Scribd
Upload
48 views4 pages

A - C L P - Application Domain: Multi-Media Systems, Recognition, Mining, Synthesis

The document discusses approximate computing for low power consumption in multimedia, recognition, mining, and synthesis systems. It notes that minor defects may still allow systems to function adequately and asks if errors in images can still allow the viewer to perceive the overall quality or if cancer can still be detected from micrographs despite some erroneous computations. The document is presented by Prof. Kaushik Roy of Purdue University on the topic of when approximate results are good enough.

Uploaded by

Dr. Sampatrao L Pinjare
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
48 views4 pages

A - C L P - Application Domain: Multi-Media Systems, Recognition, Mining, Synthesis

The document discusses approximate computing for low power consumption in multimedia, recognition, mining, and synthesis systems. It notes that minor defects may still allow systems to function adequately and asks if errors in images can still allow the viewer to perceive the overall quality or if cancer can still be detected from micrographs despite some erroneous computations. The document is presented by Prof. Kaushik Roy of Purdue University on the topic of when approximate results are good enough.

Uploaded by

Dr. Sampatrao L Pinjare
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 4

APPROXIMATE-COMPUTING FOR

LOW POWER CONSUMPTION --


Application Domain: Multi-media
systems, Recognition, Mining,
Synthesis….

Prof. Kaushik Roy


@ Purdue Univ.
When approximate is ‘good enough’

• Do few scratches make your car useless?

Prof. Kaushik Roy


@ Purdue Univ.
When approximate is ‘good enough’

• Can you quantify the image quality visually?

Prof. Kaushik Roy


@ Purdue Univ.
Error Resilient Applications: 20000ft view

• Error Resilient Applications: Produce acceptable outputs even


in the presence of imperfect implementation
• Can you spot the differences?
Input Image
Image
Segmentation
(Kmeans
Clustering)

Fully correct
computations

Application: Cancer
detection from
histological micrographs

10% erroneous computations 5% erroneous computations


Prof. Kaushik Roy
@ Purdue Univ.

You might also like