Approximate computing: Difference between revisions

Revision as of 12:11, 25 May 2021

Revision as of 12:32, 26 May 2021

Line 16:

==Application areas==

Approximate computing has been used in a variety of domains where the applications are error-tolerant, such as [[multimedia]] processing, [[machine learning]], [[signal processing]], [[Computational science|scientific computing]]. Therefore, etc.approximate Googlecomputing is usingmostly thisdriven approachby inapplications theirthat [[Tensorare processingrelated unit]]sto (TPUhuman perception/cognition and have inherent error resilience. Many of these applications are based on statistical or probabilistic computation, asuch customas ASIC)different approximations can be made to better suit the desired objectives.<ref>{{Citationcite journal |last1=Liu |first1=Weiqiang |last2=Lombardi |first2=Fabrizio |last3=Schulte |first3=Michael |title=Approximate Computing: From Circuits to Applications |journal=Proceedings of the IEEE needed|date=Dec 2020 |volume=108 |issue=12 |page=2103 |doi=10.1109/JPROC.2020.3033361 |access-date=26 May 2021 | url=https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=9264836}}.</ref>

One notable application in [[machine learning]] is that Google is using this approach in their [[Tensor processing unit]]s (TPU, a custom ASIC). <ref>{{cite journal |last1=Liu |first1=Weiqiang |last2=Lombardi |first2=Fabrizio |last3=Schulte |first3=Michael |title=Approximate Computing: From Circuits to Applications |journal=Proceedings of the IEEE |date=Dec 2020 |volume=108 |issue=12 |page=2104 |doi=10.1109/JPROC.2020.3033361 |access-date=26 May 2021 | url=https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=9264836}}</ref>

==Derived paradigms==