Designing Communication Receivers Using Machine Learning Techniques is coming
at 11/20/2017 - 4:00pm

LPSC 125
Mon, 11/20/2017 - 4:00pm

Brian M. Kurkoski
Associate Professor, Japan Advanced Institute of Science and Technology

Abstract:
Your smartphone has many communications receivers, not only in its various
wireless interfaces, but in the flash memory controller as well.  In
fixed-precision VLSI receivers, reducing the number of bits used to represent
messages will reduce power consumption and increase battery life. This
presentation describes the design of fixed-precision receivers from an
information theory perspective. This can be called "hardware-aware
information theory" because the objective is to maximize mutual information
(an information theory quantity) while minimizing the number of message bits
(in the hardware implementation).  Results from machine learning play a key
role, because quantization can be seen as classification.  Numerical results
show that LDPC decoders based on the proposed max-LUT method can outperform
belief-propagation decoders.

Bio:

Read more:
http://eecs.oregonstate.edu/colloquium/designing-communication-receivers... 
[1]


[1] 
http://eecs.oregonstate.edu/colloquium/designing-communication-receivers-using-machine-learning-techniques
_______________________________________________
Colloquium mailing list
[email protected]
https://secure.engr.oregonstate.edu/mailman/listinfo/colloquium

Reply via email to