TY - GEN
T1 - Distributed Boosting Classifiers over Noisy Channels
AU - Kim, Yongjune
AU - Cassuto, Yuval
AU - Varshney, Lav R.
N1 - Publisher Copyright:
© 2020 IEEE.
PY - 2020/11/1
Y1 - 2020/11/1
N2 - We present a principled framework to address resource allocation for realizing boosting algorithms on substrates with communication noise. Boosting classifiers (e.g., AdaBoost) make a final decision via a weighted vote from local decisions of many base classifiers (weak classifiers). Suppose the base classifiers' outputs are communicated over noisy channels; these noisy outputs will degrade the final classification accuracy. We show this degradation can be effectively reduced by allocating more system resources for more important base classifiers. We formulate resource optimization problems in terms of importance metrics for boosting. Moreover, we show that the optimized noisy boosting classifiers can be more robust than bagging for noise during inference (test stage). We provide numerical evidence to demonstrate the benefits of our approach.
AB - We present a principled framework to address resource allocation for realizing boosting algorithms on substrates with communication noise. Boosting classifiers (e.g., AdaBoost) make a final decision via a weighted vote from local decisions of many base classifiers (weak classifiers). Suppose the base classifiers' outputs are communicated over noisy channels; these noisy outputs will degrade the final classification accuracy. We show this degradation can be effectively reduced by allocating more system resources for more important base classifiers. We formulate resource optimization problems in terms of importance metrics for boosting. Moreover, we show that the optimized noisy boosting classifiers can be more robust than bagging for noise during inference (test stage). We provide numerical evidence to demonstrate the benefits of our approach.
UR - http://www.scopus.com/inward/record.url?scp=85107730545&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85107730545&partnerID=8YFLogxK
U2 - 10.1109/IEEECONF51394.2020.9443551
DO - 10.1109/IEEECONF51394.2020.9443551
M3 - Conference contribution
AN - SCOPUS:85107730545
T3 - Conference Record - Asilomar Conference on Signals, Systems and Computers
SP - 1491
EP - 1496
BT - Conference Record of the 54th Asilomar Conference on Signals, Systems and Computers, ACSSC 2020
A2 - Matthews, Michael B.
PB - IEEE Computer Society
T2 - 54th Asilomar Conference on Signals, Systems and Computers, ACSSC 2020
Y2 - 1 November 2020 through 5 November 2020
ER -