower numerical precision deep learning inference and training
A Rodriguez, E Segal, E Meiri, E Fomenko, YJ Kim… - Intel White Paper, 2018 - intel.com
Most commercial deep learning applications today use 32-bits of floating point precision
(𝑓𝑝32) for training and inference workloads. Various researchers have demonstrated that …
(𝑓𝑝32) for training and inference workloads. Various researchers have demonstrated that …
Adaptive neural networks for efficient inference
We present an approach to adaptively utilize deep neural networks in order to reduce the
evaluation time on new examples without loss of accuracy. Rather than attempting to …
evaluation time on new examples without loss of accuracy. Rather than attempting to …
Compute and energy consumption trends in deep learning inference
R Desislavov, F Martínez-Plumed… - arXiv preprint arXiv …, 2021 - arxiv.org
The progress of some AI paradigms such as deep learning is said to be linked to an
exponential growth in the number of parameters. There are many studies corroborating …
exponential growth in the number of parameters. There are many studies corroborating …
Communication-computation trade-off in resource-constrained edge inference
The recent breakthrough in artificial intelligence (AI), especially deep neural networks
(DNNs), has affected every branch of science and technology. Particularly, edge AI has …
(DNNs), has affected every branch of science and technology. Particularly, edge AI has …