Tunable Floating-Point for Artificial Neural Networks
| Autor: | Marta Franceschi, Alberto Nannarelli, Maurizio Valle |
|---|---|
| Rok vydání: | 2018 |
| Předmět: |
Floating point
Computer science Computation Energy-efficient design Power efficient 02 engineering and technology 01 natural sciences Machine learning approaches Deep neural networks Machine learning 0103 physical sciences 0202 electrical engineering electronic engineering information engineering Power efficiency Representation (mathematics) Digital signal processing Floating points 010302 applied physics Flexibility (engineering) Floating-point Artificial neural network Energy efficiency Neural networks Digital system Power efficient Digital arithmetic neural networks power efficiency business.industry 020202 computer hardware & architecture Significand Computer engineering Digital system business Electrical efficiency Digital arithmetic Neural networks |
| Zdroj: | ICECS Franceschi, M, Nannarelli, A & Valle, M 2018, Tunable Floating-Point for Artificial Neural Networks . in Proceedings of 25th IEEE International Conference on Electronics Circuits and Systems . IEEE, pp. 289-292, 2018 IEEE 25th International Conference on Electronics, Circuits and Systems, Bordeaux, France, 09/12/2018 . https://doi.org/10.1109/ICECS.2018.8617900 |
| Popis: | Approximate computing has emerged as a promising approach to energy-efficient design of digital systems in many domains such as digital signal processing, robotics, and machine learning. Numerous studies report that employing different data formats in Deep Neural Networks (DNNs), the dominant Machine Learning approach, could allow substantial improvements in power efficiency considering an acceptable quality for results. In this work, the application of Tunable Floating-Point (TFP) precision to DNN is presented. In TFP different precisions for different operations can be set by selecting a specific number of bits for significand and exponent in the floating-point representation. Flexibility in tuning the precision of given layers of theneural network may result in a more power efficient computation. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|